SPB Git

Repository Public License

conversion-matrix.md 637.2 KB

# ORIX Convert — conversion matrix

Generated file — do not edit. Produced by Scripts/matrix.sh (orix-convert matrix --markdown) from the live engine registry of the build: every row is a ConversionPath an engine declares, and every total below is computed from those rows. The Tests and Performance columns are read from the engine notes in docs/compatibility/engines/. The end-to-end result of each path is in docs/reports/e2e-matrix.md.

Build identity (no date, no commit: the fingerprint changes exactly when the matrix changes):

  • Tool: orix-convert 1.0.0
  • Engine packages in the build (12): EngineImage, EnginePDF, EngineVector, EngineMedia, EngineDocuments, EngineData, EngineArchive, EngineFont, EngineSubtitle, EngineModel3D, EngineGeo, EngineScientific
  • Engines (15): image, pdf, vector, video, audio, documents, ebooks, presentations, data, archive, font, subtitle, model3d, geo, scientific
  • Registry fingerprint (SHA-256): bb10f953b1dda4758eaef85a42eadbcaaee948576713005cb6fe646747f33edf

# Totals

Measure Count
Formats with at least one path 117
…readable (source of a path) 117
…writable (destination of a path) 111
Distinct source → destination pairs 1533
…of which with a supported path 1469
…of which reachable by a plain conversion 1328
Paths (conversions and operations) 1613
…plain conversions 1328
…operation paths 285
Named operations 29
Paths of unavailable engines 0
Engine Paths
image 443
pdf 24
vector 26
video 302
audio 132
documents 82
ebooks 23
presentations 24
data 219
archive 119
font 11
subtitle 30
model3d 132
geo 36
scientific 10
Status Paths
experimental 64
supported 1549
Fidelity Paths
lossless 292
lossy 635
partial 280
structural 406
Source category Paths
image 472
vector 4
video 302
audio 132
document 91
pdf 17
spreadsheet 42
data 182
presentation 24
archive 108
model3D 132
font 11
subtitle 30
ebook 14
geospatial 31
scientific 10
other 11

Tests column: 1495 path(s) carry the test named in the engine notes, 118 the engine's matrix suite (the notes name no test for them), 0 none. Performance column: 48 path(s) have a number measured by the engine's author and quoted from the notes with its scenario; "—" means no measurement exists for that path.

# image sources (472 paths)

Source Destination Operation Engine Status Fidelity Preserves Limitations Tests Performance
apng apng — image supported lossless animation, colorProfile, dimensions, transparency — MatrixTests.everyPath —
apng avif — image supported lossy colorProfile, dimensions, transparency Animated sources: only the first frame is kept. Lossy compression; the PSNR of the result is measured. At most 10 bits per component. MatrixTests.everyPath —
apng bmp — image supported lossy dimensions, transparency Animated sources: only the first frame is kept. Samples deeper than 8-bit are reduced to 8-bit. No embedded colour profile: pixels are converted to sRGB. Transparency is stored in a 32-bit BMP; some programs ignore BMP alpha. MatrixTests.everyPath —
apng exr — image supported lossy dimensions, transparency Animated sources: only the first frame is kept. Pixels are stored as linear values with sRGB / Rec. 709 primaries. 8-bit sources are stored as 16-bit half floats, deeper sources as 32-bit floats. MatrixTests.everyPath —
apng gif — image supported lossy animation, dimensions, transparency GIF stores at most 256 colours per frame and on/off transparency. Samples deeper than 8-bit are reduced to 8-bit. No embedded colour profile: pixels are converted to sRGB. MatrixTests.everyPath —
apng hdr — image supported lossy dimensions Animated sources: only the first frame is kept. HDR has no transparency: transparent areas are filled with the background colour. Pixels are stored as linear values with sRGB / Rec. 709 primaries. RGBE keeps an 8-bit mantissa per channel. MatrixTests.everyPath —
apng heic — image supported lossy colorProfile, dimensions, transparency Animated sources: only the first frame is kept. Lossy compression; the PSNR of the result is measured. At most 10 bits per component. MatrixTests.everyPath —
apng heif — image supported lossy colorProfile, dimensions, transparency Animated sources: only the first frame is kept. Lossy compression; the PSNR of the result is measured. At most 10 bits per component. Written as an HEVC-coded HEIF file (the same encoding as HEIC). MatrixTests.everyPath —
apng icns — image supported lossy transparency Animated sources: only the first frame is kept. Samples deeper than 8-bit are reduced to 8-bit. No embedded colour profile: pixels are converted to sRGB. The image is resampled to the standard icon sizes, fitted in a transparent square. MatrixTests.everyPath —
apng ico — image supported lossy transparency Animated sources: only the first frame is kept. Samples deeper than 8-bit are reduced to 8-bit. No embedded colour profile: pixels are converted to sRGB. The image is resampled to the standard icon sizes, fitted in a transparent square. MatrixTests.everyPath —
apng jp2 — image supported lossy dimensions, transparency Animated sources: only the first frame is kept. Lossy by default (a lossless mode is available); the PSNR of the result is measured. Samples deeper than 8-bit are reduced to 8-bit. No embedded colour profile: pixels are converted to sRGB. MatrixTests.everyPath —
apng jpeg — image supported lossy colorProfile, dimensions Animated sources: only the first frame is kept. Lossy compression; the PSNR of the result is measured. JPEG has no transparency: transparent areas are filled with the background colour. Samples deeper than 8-bit are reduced to 8-bit. Largest side: 65535 pixels. MatrixTests.everyPath —
apng pbm — image supported lossy dimensions Animated sources: only the first frame is kept. PBM has no transparency: transparent areas are filled with the background colour. Samples deeper than 8-bit are reduced to 8-bit. No embedded colour profile: pixels are converted to sRGB. Every pixel becomes black or white (50 % threshold). MatrixTests.everyPath —
apng pgm — image supported lossy dimensions Animated sources: only the first frame is kept. PGM has no transparency: transparent areas are filled with the background colour. No embedded colour profile: pixels are converted to sRGB. Colour is reduced to gray (luma). MatrixTests.everyPath —
apng png — image supported lossless colorProfile, dimensions, transparency Animated sources: only the first frame is kept. MatrixTests.everyPath —
apng ppm — image supported lossy dimensions Animated sources: only the first frame is kept. PPM has no transparency: transparent areas are filled with the background colour. No embedded colour profile: pixels are converted to sRGB. MatrixTests.everyPath —
apng psd — image supported lossy colorProfile, dimensions, transparency Animated sources: only the first frame is kept. A flat, single-layer PSD is written; partially transparent pixels are stored premultiplied. MatrixTests.everyPath —
apng tga — image supported lossy dimensions, transparency Animated sources: only the first frame is kept. Samples deeper than 8-bit are reduced to 8-bit. No embedded colour profile: pixels are converted to sRGB. MatrixTests.everyPath —
apng tiff — image supported lossless colorProfile, dimensions, transparency Animated sources: only the first frame is kept. MatrixTests.everyPath —
apng webp — image supported lossy animation, colorProfile, dimensions, transparency Lossy by default (a lossless mode is available); the PSNR of the result is measured. Samples deeper than 8-bit are reduced to 8-bit. Largest side: 16383 pixels. MatrixTests.everyPath —
avif apng — image supported lossless colorProfile, dimensions, transparency EXIF, IPTC and XMP metadata are not written. A still source gives a single-frame PNG. MatrixTests.everyPath —
avif avif — image supported lossy colorProfile, dimensions, metadata, transparency Lossy compression; the PSNR of the result is measured. At most 10 bits per component. MatrixTests.everyPath —
avif bmp — image supported lossy dimensions, transparency Samples deeper than 8-bit are reduced to 8-bit. No embedded colour profile: pixels are converted to sRGB. EXIF, IPTC and XMP metadata are not written. Transparency is stored in a 32-bit BMP; some programs ignore BMP alpha. MatrixTests.everyPath —
avif exr — image supported lossy dimensions, transparency Pixels are stored as linear values with sRGB / Rec. 709 primaries. EXIF, IPTC and XMP metadata are not written. 8-bit sources are stored as 16-bit half floats, deeper sources as 32-bit floats. MatrixTests.everyPath —
avif gif — image supported lossy dimensions, transparency GIF stores at most 256 colours per frame and on/off transparency. Samples deeper than 8-bit are reduced to 8-bit. No embedded colour profile: pixels are converted to sRGB. EXIF, IPTC and XMP metadata are not written. MatrixTests.everyPath —
avif hdr — image supported lossy dimensions HDR has no transparency: transparent areas are filled with the background colour. Pixels are stored as linear values with sRGB / Rec. 709 primaries. EXIF, IPTC and XMP metadata are not written. RGBE keeps an 8-bit mantissa per channel. MatrixTests.everyPath —
avif heic — image supported lossy colorProfile, dimensions, metadata, transparency Lossy compression; the PSNR of the result is measured. At most 10 bits per component. MatrixTests.everyPath —
avif heif — image supported lossy colorProfile, dimensions, metadata, transparency Lossy compression; the PSNR of the result is measured. At most 10 bits per component. Written as an HEVC-coded HEIF file (the same encoding as HEIC). MatrixTests.everyPath —
avif icns — image supported lossy transparency Samples deeper than 8-bit are reduced to 8-bit. No embedded colour profile: pixels are converted to sRGB. EXIF, IPTC and XMP metadata are not written. The image is resampled to the standard icon sizes, fitted in a transparent square. MatrixTests.everyPath —
avif ico — image supported lossy transparency Samples deeper than 8-bit are reduced to 8-bit. No embedded colour profile: pixels are converted to sRGB. EXIF, IPTC and XMP metadata are not written. The image is resampled to the standard icon sizes, fitted in a transparent square. MatrixTests.everyPath —
avif jp2 — image supported lossy dimensions, transparency Lossy by default (a lossless mode is available); the PSNR of the result is measured. Samples deeper than 8-bit are reduced to 8-bit. No embedded colour profile: pixels are converted to sRGB. EXIF, IPTC and XMP metadata are not written. MatrixTests.everyPath —
avif jpeg — image supported lossy colorProfile, dimensions, metadata Lossy compression; the PSNR of the result is measured. JPEG has no transparency: transparent areas are filled with the background colour. Samples deeper than 8-bit are reduced to 8-bit. Largest side: 65535 pixels. MatrixTests.everyPath —
avif pbm — image supported lossy dimensions PBM has no transparency: transparent areas are filled with the background colour. Samples deeper than 8-bit are reduced to 8-bit. No embedded colour profile: pixels are converted to sRGB. EXIF, IPTC and XMP metadata are not written. Every pixel becomes black or white (50 % threshold). MatrixTests.everyPath —
avif pdf image.vectorize vector supported lossy dimensions, transparency The drawing is an approximation made of flat colour shapes: gradients, textures and photographs become bands of colour. Partial transparency is not kept: every shape is opaque. Only the first frame of an animated or multi-page image is traced. Text becomes outlines; it is not recognised as text. Metadata and colour profiles are not carried over (colours are converted to sRGB). tracesEverySource("avif") —
avif pgm — image supported lossy dimensions PGM has no transparency: transparent areas are filled with the background colour. No embedded colour profile: pixels are converted to sRGB. EXIF, IPTC and XMP metadata are not written. Colour is reduced to gray (luma). MatrixTests.everyPath —
avif png — image supported lossless colorProfile, dimensions, metadata, transparency — MatrixTests.everyPath —
avif ppm — image supported lossy dimensions PPM has no transparency: transparent areas are filled with the background colour. No embedded colour profile: pixels are converted to sRGB. EXIF, IPTC and XMP metadata are not written. MatrixTests.everyPath —
avif psd — image supported lossy colorProfile, dimensions, transparency EXIF, IPTC and XMP metadata are not written. A flat, single-layer PSD is written; partially transparent pixels are stored premultiplied. MatrixTests.everyPath —
avif svg image.vectorize vector supported lossy dimensions, transparency The drawing is an approximation made of flat colour shapes: gradients, textures and photographs become bands of colour. Partial transparency is not kept: every shape is opaque. Only the first frame of an animated or multi-page image is traced. Text becomes outlines; it is not recognised as text. Metadata and colour profiles are not carried over (colours are converted to sRGB). tracesEverySource("avif") —
avif tga — image supported lossy dimensions, transparency Samples deeper than 8-bit are reduced to 8-bit. No embedded colour profile: pixels are converted to sRGB. EXIF, IPTC and XMP metadata are not written. MatrixTests.everyPath —
avif tiff — image supported lossless colorProfile, dimensions, metadata, transparency — MatrixTests.everyPath —
avif webp — image supported lossy colorProfile, dimensions, metadata, transparency Lossy by default (a lossless mode is available); the PSNR of the result is measured. Samples deeper than 8-bit are reduced to 8-bit. Largest side: 16383 pixels. MatrixTests.everyPath —
bmp apng — image supported lossless colorProfile, dimensions, transparency A still source gives a single-frame PNG. MatrixTests.everyPath —
bmp avif — image supported lossy colorProfile, dimensions, transparency Lossy compression; the PSNR of the result is measured. At most 10 bits per component. MatrixTests.everyPath —
bmp bmp — image supported lossless dimensions, transparency No embedded colour profile: pixels are converted to sRGB. Transparency is stored in a 32-bit BMP; some programs ignore BMP alpha. MatrixTests.everyPath —
bmp exr — image supported lossy dimensions, transparency Pixels are stored as linear values with sRGB / Rec. 709 primaries. 8-bit sources are stored as 16-bit half floats, deeper sources as 32-bit floats. MatrixTests.everyPath —
bmp gif — image supported lossy dimensions, transparency GIF stores at most 256 colours per frame and on/off transparency. No embedded colour profile: pixels are converted to sRGB. MatrixTests.everyPath —
bmp hdr — image supported lossy dimensions HDR has no transparency: transparent areas are filled with the background colour. Pixels are stored as linear values with sRGB / Rec. 709 primaries. RGBE keeps an 8-bit mantissa per channel. MatrixTests.everyPath —
bmp heic — image supported lossy colorProfile, dimensions, transparency Lossy compression; the PSNR of the result is measured. At most 10 bits per component. MatrixTests.everyPath —
bmp heif — image supported lossy colorProfile, dimensions, transparency Lossy compression; the PSNR of the result is measured. At most 10 bits per component. Written as an HEVC-coded HEIF file (the same encoding as HEIC). MatrixTests.everyPath —
bmp icns — image supported lossy transparency No embedded colour profile: pixels are converted to sRGB. The image is resampled to the standard icon sizes, fitted in a transparent square. MatrixTests.everyPath —
bmp ico — image supported lossy transparency No embedded colour profile: pixels are converted to sRGB. The image is resampled to the standard icon sizes, fitted in a transparent square. MatrixTests.everyPath —
bmp jp2 — image supported lossy dimensions, transparency Lossy by default (a lossless mode is available); the PSNR of the result is measured. No embedded colour profile: pixels are converted to sRGB. MatrixTests.everyPath —
bmp jpeg — image supported lossy colorProfile, dimensions Lossy compression; the PSNR of the result is measured. JPEG has no transparency: transparent areas are filled with the background colour. Largest side: 65535 pixels. MatrixTests.everyPath —
bmp pbm — image supported lossy dimensions PBM has no transparency: transparent areas are filled with the background colour. No embedded colour profile: pixels are converted to sRGB. Every pixel becomes black or white (50 % threshold). MatrixTests.everyPath —
bmp pdf — pdf supported lossless dimensions One image per page; EXIF orientation is applied. Other metadata (EXIF, GPS, captions) is not copied into the PDF. With “Recompress as JPEG”, opaque images become lossy JPEG; images with transparency always stay lossless. ImagesIntoPDF/everySourceFormat (7 cases), jpegIsEmbeddedWithoutRecompression, losslessByDefaultAndJPEGOnRequest, exifOrientationIsApplied (4 cases), pageSizesMarginsOrderAndPasswords, optionalTextLayer, rejectsWhatIsNotAnImage —
bmp pdf image.vectorize vector supported lossy dimensions, transparency The drawing is an approximation made of flat colour shapes: gradients, textures and photographs become bands of colour. Partial transparency is not kept: every shape is opaque. Only the first frame of an animated or multi-page image is traced. Text becomes outlines; it is not recognised as text. Metadata and colour profiles are not carried over (colours are converted to sRGB). tracesEverySource("bmp") —
bmp pgm — image supported lossy dimensions PGM has no transparency: transparent areas are filled with the background colour. No embedded colour profile: pixels are converted to sRGB. Colour is reduced to gray (luma). MatrixTests.everyPath —
bmp png — image supported lossless colorProfile, dimensions, transparency — MatrixTests.everyPath —
bmp ppm — image supported lossy dimensions PPM has no transparency: transparent areas are filled with the background colour. No embedded colour profile: pixels are converted to sRGB. MatrixTests.everyPath —
bmp psd — image supported lossy colorProfile, dimensions, transparency A flat, single-layer PSD is written; partially transparent pixels are stored premultiplied. MatrixTests.everyPath —
bmp svg image.vectorize vector supported lossy dimensions, transparency The drawing is an approximation made of flat colour shapes: gradients, textures and photographs become bands of colour. Partial transparency is not kept: every shape is opaque. Only the first frame of an animated or multi-page image is traced. Text becomes outlines; it is not recognised as text. Metadata and colour profiles are not carried over (colours are converted to sRGB). tracesEverySource("bmp") —
bmp tga — image supported lossless dimensions, transparency No embedded colour profile: pixels are converted to sRGB. MatrixTests.everyPath —
bmp tiff — image supported lossless colorProfile, dimensions, transparency — MatrixTests.everyPath —
bmp webp — image supported lossy colorProfile, dimensions, transparency Lossy by default (a lossless mode is available); the PSNR of the result is measured. Largest side: 16383 pixels. MatrixTests.everyPath —
dng apng — image experimental lossy colorProfile, dimensions RAW development is Apple's default rendering (ImageIO): no exposure, white-balance or noise controls. Tested only with a linear DNG generated in code; camera DNG files rely on the camera support of macOS. EXIF, IPTC and XMP metadata are not written. A still source gives a single-frame PNG. MatrixTests.everyPath —
dng avif — image experimental lossy colorProfile, dimensions, metadata RAW development is Apple's default rendering (ImageIO): no exposure, white-balance or noise controls. Tested only with a linear DNG generated in code; camera DNG files rely on the camera support of macOS. Lossy compression; the PSNR of the result is measured. At most 10 bits per component. MatrixTests.everyPath —
dng bmp — image experimental lossy dimensions RAW development is Apple's default rendering (ImageIO): no exposure, white-balance or noise controls. Tested only with a linear DNG generated in code; camera DNG files rely on the camera support of macOS. Samples deeper than 8-bit are reduced to 8-bit. No embedded colour profile: pixels are converted to sRGB. EXIF, IPTC and XMP metadata are not written. MatrixTests.everyPath —
dng exr — image experimental lossy dimensions RAW development is Apple's default rendering (ImageIO): no exposure, white-balance or noise controls. Tested only with a linear DNG generated in code; camera DNG files rely on the camera support of macOS. Pixels are stored as linear values with sRGB / Rec. 709 primaries. EXIF, IPTC and XMP metadata are not written. 8-bit sources are stored as 16-bit half floats, deeper sources as 32-bit floats. MatrixTests.everyPath —
dng gif — image experimental lossy dimensions RAW development is Apple's default rendering (ImageIO): no exposure, white-balance or noise controls. Tested only with a linear DNG generated in code; camera DNG files rely on the camera support of macOS. GIF stores at most 256 colours per frame and on/off transparency. Samples deeper than 8-bit are reduced to 8-bit. No embedded colour profile: pixels are converted to sRGB. EXIF, IPTC and XMP metadata are not written. MatrixTests.everyPath —
dng hdr — image experimental lossy dimensions RAW development is Apple's default rendering (ImageIO): no exposure, white-balance or noise controls. Tested only with a linear DNG generated in code; camera DNG files rely on the camera support of macOS. Pixels are stored as linear values with sRGB / Rec. 709 primaries. EXIF, IPTC and XMP metadata are not written. RGBE keeps an 8-bit mantissa per channel. MatrixTests.everyPath —
dng heic — image experimental lossy colorProfile, dimensions, metadata RAW development is Apple's default rendering (ImageIO): no exposure, white-balance or noise controls. Tested only with a linear DNG generated in code; camera DNG files rely on the camera support of macOS. Lossy compression; the PSNR of the result is measured. At most 10 bits per component. MatrixTests.everyPath —
dng heif — image experimental lossy colorProfile, dimensions, metadata RAW development is Apple's default rendering (ImageIO): no exposure, white-balance or noise controls. Tested only with a linear DNG generated in code; camera DNG files rely on the camera support of macOS. Lossy compression; the PSNR of the result is measured. At most 10 bits per component. Written as an HEVC-coded HEIF file (the same encoding as HEIC). MatrixTests.everyPath —
dng icns — image experimental lossy — RAW development is Apple's default rendering (ImageIO): no exposure, white-balance or noise controls. Tested only with a linear DNG generated in code; camera DNG files rely on the camera support of macOS. Samples deeper than 8-bit are reduced to 8-bit. No embedded colour profile: pixels are converted to sRGB. EXIF, IPTC and XMP metadata are not written. The image is resampled to the standard icon sizes, fitted in a transparent square. MatrixTests.everyPath —
dng ico — image experimental lossy — RAW development is Apple's default rendering (ImageIO): no exposure, white-balance or noise controls. Tested only with a linear DNG generated in code; camera DNG files rely on the camera support of macOS. Samples deeper than 8-bit are reduced to 8-bit. No embedded colour profile: pixels are converted to sRGB. EXIF, IPTC and XMP metadata are not written. The image is resampled to the standard icon sizes, fitted in a transparent square. MatrixTests.everyPath —
dng jp2 — image experimental lossy dimensions RAW development is Apple's default rendering (ImageIO): no exposure, white-balance or noise controls. Tested only with a linear DNG generated in code; camera DNG files rely on the camera support of macOS. Lossy by default (a lossless mode is available); the PSNR of the result is measured. Samples deeper than 8-bit are reduced to 8-bit. No embedded colour profile: pixels are converted to sRGB. EXIF, IPTC and XMP metadata are not written. MatrixTests.everyPath —
dng jpeg — image experimental lossy colorProfile, dimensions, metadata RAW development is Apple's default rendering (ImageIO): no exposure, white-balance or noise controls. Tested only with a linear DNG generated in code; camera DNG files rely on the camera support of macOS. Lossy compression; the PSNR of the result is measured. Samples deeper than 8-bit are reduced to 8-bit. Largest side: 65535 pixels. MatrixTests.everyPath —
dng pbm — image experimental lossy dimensions RAW development is Apple's default rendering (ImageIO): no exposure, white-balance or noise controls. Tested only with a linear DNG generated in code; camera DNG files rely on the camera support of macOS. Samples deeper than 8-bit are reduced to 8-bit. No embedded colour profile: pixels are converted to sRGB. EXIF, IPTC and XMP metadata are not written. Every pixel becomes black or white (50 % threshold). MatrixTests.everyPath —
dng pgm — image experimental lossy dimensions RAW development is Apple's default rendering (ImageIO): no exposure, white-balance or noise controls. Tested only with a linear DNG generated in code; camera DNG files rely on the camera support of macOS. No embedded colour profile: pixels are converted to sRGB. EXIF, IPTC and XMP metadata are not written. Colour is reduced to gray (luma). MatrixTests.everyPath —
dng png — image experimental lossy colorProfile, dimensions, metadata RAW development is Apple's default rendering (ImageIO): no exposure, white-balance or noise controls. Tested only with a linear DNG generated in code; camera DNG files rely on the camera support of macOS. MatrixTests.everyPath —
dng ppm — image experimental lossy dimensions RAW development is Apple's default rendering (ImageIO): no exposure, white-balance or noise controls. Tested only with a linear DNG generated in code; camera DNG files rely on the camera support of macOS. No embedded colour profile: pixels are converted to sRGB. EXIF, IPTC and XMP metadata are not written. MatrixTests.everyPath —
dng psd — image experimental lossy colorProfile, dimensions RAW development is Apple's default rendering (ImageIO): no exposure, white-balance or noise controls. Tested only with a linear DNG generated in code; camera DNG files rely on the camera support of macOS. EXIF, IPTC and XMP metadata are not written. A flat, single-layer PSD is written; partially transparent pixels are stored premultiplied. MatrixTests.everyPath —
dng tga — image experimental lossy dimensions RAW development is Apple's default rendering (ImageIO): no exposure, white-balance or noise controls. Tested only with a linear DNG generated in code; camera DNG files rely on the camera support of macOS. Samples deeper than 8-bit are reduced to 8-bit. No embedded colour profile: pixels are converted to sRGB. EXIF, IPTC and XMP metadata are not written. MatrixTests.everyPath —
dng tiff — image experimental lossy colorProfile, dimensions, metadata RAW development is Apple's default rendering (ImageIO): no exposure, white-balance or noise controls. Tested only with a linear DNG generated in code; camera DNG files rely on the camera support of macOS. MatrixTests.everyPath —
dng webp — image experimental lossy colorProfile, dimensions, metadata RAW development is Apple's default rendering (ImageIO): no exposure, white-balance or noise controls. Tested only with a linear DNG generated in code; camera DNG files rely on the camera support of macOS. Lossy by default (a lossless mode is available); the PSNR of the result is measured. Samples deeper than 8-bit are reduced to 8-bit. Largest side: 16383 pixels. MatrixTests.everyPath —
exr apng — image supported lossy colorProfile, dimensions, transparency HDR values are tone-mapped into the SDR range. Samples deeper than 16-bit are reduced to 16-bit. A still source gives a single-frame PNG. MatrixTests.everyPath —
exr avif — image supported lossy colorProfile, dimensions, transparency Lossy compression; the PSNR of the result is measured. HDR values are tone-mapped into the SDR range. Samples deeper than 16-bit are reduced to 16-bit. At most 10 bits per component. MatrixTests.everyPath —
exr bmp — image supported lossy dimensions, transparency HDR values are tone-mapped into the SDR range. Samples deeper than 8-bit are reduced to 8-bit. No embedded colour profile: pixels are converted to sRGB. Transparency is stored in a 32-bit BMP; some programs ignore BMP alpha. MatrixTests.everyPath —
exr exr — image supported lossy dimensions, hdr, transparency Pixels are stored as linear values with sRGB / Rec. 709 primaries. 8-bit sources are stored as 16-bit half floats, deeper sources as 32-bit floats. MatrixTests.everyPath —
exr gif — image supported lossy dimensions, transparency GIF stores at most 256 colours per frame and on/off transparency. HDR values are tone-mapped into the SDR range. Samples deeper than 8-bit are reduced to 8-bit. No embedded colour profile: pixels are converted to sRGB. MatrixTests.everyPath —
exr hdr — image supported lossy dimensions, hdr HDR has no transparency: transparent areas are filled with the background colour. Pixels are stored as linear values with sRGB / Rec. 709 primaries. RGBE keeps an 8-bit mantissa per channel. MatrixTests.everyPath —
exr heic — image supported lossy colorProfile, dimensions, transparency Lossy compression; the PSNR of the result is measured. HDR values are tone-mapped into the SDR range. Samples deeper than 16-bit are reduced to 16-bit. At most 10 bits per component. MatrixTests.everyPath —
exr heif — image supported lossy colorProfile, dimensions, transparency Lossy compression; the PSNR of the result is measured. HDR values are tone-mapped into the SDR range. Samples deeper than 16-bit are reduced to 16-bit. At most 10 bits per component. Written as an HEVC-coded HEIF file (the same encoding as HEIC). MatrixTests.everyPath —
exr icns — image supported lossy transparency HDR values are tone-mapped into the SDR range. Samples deeper than 8-bit are reduced to 8-bit. No embedded colour profile: pixels are converted to sRGB. The image is resampled to the standard icon sizes, fitted in a transparent square. MatrixTests.everyPath —
exr ico — image supported lossy transparency HDR values are tone-mapped into the SDR range. Samples deeper than 8-bit are reduced to 8-bit. No embedded colour profile: pixels are converted to sRGB. The image is resampled to the standard icon sizes, fitted in a transparent square. MatrixTests.everyPath —
exr jp2 — image supported lossy dimensions, transparency Lossy by default (a lossless mode is available); the PSNR of the result is measured. HDR values are tone-mapped into the SDR range. Samples deeper than 8-bit are reduced to 8-bit. No embedded colour profile: pixels are converted to sRGB. MatrixTests.everyPath —
exr jpeg — image supported lossy colorProfile, dimensions Lossy compression; the PSNR of the result is measured. JPEG has no transparency: transparent areas are filled with the background colour. HDR values are tone-mapped into the SDR range. Samples deeper than 8-bit are reduced to 8-bit. Largest side: 65535 pixels. MatrixTests.everyPath —
exr pbm — image supported lossy dimensions PBM has no transparency: transparent areas are filled with the background colour. HDR values are tone-mapped into the SDR range. Samples deeper than 8-bit are reduced to 8-bit. No embedded colour profile: pixels are converted to sRGB. Every pixel becomes black or white (50 % threshold). MatrixTests.everyPath —
exr pgm — image supported lossy dimensions PGM has no transparency: transparent areas are filled with the background colour. HDR values are tone-mapped into the SDR range. Samples deeper than 16-bit are reduced to 16-bit. No embedded colour profile: pixels are converted to sRGB. Colour is reduced to gray (luma). MatrixTests.everyPath —
exr png — image supported lossy colorProfile, dimensions, transparency HDR values are tone-mapped into the SDR range. Samples deeper than 16-bit are reduced to 16-bit. MatrixTests.everyPath —
exr ppm — image supported lossy dimensions PPM has no transparency: transparent areas are filled with the background colour. HDR values are tone-mapped into the SDR range. Samples deeper than 16-bit are reduced to 16-bit. No embedded colour profile: pixels are converted to sRGB. MatrixTests.everyPath —
exr psd — image supported lossy colorProfile, dimensions, transparency HDR values are tone-mapped into the SDR range. Samples deeper than 16-bit are reduced to 16-bit. A flat, single-layer PSD is written; partially transparent pixels are stored premultiplied. MatrixTests.everyPath —
exr tga — image supported lossy dimensions, transparency HDR values are tone-mapped into the SDR range. Samples deeper than 8-bit are reduced to 8-bit. No embedded colour profile: pixels are converted to sRGB. MatrixTests.everyPath —
exr tiff — image supported lossless colorProfile, dimensions, hdr, transparency — MatrixTests.everyPath —
exr webp — image supported lossy colorProfile, dimensions, transparency Lossy by default (a lossless mode is available); the PSNR of the result is measured. HDR values are tone-mapped into the SDR range. Samples deeper than 8-bit are reduced to 8-bit. Largest side: 16383 pixels. MatrixTests.everyPath —
gif apng — image supported lossless animation, colorProfile, dimensions, transparency — MatrixTests.everyPath —
gif avif — image supported lossy colorProfile, dimensions, transparency Animated sources: only the first frame is kept. Lossy compression; the PSNR of the result is measured. At most 10 bits per component. MatrixTests.everyPath —
gif bmp — image supported lossless dimensions, transparency Animated sources: only the first frame is kept. No embedded colour profile: pixels are converted to sRGB. Transparency is stored in a 32-bit BMP; some programs ignore BMP alpha. MatrixTests.everyPath —
gif exr — image supported lossy dimensions, transparency Animated sources: only the first frame is kept. Pixels are stored as linear values with sRGB / Rec. 709 primaries. 8-bit sources are stored as 16-bit half floats, deeper sources as 32-bit floats. MatrixTests.everyPath —
gif gif — image supported lossy animation, dimensions, transparency GIF stores at most 256 colours per frame and on/off transparency. No embedded colour profile: pixels are converted to sRGB. MatrixTests.everyPath —
gif hdr — image supported lossy dimensions Animated sources: only the first frame is kept. HDR has no transparency: transparent areas are filled with the background colour. Pixels are stored as linear values with sRGB / Rec. 709 primaries. RGBE keeps an 8-bit mantissa per channel. MatrixTests.everyPath —
gif heic — image supported lossy colorProfile, dimensions, transparency Animated sources: only the first frame is kept. Lossy compression; the PSNR of the result is measured. At most 10 bits per component. MatrixTests.everyPath —
gif heif — image supported lossy colorProfile, dimensions, transparency Animated sources: only the first frame is kept. Lossy compression; the PSNR of the result is measured. At most 10 bits per component. Written as an HEVC-coded HEIF file (the same encoding as HEIC). MatrixTests.everyPath —
gif icns — image supported lossy transparency Animated sources: only the first frame is kept. No embedded colour profile: pixels are converted to sRGB. The image is resampled to the standard icon sizes, fitted in a transparent square. MatrixTests.everyPath —
gif ico — image supported lossy transparency Animated sources: only the first frame is kept. No embedded colour profile: pixels are converted to sRGB. The image is resampled to the standard icon sizes, fitted in a transparent square. MatrixTests.everyPath —
gif jp2 — image supported lossy dimensions, transparency Animated sources: only the first frame is kept. Lossy by default (a lossless mode is available); the PSNR of the result is measured. No embedded colour profile: pixels are converted to sRGB. MatrixTests.everyPath —
gif jpeg — image supported lossy colorProfile, dimensions Animated sources: only the first frame is kept. Lossy compression; the PSNR of the result is measured. JPEG has no transparency: transparent areas are filled with the background colour. Largest side: 65535 pixels. MatrixTests.everyPath —
gif pbm — image supported lossy dimensions Animated sources: only the first frame is kept. PBM has no transparency: transparent areas are filled with the background colour. No embedded colour profile: pixels are converted to sRGB. Every pixel becomes black or white (50 % threshold). MatrixTests.everyPath —
gif pdf — pdf supported lossless dimensions One image per page; EXIF orientation is applied. Other metadata (EXIF, GPS, captions) is not copied into the PDF. With “Recompress as JPEG”, opaque images become lossy JPEG; images with transparency always stay lossless. Animated or multi-image files contribute their first frame only. ImagesIntoPDF/everySourceFormat (7 cases), jpegIsEmbeddedWithoutRecompression, losslessByDefaultAndJPEGOnRequest, exifOrientationIsApplied (4 cases), pageSizesMarginsOrderAndPasswords, optionalTextLayer, rejectsWhatIsNotAnImage —
gif pdf image.vectorize vector supported lossy dimensions, transparency The drawing is an approximation made of flat colour shapes: gradients, textures and photographs become bands of colour. Partial transparency is not kept: every shape is opaque. Only the first frame of an animated or multi-page image is traced. Text becomes outlines; it is not recognised as text. Metadata and colour profiles are not carried over (colours are converted to sRGB). tracesEverySource("gif") —
gif pgm — image supported lossy dimensions Animated sources: only the first frame is kept. PGM has no transparency: transparent areas are filled with the background colour. No embedded colour profile: pixels are converted to sRGB. Colour is reduced to gray (luma). MatrixTests.everyPath —
gif png — image supported lossless colorProfile, dimensions, transparency Animated sources: only the first frame is kept. MatrixTests.everyPath —
gif ppm — image supported lossy dimensions Animated sources: only the first frame is kept. PPM has no transparency: transparent areas are filled with the background colour. No embedded colour profile: pixels are converted to sRGB. MatrixTests.everyPath —
gif psd — image supported lossy colorProfile, dimensions, transparency Animated sources: only the first frame is kept. A flat, single-layer PSD is written; partially transparent pixels are stored premultiplied. MatrixTests.everyPath —
gif svg image.vectorize vector supported lossy dimensions, transparency The drawing is an approximation made of flat colour shapes: gradients, textures and photographs become bands of colour. Partial transparency is not kept: every shape is opaque. Only the first frame of an animated or multi-page image is traced. Text becomes outlines; it is not recognised as text. Metadata and colour profiles are not carried over (colours are converted to sRGB). tracesEverySource("gif") —
gif tga — image supported lossless dimensions, transparency Animated sources: only the first frame is kept. No embedded colour profile: pixels are converted to sRGB. MatrixTests.everyPath —
gif tiff — image supported lossless colorProfile, dimensions, transparency Animated sources: only the first frame is kept. MatrixTests.everyPath —
gif webp — image supported lossy animation, colorProfile, dimensions, transparency Lossy by default (a lossless mode is available); the PSNR of the result is measured. Largest side: 16383 pixels. MatrixTests.everyPath —
hdr apng — image supported lossy colorProfile, dimensions HDR values are tone-mapped into the SDR range. Samples deeper than 16-bit are reduced to 16-bit. A still source gives a single-frame PNG. MatrixTests.everyPath —
hdr avif — image supported lossy colorProfile, dimensions Lossy compression; the PSNR of the result is measured. HDR values are tone-mapped into the SDR range. Samples deeper than 16-bit are reduced to 16-bit. At most 10 bits per component. MatrixTests.everyPath —
hdr bmp — image supported lossy dimensions HDR values are tone-mapped into the SDR range. Samples deeper than 8-bit are reduced to 8-bit. No embedded colour profile: pixels are converted to sRGB. MatrixTests.everyPath —
hdr exr — image supported lossy dimensions, hdr Pixels are stored as linear values with sRGB / Rec. 709 primaries. 8-bit sources are stored as 16-bit half floats, deeper sources as 32-bit floats. MatrixTests.everyPath —
hdr gif — image supported lossy dimensions GIF stores at most 256 colours per frame and on/off transparency. HDR values are tone-mapped into the SDR range. Samples deeper than 8-bit are reduced to 8-bit. No embedded colour profile: pixels are converted to sRGB. MatrixTests.everyPath —
hdr hdr — image supported lossy dimensions, hdr Pixels are stored as linear values with sRGB / Rec. 709 primaries. RGBE keeps an 8-bit mantissa per channel. MatrixTests.everyPath —
hdr heic — image supported lossy colorProfile, dimensions Lossy compression; the PSNR of the result is measured. HDR values are tone-mapped into the SDR range. Samples deeper than 16-bit are reduced to 16-bit. At most 10 bits per component. MatrixTests.everyPath —
hdr heif — image supported lossy colorProfile, dimensions Lossy compression; the PSNR of the result is measured. HDR values are tone-mapped into the SDR range. Samples deeper than 16-bit are reduced to 16-bit. At most 10 bits per component. Written as an HEVC-coded HEIF file (the same encoding as HEIC). MatrixTests.everyPath —
hdr icns — image supported lossy — HDR values are tone-mapped into the SDR range. Samples deeper than 8-bit are reduced to 8-bit. No embedded colour profile: pixels are converted to sRGB. The image is resampled to the standard icon sizes, fitted in a transparent square. MatrixTests.everyPath —
hdr ico — image supported lossy — HDR values are tone-mapped into the SDR range. Samples deeper than 8-bit are reduced to 8-bit. No embedded colour profile: pixels are converted to sRGB. The image is resampled to the standard icon sizes, fitted in a transparent square. MatrixTests.everyPath —
hdr jp2 — image supported lossy dimensions Lossy by default (a lossless mode is available); the PSNR of the result is measured. HDR values are tone-mapped into the SDR range. Samples deeper than 8-bit are reduced to 8-bit. No embedded colour profile: pixels are converted to sRGB. MatrixTests.everyPath —
hdr jpeg — image supported lossy colorProfile, dimensions Lossy compression; the PSNR of the result is measured. HDR values are tone-mapped into the SDR range. Samples deeper than 8-bit are reduced to 8-bit. Largest side: 65535 pixels. MatrixTests.everyPath —
hdr pbm — image supported lossy dimensions HDR values are tone-mapped into the SDR range. Samples deeper than 8-bit are reduced to 8-bit. No embedded colour profile: pixels are converted to sRGB. Every pixel becomes black or white (50 % threshold). MatrixTests.everyPath —
hdr pgm — image supported lossy dimensions HDR values are tone-mapped into the SDR range. Samples deeper than 16-bit are reduced to 16-bit. No embedded colour profile: pixels are converted to sRGB. Colour is reduced to gray (luma). MatrixTests.everyPath —
hdr png — image supported lossy colorProfile, dimensions HDR values are tone-mapped into the SDR range. Samples deeper than 16-bit are reduced to 16-bit. MatrixTests.everyPath —
hdr ppm — image supported lossy dimensions HDR values are tone-mapped into the SDR range. Samples deeper than 16-bit are reduced to 16-bit. No embedded colour profile: pixels are converted to sRGB. MatrixTests.everyPath —
hdr psd — image supported lossy colorProfile, dimensions HDR values are tone-mapped into the SDR range. Samples deeper than 16-bit are reduced to 16-bit. A flat, single-layer PSD is written; partially transparent pixels are stored premultiplied. MatrixTests.everyPath —
hdr tga — image supported lossy dimensions HDR values are tone-mapped into the SDR range. Samples deeper than 8-bit are reduced to 8-bit. No embedded colour profile: pixels are converted to sRGB. MatrixTests.everyPath —
hdr tiff — image supported lossless colorProfile, dimensions, hdr — MatrixTests.everyPath —
hdr webp — image supported lossy colorProfile, dimensions Lossy by default (a lossless mode is available); the PSNR of the result is measured. HDR values are tone-mapped into the SDR range. Samples deeper than 8-bit are reduced to 8-bit. Largest side: 16383 pixels. MatrixTests.everyPath —
heic apng — image supported lossless colorProfile, dimensions, transparency EXIF, IPTC and XMP metadata are not written. A still source gives a single-frame PNG. MatrixTests.everyPath —
heic avif — image supported lossy colorProfile, dimensions, metadata, transparency Lossy compression; the PSNR of the result is measured. At most 10 bits per component. MatrixTests.everyPath —
heic bmp — image supported lossy dimensions, transparency Samples deeper than 8-bit are reduced to 8-bit. No embedded colour profile: pixels are converted to sRGB. EXIF, IPTC and XMP metadata are not written. Transparency is stored in a 32-bit BMP; some programs ignore BMP alpha. MatrixTests.everyPath —
heic exr — image supported lossy dimensions, transparency Pixels are stored as linear values with sRGB / Rec. 709 primaries. EXIF, IPTC and XMP metadata are not written. 8-bit sources are stored as 16-bit half floats, deeper sources as 32-bit floats. MatrixTests.everyPath —
heic gif — image supported lossy dimensions, transparency GIF stores at most 256 colours per frame and on/off transparency. Samples deeper than 8-bit are reduced to 8-bit. No embedded colour profile: pixels are converted to sRGB. EXIF, IPTC and XMP metadata are not written. MatrixTests.everyPath —
heic hdr — image supported lossy dimensions HDR has no transparency: transparent areas are filled with the background colour. Pixels are stored as linear values with sRGB / Rec. 709 primaries. EXIF, IPTC and XMP metadata are not written. RGBE keeps an 8-bit mantissa per channel. MatrixTests.everyPath —
heic heic — image supported lossy colorProfile, dimensions, metadata, transparency Lossy compression; the PSNR of the result is measured. At most 10 bits per component. MatrixTests.everyPath —
heic heif — image supported lossy colorProfile, dimensions, metadata, transparency Lossy compression; the PSNR of the result is measured. At most 10 bits per component. Written as an HEVC-coded HEIF file (the same encoding as HEIC). MatrixTests.everyPath —
heic icns — image supported lossy transparency Samples deeper than 8-bit are reduced to 8-bit. No embedded colour profile: pixels are converted to sRGB. EXIF, IPTC and XMP metadata are not written. The image is resampled to the standard icon sizes, fitted in a transparent square. MatrixTests.everyPath —
heic ico — image supported lossy transparency Samples deeper than 8-bit are reduced to 8-bit. No embedded colour profile: pixels are converted to sRGB. EXIF, IPTC and XMP metadata are not written. The image is resampled to the standard icon sizes, fitted in a transparent square. MatrixTests.everyPath —
heic jp2 — image supported lossy dimensions, transparency Lossy by default (a lossless mode is available); the PSNR of the result is measured. Samples deeper than 8-bit are reduced to 8-bit. No embedded colour profile: pixels are converted to sRGB. EXIF, IPTC and XMP metadata are not written. MatrixTests.everyPath —
heic jpeg — image supported lossy colorProfile, dimensions, metadata Lossy compression; the PSNR of the result is measured. JPEG has no transparency: transparent areas are filled with the background colour. Samples deeper than 8-bit are reduced to 8-bit. Largest side: 65535 pixels. MatrixTests.everyPath 9.19 s (wall time; HEIC → JPEG (q 85))
heic pbm — image supported lossy dimensions PBM has no transparency: transparent areas are filled with the background colour. Samples deeper than 8-bit are reduced to 8-bit. No embedded colour profile: pixels are converted to sRGB. EXIF, IPTC and XMP metadata are not written. Every pixel becomes black or white (50 % threshold). MatrixTests.everyPath —
heic pdf — pdf supported lossless dimensions One image per page; EXIF orientation is applied. Other metadata (EXIF, GPS, captions) is not copied into the PDF. With “Recompress as JPEG”, opaque images become lossy JPEG; images with transparency always stay lossless. Animated or multi-image files contribute their first frame only. ImagesIntoPDF/everySourceFormat (7 cases), jpegIsEmbeddedWithoutRecompression, losslessByDefaultAndJPEGOnRequest, exifOrientationIsApplied (4 cases), pageSizesMarginsOrderAndPasswords, optionalTextLayer, rejectsWhatIsNotAnImage —
heic pdf image.vectorize vector supported lossy dimensions, transparency The drawing is an approximation made of flat colour shapes: gradients, textures and photographs become bands of colour. Partial transparency is not kept: every shape is opaque. Only the first frame of an animated or multi-page image is traced. Text becomes outlines; it is not recognised as text. Metadata and colour profiles are not carried over (colours are converted to sRGB). tracesEverySource("heic") —
heic pgm — image supported lossy dimensions PGM has no transparency: transparent areas are filled with the background colour. No embedded colour profile: pixels are converted to sRGB. EXIF, IPTC and XMP metadata are not written. Colour is reduced to gray (luma). MatrixTests.everyPath —
heic png — image supported lossless colorProfile, dimensions, metadata, transparency — MatrixTests.everyPath —
heic ppm — image supported lossy dimensions PPM has no transparency: transparent areas are filled with the background colour. No embedded colour profile: pixels are converted to sRGB. EXIF, IPTC and XMP metadata are not written. MatrixTests.everyPath —
heic psd — image supported lossy colorProfile, dimensions, transparency EXIF, IPTC and XMP metadata are not written. A flat, single-layer PSD is written; partially transparent pixels are stored premultiplied. MatrixTests.everyPath —
heic svg image.vectorize vector supported lossy dimensions, transparency The drawing is an approximation made of flat colour shapes: gradients, textures and photographs become bands of colour. Partial transparency is not kept: every shape is opaque. Only the first frame of an animated or multi-page image is traced. Text becomes outlines; it is not recognised as text. Metadata and colour profiles are not carried over (colours are converted to sRGB). tracesEverySource("heic") —
heic tga — image supported lossy dimensions, transparency Samples deeper than 8-bit are reduced to 8-bit. No embedded colour profile: pixels are converted to sRGB. EXIF, IPTC and XMP metadata are not written. MatrixTests.everyPath —
heic tiff — image supported lossless colorProfile, dimensions, metadata, transparency — MatrixTests.everyPath —
heic webp — image supported lossy colorProfile, dimensions, metadata, transparency Lossy by default (a lossless mode is available); the PSNR of the result is measured. Samples deeper than 8-bit are reduced to 8-bit. Largest side: 16383 pixels. MatrixTests.everyPath —
heif apng — image supported lossless colorProfile, dimensions, transparency EXIF, IPTC and XMP metadata are not written. A still source gives a single-frame PNG. MatrixTests.everyPath —
heif avif — image supported lossy colorProfile, dimensions, metadata, transparency Lossy compression; the PSNR of the result is measured. At most 10 bits per component. MatrixTests.everyPath —
heif bmp — image supported lossy dimensions, transparency Samples deeper than 8-bit are reduced to 8-bit. No embedded colour profile: pixels are converted to sRGB. EXIF, IPTC and XMP metadata are not written. Transparency is stored in a 32-bit BMP; some programs ignore BMP alpha. MatrixTests.everyPath —
heif exr — image supported lossy dimensions, transparency Pixels are stored as linear values with sRGB / Rec. 709 primaries. EXIF, IPTC and XMP metadata are not written. 8-bit sources are stored as 16-bit half floats, deeper sources as 32-bit floats. MatrixTests.everyPath —
heif gif — image supported lossy dimensions, transparency GIF stores at most 256 colours per frame and on/off transparency. Samples deeper than 8-bit are reduced to 8-bit. No embedded colour profile: pixels are converted to sRGB. EXIF, IPTC and XMP metadata are not written. MatrixTests.everyPath —
heif hdr — image supported lossy dimensions HDR has no transparency: transparent areas are filled with the background colour. Pixels are stored as linear values with sRGB / Rec. 709 primaries. EXIF, IPTC and XMP metadata are not written. RGBE keeps an 8-bit mantissa per channel. MatrixTests.everyPath —
heif heic — image supported lossy colorProfile, dimensions, metadata, transparency Lossy compression; the PSNR of the result is measured. At most 10 bits per component. MatrixTests.everyPath —
heif heif — image supported lossy colorProfile, dimensions, metadata, transparency Lossy compression; the PSNR of the result is measured. At most 10 bits per component. Written as an HEVC-coded HEIF file (the same encoding as HEIC). MatrixTests.everyPath —
heif icns — image supported lossy transparency Samples deeper than 8-bit are reduced to 8-bit. No embedded colour profile: pixels are converted to sRGB. EXIF, IPTC and XMP metadata are not written. The image is resampled to the standard icon sizes, fitted in a transparent square. MatrixTests.everyPath —
heif ico — image supported lossy transparency Samples deeper than 8-bit are reduced to 8-bit. No embedded colour profile: pixels are converted to sRGB. EXIF, IPTC and XMP metadata are not written. The image is resampled to the standard icon sizes, fitted in a transparent square. MatrixTests.everyPath —
heif jp2 — image supported lossy dimensions, transparency Lossy by default (a lossless mode is available); the PSNR of the result is measured. Samples deeper than 8-bit are reduced to 8-bit. No embedded colour profile: pixels are converted to sRGB. EXIF, IPTC and XMP metadata are not written. MatrixTests.everyPath —
heif jpeg — image supported lossy colorProfile, dimensions, metadata Lossy compression; the PSNR of the result is measured. JPEG has no transparency: transparent areas are filled with the background colour. Samples deeper than 8-bit are reduced to 8-bit. Largest side: 65535 pixels. MatrixTests.everyPath —
heif pbm — image supported lossy dimensions PBM has no transparency: transparent areas are filled with the background colour. Samples deeper than 8-bit are reduced to 8-bit. No embedded colour profile: pixels are converted to sRGB. EXIF, IPTC and XMP metadata are not written. Every pixel becomes black or white (50 % threshold). MatrixTests.everyPath —
heif pgm — image supported lossy dimensions PGM has no transparency: transparent areas are filled with the background colour. No embedded colour profile: pixels are converted to sRGB. EXIF, IPTC and XMP metadata are not written. Colour is reduced to gray (luma). MatrixTests.everyPath —
heif png — image supported lossless colorProfile, dimensions, metadata, transparency — MatrixTests.everyPath —
heif ppm — image supported lossy dimensions PPM has no transparency: transparent areas are filled with the background colour. No embedded colour profile: pixels are converted to sRGB. EXIF, IPTC and XMP metadata are not written. MatrixTests.everyPath —
heif psd — image supported lossy colorProfile, dimensions, transparency EXIF, IPTC and XMP metadata are not written. A flat, single-layer PSD is written; partially transparent pixels are stored premultiplied. MatrixTests.everyPath —
heif tga — image supported lossy dimensions, transparency Samples deeper than 8-bit are reduced to 8-bit. No embedded colour profile: pixels are converted to sRGB. EXIF, IPTC and XMP metadata are not written. MatrixTests.everyPath —
heif tiff — image supported lossless colorProfile, dimensions, metadata, transparency — MatrixTests.everyPath —
heif webp — image supported lossy colorProfile, dimensions, metadata, transparency Lossy by default (a lossless mode is available); the PSNR of the result is measured. Samples deeper than 8-bit are reduced to 8-bit. Largest side: 16383 pixels. MatrixTests.everyPath —
icns apng — image supported lossless colorProfile, dimensions, transparency Only the largest representation of the icon is read. A still source gives a single-frame PNG. MatrixTests.everyPath —
icns avif — image supported lossy colorProfile, dimensions, transparency Only the largest representation of the icon is read. Lossy compression; the PSNR of the result is measured. At most 10 bits per component. MatrixTests.everyPath —
icns bmp — image supported lossless dimensions, transparency Only the largest representation of the icon is read. No embedded colour profile: pixels are converted to sRGB. Transparency is stored in a 32-bit BMP; some programs ignore BMP alpha. MatrixTests.everyPath —
icns exr — image supported lossy dimensions, transparency Only the largest representation of the icon is read. Pixels are stored as linear values with sRGB / Rec. 709 primaries. 8-bit sources are stored as 16-bit half floats, deeper sources as 32-bit floats. MatrixTests.everyPath —
icns gif — image supported lossy dimensions, transparency Only the largest representation of the icon is read. GIF stores at most 256 colours per frame and on/off transparency. No embedded colour profile: pixels are converted to sRGB. MatrixTests.everyPath —
icns hdr — image supported lossy dimensions Only the largest representation of the icon is read. HDR has no transparency: transparent areas are filled with the background colour. Pixels are stored as linear values with sRGB / Rec. 709 primaries. RGBE keeps an 8-bit mantissa per channel. MatrixTests.everyPath —
icns heic — image supported lossy colorProfile, dimensions, transparency Only the largest representation of the icon is read. Lossy compression; the PSNR of the result is measured. At most 10 bits per component. MatrixTests.everyPath —
icns heif — image supported lossy colorProfile, dimensions, transparency Only the largest representation of the icon is read. Lossy compression; the PSNR of the result is measured. At most 10 bits per component. Written as an HEVC-coded HEIF file (the same encoding as HEIC). MatrixTests.everyPath —
icns icns — image supported lossy transparency Only the largest representation of the icon is read. No embedded colour profile: pixels are converted to sRGB. The image is resampled to the standard icon sizes, fitted in a transparent square. MatrixTests.everyPath —
icns ico — image supported lossy transparency Only the largest representation of the icon is read. No embedded colour profile: pixels are converted to sRGB. The image is resampled to the standard icon sizes, fitted in a transparent square. MatrixTests.everyPath —
icns jp2 — image supported lossy dimensions, transparency Only the largest representation of the icon is read. Lossy by default (a lossless mode is available); the PSNR of the result is measured. No embedded colour profile: pixels are converted to sRGB. MatrixTests.everyPath —
icns jpeg — image supported lossy colorProfile, dimensions Only the largest representation of the icon is read. Lossy compression; the PSNR of the result is measured. JPEG has no transparency: transparent areas are filled with the background colour. Largest side: 65535 pixels. MatrixTests.everyPath —
icns pbm — image supported lossy dimensions Only the largest representation of the icon is read. PBM has no transparency: transparent areas are filled with the background colour. No embedded colour profile: pixels are converted to sRGB. Every pixel becomes black or white (50 % threshold). MatrixTests.everyPath —
icns pgm — image supported lossy dimensions Only the largest representation of the icon is read. PGM has no transparency: transparent areas are filled with the background colour. No embedded colour profile: pixels are converted to sRGB. Colour is reduced to gray (luma). MatrixTests.everyPath —
icns png — image supported lossless colorProfile, dimensions, transparency Only the largest representation of the icon is read. MatrixTests.everyPath —
icns ppm — image supported lossy dimensions Only the largest representation of the icon is read. PPM has no transparency: transparent areas are filled with the background colour. No embedded colour profile: pixels are converted to sRGB. MatrixTests.everyPath —
icns psd — image supported lossy colorProfile, dimensions, transparency Only the largest representation of the icon is read. A flat, single-layer PSD is written; partially transparent pixels are stored premultiplied. MatrixTests.everyPath —
icns tga — image supported lossless dimensions, transparency Only the largest representation of the icon is read. No embedded colour profile: pixels are converted to sRGB. MatrixTests.everyPath —
icns tiff — image supported lossless colorProfile, dimensions, transparency Only the largest representation of the icon is read. MatrixTests.everyPath —
icns webp — image supported lossy colorProfile, dimensions, transparency Only the largest representation of the icon is read. Lossy by default (a lossless mode is available); the PSNR of the result is measured. Largest side: 16383 pixels. MatrixTests.everyPath —
ico apng — image supported lossless colorProfile, dimensions, transparency Only the largest representation of the icon is read. A still source gives a single-frame PNG. MatrixTests.everyPath —
ico avif — image supported lossy colorProfile, dimensions, transparency Only the largest representation of the icon is read. Lossy compression; the PSNR of the result is measured. At most 10 bits per component. MatrixTests.everyPath —
ico bmp — image supported lossless dimensions, transparency Only the largest representation of the icon is read. No embedded colour profile: pixels are converted to sRGB. Transparency is stored in a 32-bit BMP; some programs ignore BMP alpha. MatrixTests.everyPath —
ico exr — image supported lossy dimensions, transparency Only the largest representation of the icon is read. Pixels are stored as linear values with sRGB / Rec. 709 primaries. 8-bit sources are stored as 16-bit half floats, deeper sources as 32-bit floats. MatrixTests.everyPath —
ico gif — image supported lossy dimensions, transparency Only the largest representation of the icon is read. GIF stores at most 256 colours per frame and on/off transparency. No embedded colour profile: pixels are converted to sRGB. MatrixTests.everyPath —
ico hdr — image supported lossy dimensions Only the largest representation of the icon is read. HDR has no transparency: transparent areas are filled with the background colour. Pixels are stored as linear values with sRGB / Rec. 709 primaries. RGBE keeps an 8-bit mantissa per channel. MatrixTests.everyPath —
ico heic — image supported lossy colorProfile, dimensions, transparency Only the largest representation of the icon is read. Lossy compression; the PSNR of the result is measured. At most 10 bits per component. MatrixTests.everyPath —
ico heif — image supported lossy colorProfile, dimensions, transparency Only the largest representation of the icon is read. Lossy compression; the PSNR of the result is measured. At most 10 bits per component. Written as an HEVC-coded HEIF file (the same encoding as HEIC). MatrixTests.everyPath —
ico icns — image supported lossy transparency Only the largest representation of the icon is read. No embedded colour profile: pixels are converted to sRGB. The image is resampled to the standard icon sizes, fitted in a transparent square. MatrixTests.everyPath —
ico ico — image supported lossy transparency Only the largest representation of the icon is read. No embedded colour profile: pixels are converted to sRGB. The image is resampled to the standard icon sizes, fitted in a transparent square. MatrixTests.everyPath —
ico jp2 — image supported lossy dimensions, transparency Only the largest representation of the icon is read. Lossy by default (a lossless mode is available); the PSNR of the result is measured. No embedded colour profile: pixels are converted to sRGB. MatrixTests.everyPath —
ico jpeg — image supported lossy colorProfile, dimensions Only the largest representation of the icon is read. Lossy compression; the PSNR of the result is measured. JPEG has no transparency: transparent areas are filled with the background colour. Largest side: 65535 pixels. MatrixTests.everyPath —
ico pbm — image supported lossy dimensions Only the largest representation of the icon is read. PBM has no transparency: transparent areas are filled with the background colour. No embedded colour profile: pixels are converted to sRGB. Every pixel becomes black or white (50 % threshold). MatrixTests.everyPath —
ico pgm — image supported lossy dimensions Only the largest representation of the icon is read. PGM has no transparency: transparent areas are filled with the background colour. No embedded colour profile: pixels are converted to sRGB. Colour is reduced to gray (luma). MatrixTests.everyPath —
ico png — image supported lossless colorProfile, dimensions, transparency Only the largest representation of the icon is read. MatrixTests.everyPath —
ico ppm — image supported lossy dimensions Only the largest representation of the icon is read. PPM has no transparency: transparent areas are filled with the background colour. No embedded colour profile: pixels are converted to sRGB. MatrixTests.everyPath —
ico psd — image supported lossy colorProfile, dimensions, transparency Only the largest representation of the icon is read. A flat, single-layer PSD is written; partially transparent pixels are stored premultiplied. MatrixTests.everyPath —
ico tga — image supported lossless dimensions, transparency Only the largest representation of the icon is read. No embedded colour profile: pixels are converted to sRGB. MatrixTests.everyPath —
ico tiff — image supported lossless colorProfile, dimensions, transparency Only the largest representation of the icon is read. MatrixTests.everyPath —
ico webp — image supported lossy colorProfile, dimensions, transparency Only the largest representation of the icon is read. Lossy by default (a lossless mode is available); the PSNR of the result is measured. Largest side: 16383 pixels. MatrixTests.everyPath —
jp2 apng — image supported lossless colorProfile, dimensions, transparency A still source gives a single-frame PNG. MatrixTests.everyPath —
jp2 avif — image supported lossy colorProfile, dimensions, transparency Lossy compression; the PSNR of the result is measured. At most 10 bits per component. MatrixTests.everyPath —
jp2 bmp — image supported lossy dimensions, transparency Samples deeper than 8-bit are reduced to 8-bit. No embedded colour profile: pixels are converted to sRGB. Transparency is stored in a 32-bit BMP; some programs ignore BMP alpha. MatrixTests.everyPath —
jp2 exr — image supported lossy dimensions, transparency Pixels are stored as linear values with sRGB / Rec. 709 primaries. 8-bit sources are stored as 16-bit half floats, deeper sources as 32-bit floats. MatrixTests.everyPath —
jp2 gif — image supported lossy dimensions, transparency GIF stores at most 256 colours per frame and on/off transparency. Samples deeper than 8-bit are reduced to 8-bit. No embedded colour profile: pixels are converted to sRGB. MatrixTests.everyPath —
jp2 hdr — image supported lossy dimensions HDR has no transparency: transparent areas are filled with the background colour. Pixels are stored as linear values with sRGB / Rec. 709 primaries. RGBE keeps an 8-bit mantissa per channel. MatrixTests.everyPath —
jp2 heic — image supported lossy colorProfile, dimensions, transparency Lossy compression; the PSNR of the result is measured. At most 10 bits per component. MatrixTests.everyPath —
jp2 heif — image supported lossy colorProfile, dimensions, transparency Lossy compression; the PSNR of the result is measured. At most 10 bits per component. Written as an HEVC-coded HEIF file (the same encoding as HEIC). MatrixTests.everyPath —
jp2 icns — image supported lossy transparency Samples deeper than 8-bit are reduced to 8-bit. No embedded colour profile: pixels are converted to sRGB. The image is resampled to the standard icon sizes, fitted in a transparent square. MatrixTests.everyPath —
jp2 ico — image supported lossy transparency Samples deeper than 8-bit are reduced to 8-bit. No embedded colour profile: pixels are converted to sRGB. The image is resampled to the standard icon sizes, fitted in a transparent square. MatrixTests.everyPath —
jp2 jp2 — image supported lossy dimensions, transparency Lossy by default (a lossless mode is available); the PSNR of the result is measured. Samples deeper than 8-bit are reduced to 8-bit. No embedded colour profile: pixels are converted to sRGB. MatrixTests.everyPath —
jp2 jpeg — image supported lossy colorProfile, dimensions Lossy compression; the PSNR of the result is measured. JPEG has no transparency: transparent areas are filled with the background colour. Samples deeper than 8-bit are reduced to 8-bit. Largest side: 65535 pixels. MatrixTests.everyPath —
jp2 pbm — image supported lossy dimensions PBM has no transparency: transparent areas are filled with the background colour. Samples deeper than 8-bit are reduced to 8-bit. No embedded colour profile: pixels are converted to sRGB. Every pixel becomes black or white (50 % threshold). MatrixTests.everyPath —
jp2 pdf image.vectorize vector supported lossy dimensions, transparency The drawing is an approximation made of flat colour shapes: gradients, textures and photographs become bands of colour. Partial transparency is not kept: every shape is opaque. Only the first frame of an animated or multi-page image is traced. Text becomes outlines; it is not recognised as text. Metadata and colour profiles are not carried over (colours are converted to sRGB). tracesEverySource("jp2") —
jp2 pgm — image supported lossy dimensions PGM has no transparency: transparent areas are filled with the background colour. No embedded colour profile: pixels are converted to sRGB. Colour is reduced to gray (luma). MatrixTests.everyPath —
jp2 png — image supported lossless colorProfile, dimensions, transparency — MatrixTests.everyPath —
jp2 ppm — image supported lossy dimensions PPM has no transparency: transparent areas are filled with the background colour. No embedded colour profile: pixels are converted to sRGB. MatrixTests.everyPath —
jp2 psd — image supported lossy colorProfile, dimensions, transparency A flat, single-layer PSD is written; partially transparent pixels are stored premultiplied. MatrixTests.everyPath —
jp2 svg image.vectorize vector supported lossy dimensions, transparency The drawing is an approximation made of flat colour shapes: gradients, textures and photographs become bands of colour. Partial transparency is not kept: every shape is opaque. Only the first frame of an animated or multi-page image is traced. Text becomes outlines; it is not recognised as text. Metadata and colour profiles are not carried over (colours are converted to sRGB). tracesEverySource("jp2") —
jp2 tga — image supported lossy dimensions, transparency Samples deeper than 8-bit are reduced to 8-bit. No embedded colour profile: pixels are converted to sRGB. MatrixTests.everyPath —
jp2 tiff — image supported lossless colorProfile, dimensions, transparency — MatrixTests.everyPath —
jp2 webp — image supported lossy colorProfile, dimensions, transparency Lossy by default (a lossless mode is available); the PSNR of the result is measured. Samples deeper than 8-bit are reduced to 8-bit. Largest side: 16383 pixels. MatrixTests.everyPath —
jpeg apng — image supported lossless colorProfile, dimensions EXIF, IPTC and XMP metadata are not written. A still source gives a single-frame PNG. MatrixTests.everyPath —
jpeg avif — image supported lossy colorProfile, dimensions, metadata Lossy compression; the PSNR of the result is measured. At most 10 bits per component. MatrixTests.everyPath —
jpeg bmp — image supported lossless dimensions No embedded colour profile: pixels are converted to sRGB. EXIF, IPTC and XMP metadata are not written. MatrixTests.everyPath —
jpeg exr — image supported lossy dimensions Pixels are stored as linear values with sRGB / Rec. 709 primaries. EXIF, IPTC and XMP metadata are not written. 8-bit sources are stored as 16-bit half floats, deeper sources as 32-bit floats. MatrixTests.everyPath —
jpeg gif — image supported lossy dimensions GIF stores at most 256 colours per frame and on/off transparency. No embedded colour profile: pixels are converted to sRGB. EXIF, IPTC and XMP metadata are not written. MatrixTests.everyPath —
jpeg hdr — image supported lossy dimensions Pixels are stored as linear values with sRGB / Rec. 709 primaries. EXIF, IPTC and XMP metadata are not written. RGBE keeps an 8-bit mantissa per channel. MatrixTests.everyPath —
jpeg heic — image supported lossy colorProfile, dimensions, metadata Lossy compression; the PSNR of the result is measured. At most 10 bits per component. MatrixTests.everyPath —
jpeg heif — image supported lossy colorProfile, dimensions, metadata Lossy compression; the PSNR of the result is measured. At most 10 bits per component. Written as an HEVC-coded HEIF file (the same encoding as HEIC). MatrixTests.everyPath —
jpeg icns — image supported lossy — No embedded colour profile: pixels are converted to sRGB. EXIF, IPTC and XMP metadata are not written. The image is resampled to the standard icon sizes, fitted in a transparent square. MatrixTests.everyPath —
jpeg ico — image supported lossy — No embedded colour profile: pixels are converted to sRGB. EXIF, IPTC and XMP metadata are not written. The image is resampled to the standard icon sizes, fitted in a transparent square. MatrixTests.everyPath —
jpeg jp2 — image supported lossy dimensions Lossy by default (a lossless mode is available); the PSNR of the result is measured. No embedded colour profile: pixels are converted to sRGB. EXIF, IPTC and XMP metadata are not written. MatrixTests.everyPath —
jpeg jpeg — image supported lossy colorProfile, dimensions, metadata Lossy compression; the PSNR of the result is measured. Largest side: 65535 pixels. MatrixTests.everyPath —
jpeg jpeg image.optimize image supported lossy colorProfile, dimensions, metadata The image is decoded and encoded again: a new generation of JPEG loss is added. MatrixTests.operations —
jpeg pbm — image supported lossy dimensions No embedded colour profile: pixels are converted to sRGB. EXIF, IPTC and XMP metadata are not written. Every pixel becomes black or white (50 % threshold). MatrixTests.everyPath —
jpeg pdf — pdf supported lossless dimensions One image per page; EXIF orientation is applied. Other metadata (EXIF, GPS, captions) is not copied into the PDF. With “Recompress as JPEG”, opaque images become lossy JPEG; images with transparency always stay lossless. JPEG files are embedded as their original compressed data: no generation loss. ImagesIntoPDF/everySourceFormat (7 cases), jpegIsEmbeddedWithoutRecompression, losslessByDefaultAndJPEGOnRequest, exifOrientationIsApplied (4 cases), pageSizesMarginsOrderAndPasswords, optionalTextLayer, rejectsWhatIsNotAnImage —
jpeg pdf image.vectorize vector supported lossy dimensions, transparency The drawing is an approximation made of flat colour shapes: gradients, textures and photographs become bands of colour. Partial transparency is not kept: every shape is opaque. Only the first frame of an animated or multi-page image is traced. Text becomes outlines; it is not recognised as text. Metadata and colour profiles are not carried over (colours are converted to sRGB). tracesEverySource("jpeg") —
jpeg pgm — image supported lossy dimensions No embedded colour profile: pixels are converted to sRGB. EXIF, IPTC and XMP metadata are not written. Colour is reduced to gray (luma). MatrixTests.everyPath —
jpeg png — image supported lossless colorProfile, dimensions, metadata — MatrixTests.everyPath —
jpeg ppm — image supported lossless dimensions No embedded colour profile: pixels are converted to sRGB. EXIF, IPTC and XMP metadata are not written. MatrixTests.everyPath —
jpeg psd — image supported lossy colorProfile, dimensions EXIF, IPTC and XMP metadata are not written. A flat, single-layer PSD is written; partially transparent pixels are stored premultiplied. MatrixTests.everyPath —
jpeg svg image.vectorize vector supported lossy dimensions, transparency The drawing is an approximation made of flat colour shapes: gradients, textures and photographs become bands of colour. Partial transparency is not kept: every shape is opaque. Only the first frame of an animated or multi-page image is traced. Text becomes outlines; it is not recognised as text. Metadata and colour profiles are not carried over (colours are converted to sRGB). tracesEverySource("jpeg") —
jpeg tga — image supported lossless dimensions No embedded colour profile: pixels are converted to sRGB. EXIF, IPTC and XMP metadata are not written. MatrixTests.everyPath —
jpeg tiff — image supported lossless colorProfile, dimensions, metadata — MatrixTests.everyPath —
jpeg webp — image supported lossy colorProfile, dimensions, metadata Lossy by default (a lossless mode is available); the PSNR of the result is measured. Largest side: 16383 pixels. MatrixTests.everyPath —
pbm apng — image supported lossless colorProfile, dimensions A still source gives a single-frame PNG. MatrixTests.everyPath —
pbm avif — image supported lossy colorProfile, dimensions Lossy compression; the PSNR of the result is measured. At most 10 bits per component. MatrixTests.everyPath —
pbm bmp — image supported lossless dimensions No embedded colour profile: pixels are converted to sRGB. MatrixTests.everyPath —
pbm exr — image supported lossy dimensions Pixels are stored as linear values with sRGB / Rec. 709 primaries. 8-bit sources are stored as 16-bit half floats, deeper sources as 32-bit floats. MatrixTests.everyPath —
pbm gif — image supported lossy dimensions GIF stores at most 256 colours per frame and on/off transparency. No embedded colour profile: pixels are converted to sRGB. MatrixTests.everyPath —
pbm hdr — image supported lossy dimensions Pixels are stored as linear values with sRGB / Rec. 709 primaries. RGBE keeps an 8-bit mantissa per channel. MatrixTests.everyPath —
pbm heic — image supported lossy colorProfile, dimensions Lossy compression; the PSNR of the result is measured. At most 10 bits per component. MatrixTests.everyPath —
pbm heif — image supported lossy colorProfile, dimensions Lossy compression; the PSNR of the result is measured. At most 10 bits per component. Written as an HEVC-coded HEIF file (the same encoding as HEIC). MatrixTests.everyPath —
pbm icns — image supported lossy — No embedded colour profile: pixels are converted to sRGB. The image is resampled to the standard icon sizes, fitted in a transparent square. MatrixTests.everyPath —
pbm ico — image supported lossy — No embedded colour profile: pixels are converted to sRGB. The image is resampled to the standard icon sizes, fitted in a transparent square. MatrixTests.everyPath —
pbm jp2 — image supported lossy dimensions Lossy by default (a lossless mode is available); the PSNR of the result is measured. No embedded colour profile: pixels are converted to sRGB. MatrixTests.everyPath —
pbm jpeg — image supported lossy colorProfile, dimensions Lossy compression; the PSNR of the result is measured. Largest side: 65535 pixels. MatrixTests.everyPath —
pbm pbm — image supported lossless dimensions No embedded colour profile: pixels are converted to sRGB. MatrixTests.everyPath —
pbm pgm — image supported lossless dimensions No embedded colour profile: pixels are converted to sRGB. MatrixTests.everyPath —
pbm png — image supported lossless colorProfile, dimensions — MatrixTests.everyPath —
pbm ppm — image supported lossless dimensions No embedded colour profile: pixels are converted to sRGB. MatrixTests.everyPath —
pbm psd — image supported lossy colorProfile, dimensions A flat, single-layer PSD is written; partially transparent pixels are stored premultiplied. MatrixTests.everyPath —
pbm tga — image supported lossless dimensions No embedded colour profile: pixels are converted to sRGB. MatrixTests.everyPath —
pbm tiff — image supported lossless colorProfile, dimensions — MatrixTests.everyPath —
pbm webp — image supported lossy colorProfile, dimensions Lossy by default (a lossless mode is available); the PSNR of the result is measured. Largest side: 16383 pixels. MatrixTests.everyPath —
pgm apng — image supported lossless colorProfile, dimensions A still source gives a single-frame PNG. MatrixTests.everyPath —
pgm avif — image supported lossy colorProfile, dimensions Lossy compression; the PSNR of the result is measured. At most 10 bits per component. MatrixTests.everyPath —
pgm bmp — image supported lossy dimensions Samples deeper than 8-bit are reduced to 8-bit. No embedded colour profile: pixels are converted to sRGB. MatrixTests.everyPath —
pgm exr — image supported lossy dimensions Pixels are stored as linear values with sRGB / Rec. 709 primaries. 8-bit sources are stored as 16-bit half floats, deeper sources as 32-bit floats. MatrixTests.everyPath —
pgm gif — image supported lossy dimensions GIF stores at most 256 colours per frame and on/off transparency. Samples deeper than 8-bit are reduced to 8-bit. No embedded colour profile: pixels are converted to sRGB. MatrixTests.everyPath —
pgm hdr — image supported lossy dimensions Pixels are stored as linear values with sRGB / Rec. 709 primaries. RGBE keeps an 8-bit mantissa per channel. MatrixTests.everyPath —
pgm heic — image supported lossy colorProfile, dimensions Lossy compression; the PSNR of the result is measured. At most 10 bits per component. MatrixTests.everyPath —
pgm heif — image supported lossy colorProfile, dimensions Lossy compression; the PSNR of the result is measured. At most 10 bits per component. Written as an HEVC-coded HEIF file (the same encoding as HEIC). MatrixTests.everyPath —
pgm icns — image supported lossy — Samples deeper than 8-bit are reduced to 8-bit. No embedded colour profile: pixels are converted to sRGB. The image is resampled to the standard icon sizes, fitted in a transparent square. MatrixTests.everyPath —
pgm ico — image supported lossy — Samples deeper than 8-bit are reduced to 8-bit. No embedded colour profile: pixels are converted to sRGB. The image is resampled to the standard icon sizes, fitted in a transparent square. MatrixTests.everyPath —
pgm jp2 — image supported lossy dimensions Lossy by default (a lossless mode is available); the PSNR of the result is measured. Samples deeper than 8-bit are reduced to 8-bit. No embedded colour profile: pixels are converted to sRGB. MatrixTests.everyPath —
pgm jpeg — image supported lossy colorProfile, dimensions Lossy compression; the PSNR of the result is measured. Samples deeper than 8-bit are reduced to 8-bit. Largest side: 65535 pixels. MatrixTests.everyPath —
pgm pbm — image supported lossy dimensions Samples deeper than 8-bit are reduced to 8-bit. No embedded colour profile: pixels are converted to sRGB. Every pixel becomes black or white (50 % threshold). MatrixTests.everyPath —
pgm pgm — image supported lossless dimensions No embedded colour profile: pixels are converted to sRGB. MatrixTests.everyPath —
pgm png — image supported lossless colorProfile, dimensions — MatrixTests.everyPath —
pgm ppm — image supported lossless dimensions No embedded colour profile: pixels are converted to sRGB. MatrixTests.everyPath —
pgm psd — image supported lossy colorProfile, dimensions A flat, single-layer PSD is written; partially transparent pixels are stored premultiplied. MatrixTests.everyPath —
pgm tga — image supported lossy dimensions Samples deeper than 8-bit are reduced to 8-bit. No embedded colour profile: pixels are converted to sRGB. MatrixTests.everyPath —
pgm tiff — image supported lossless colorProfile, dimensions — MatrixTests.everyPath —
pgm webp — image supported lossy colorProfile, dimensions Lossy by default (a lossless mode is available); the PSNR of the result is measured. Samples deeper than 8-bit are reduced to 8-bit. Largest side: 16383 pixels. MatrixTests.everyPath —
png apng — image supported lossless colorProfile, dimensions, transparency EXIF, IPTC and XMP metadata are not written. A still source gives a single-frame PNG. MatrixTests.everyPath —
png avif — image supported lossy colorProfile, dimensions, metadata, transparency Lossy compression; the PSNR of the result is measured. At most 10 bits per component. MatrixTests.everyPath —
png bmp — image supported lossy dimensions, transparency Samples deeper than 8-bit are reduced to 8-bit. No embedded colour profile: pixels are converted to sRGB. EXIF, IPTC and XMP metadata are not written. Transparency is stored in a 32-bit BMP; some programs ignore BMP alpha. MatrixTests.everyPath —
png exr — image supported lossy dimensions, transparency Pixels are stored as linear values with sRGB / Rec. 709 primaries. EXIF, IPTC and XMP metadata are not written. 8-bit sources are stored as 16-bit half floats, deeper sources as 32-bit floats. MatrixTests.everyPath —
png gif — image supported lossy dimensions, transparency GIF stores at most 256 colours per frame and on/off transparency. Samples deeper than 8-bit are reduced to 8-bit. No embedded colour profile: pixels are converted to sRGB. EXIF, IPTC and XMP metadata are not written. MatrixTests.everyPath —
png hdr — image supported lossy dimensions HDR has no transparency: transparent areas are filled with the background colour. Pixels are stored as linear values with sRGB / Rec. 709 primaries. EXIF, IPTC and XMP metadata are not written. RGBE keeps an 8-bit mantissa per channel. MatrixTests.everyPath —
png heic — image supported lossy colorProfile, dimensions, metadata, transparency Lossy compression; the PSNR of the result is measured. At most 10 bits per component. MatrixTests.everyPath 23.38 s (wall time; PNG → HEIC (q 80))
png heif — image supported lossy colorProfile, dimensions, metadata, transparency Lossy compression; the PSNR of the result is measured. At most 10 bits per component. Written as an HEVC-coded HEIF file (the same encoding as HEIC). MatrixTests.everyPath —
png icns — image supported lossy transparency Samples deeper than 8-bit are reduced to 8-bit. No embedded colour profile: pixels are converted to sRGB. EXIF, IPTC and XMP metadata are not written. The image is resampled to the standard icon sizes, fitted in a transparent square. MatrixTests.everyPath —
png ico — image supported lossy transparency Samples deeper than 8-bit are reduced to 8-bit. No embedded colour profile: pixels are converted to sRGB. EXIF, IPTC and XMP metadata are not written. The image is resampled to the standard icon sizes, fitted in a transparent square. MatrixTests.everyPath —
png jp2 — image supported lossy dimensions, transparency Lossy by default (a lossless mode is available); the PSNR of the result is measured. Samples deeper than 8-bit are reduced to 8-bit. No embedded colour profile: pixels are converted to sRGB. EXIF, IPTC and XMP metadata are not written. MatrixTests.everyPath —
png jpeg — image supported lossy colorProfile, dimensions, metadata Lossy compression; the PSNR of the result is measured. JPEG has no transparency: transparent areas are filled with the background colour. Samples deeper than 8-bit are reduced to 8-bit. Largest side: 65535 pixels. MatrixTests.everyPath 9.76 s (wall time; PNG → JPEG (q 85))
png pbm — image supported lossy dimensions PBM has no transparency: transparent areas are filled with the background colour. Samples deeper than 8-bit are reduced to 8-bit. No embedded colour profile: pixels are converted to sRGB. EXIF, IPTC and XMP metadata are not written. Every pixel becomes black or white (50 % threshold). MatrixTests.everyPath —
png pdf — pdf supported lossless dimensions, transparency One image per page; EXIF orientation is applied. Other metadata (EXIF, GPS, captions) is not copied into the PDF. With “Recompress as JPEG”, opaque images become lossy JPEG; images with transparency always stay lossless. Animated or multi-image files contribute their first frame only. ImagesIntoPDF/everySourceFormat (7 cases), jpegIsEmbeddedWithoutRecompression, losslessByDefaultAndJPEGOnRequest, exifOrientationIsApplied (4 cases), pageSizesMarginsOrderAndPasswords, optionalTextLayer, rejectsWhatIsNotAnImage —
png pdf image.vectorize vector supported lossy dimensions, transparency The drawing is an approximation made of flat colour shapes: gradients, textures and photographs become bands of colour. Partial transparency is not kept: every shape is opaque. Only the first frame of an animated or multi-page image is traced. Text becomes outlines; it is not recognised as text. Metadata and colour profiles are not carried over (colours are converted to sRGB). tracesEverySource("png") —
png pgm — image supported lossy dimensions PGM has no transparency: transparent areas are filled with the background colour. No embedded colour profile: pixels are converted to sRGB. EXIF, IPTC and XMP metadata are not written. Colour is reduced to gray (luma). MatrixTests.everyPath —
png png — image supported lossless colorProfile, dimensions, metadata, transparency — MatrixTests.everyPath —
png png image.optimize image supported lossless colorProfile, dimensions, metadata, transparency Pixels are unchanged; only the compression is redone. MatrixTests.operations —
png ppm — image supported lossy dimensions PPM has no transparency: transparent areas are filled with the background colour. No embedded colour profile: pixels are converted to sRGB. EXIF, IPTC and XMP metadata are not written. MatrixTests.everyPath —
png psd — image supported lossy colorProfile, dimensions, transparency EXIF, IPTC and XMP metadata are not written. A flat, single-layer PSD is written; partially transparent pixels are stored premultiplied. MatrixTests.everyPath —
png svg image.vectorize vector supported lossy dimensions, transparency The drawing is an approximation made of flat colour shapes: gradients, textures and photographs become bands of colour. Partial transparency is not kept: every shape is opaque. Only the first frame of an animated or multi-page image is traced. Text becomes outlines; it is not recognised as text. Metadata and colour profiles are not carried over (colours are converted to sRGB). tracesEverySource("png") —
png tga — image supported lossy dimensions, transparency Samples deeper than 8-bit are reduced to 8-bit. No embedded colour profile: pixels are converted to sRGB. EXIF, IPTC and XMP metadata are not written. MatrixTests.everyPath —
png tiff — image supported lossless colorProfile, dimensions, metadata, transparency — MatrixTests.everyPath —
png webp — image supported lossy colorProfile, dimensions, metadata, transparency Lossy by default (a lossless mode is available); the PSNR of the result is measured. Samples deeper than 8-bit are reduced to 8-bit. Largest side: 16383 pixels. MatrixTests.everyPath 24.39 s (wall time; PNG → WebP (q 80, method 4))
ppm apng — image supported lossless colorProfile, dimensions A still source gives a single-frame PNG. MatrixTests.everyPath —
ppm avif — image supported lossy colorProfile, dimensions Lossy compression; the PSNR of the result is measured. At most 10 bits per component. MatrixTests.everyPath —
ppm bmp — image supported lossy dimensions Samples deeper than 8-bit are reduced to 8-bit. No embedded colour profile: pixels are converted to sRGB. MatrixTests.everyPath —
ppm exr — image supported lossy dimensions Pixels are stored as linear values with sRGB / Rec. 709 primaries. 8-bit sources are stored as 16-bit half floats, deeper sources as 32-bit floats. MatrixTests.everyPath —
ppm gif — image supported lossy dimensions GIF stores at most 256 colours per frame and on/off transparency. Samples deeper than 8-bit are reduced to 8-bit. No embedded colour profile: pixels are converted to sRGB. MatrixTests.everyPath —
ppm hdr — image supported lossy dimensions Pixels are stored as linear values with sRGB / Rec. 709 primaries. RGBE keeps an 8-bit mantissa per channel. MatrixTests.everyPath —
ppm heic — image supported lossy colorProfile, dimensions Lossy compression; the PSNR of the result is measured. At most 10 bits per component. MatrixTests.everyPath —
ppm heif — image supported lossy colorProfile, dimensions Lossy compression; the PSNR of the result is measured. At most 10 bits per component. Written as an HEVC-coded HEIF file (the same encoding as HEIC). MatrixTests.everyPath —
ppm icns — image supported lossy — Samples deeper than 8-bit are reduced to 8-bit. No embedded colour profile: pixels are converted to sRGB. The image is resampled to the standard icon sizes, fitted in a transparent square. MatrixTests.everyPath —
ppm ico — image supported lossy — Samples deeper than 8-bit are reduced to 8-bit. No embedded colour profile: pixels are converted to sRGB. The image is resampled to the standard icon sizes, fitted in a transparent square. MatrixTests.everyPath —
ppm jp2 — image supported lossy dimensions Lossy by default (a lossless mode is available); the PSNR of the result is measured. Samples deeper than 8-bit are reduced to 8-bit. No embedded colour profile: pixels are converted to sRGB. MatrixTests.everyPath —
ppm jpeg — image supported lossy colorProfile, dimensions Lossy compression; the PSNR of the result is measured. Samples deeper than 8-bit are reduced to 8-bit. Largest side: 65535 pixels. MatrixTests.everyPath —
ppm pbm — image supported lossy dimensions Samples deeper than 8-bit are reduced to 8-bit. No embedded colour profile: pixels are converted to sRGB. Every pixel becomes black or white (50 % threshold). MatrixTests.everyPath —
ppm pgm — image supported lossy dimensions No embedded colour profile: pixels are converted to sRGB. Colour is reduced to gray (luma). MatrixTests.everyPath —
ppm png — image supported lossless colorProfile, dimensions — MatrixTests.everyPath —
ppm ppm — image supported lossless dimensions No embedded colour profile: pixels are converted to sRGB. MatrixTests.everyPath —
ppm psd — image supported lossy colorProfile, dimensions A flat, single-layer PSD is written; partially transparent pixels are stored premultiplied. MatrixTests.everyPath —
ppm tga — image supported lossy dimensions Samples deeper than 8-bit are reduced to 8-bit. No embedded colour profile: pixels are converted to sRGB. MatrixTests.everyPath —
ppm tiff — image supported lossless colorProfile, dimensions — MatrixTests.everyPath —
ppm webp — image supported lossy colorProfile, dimensions Lossy by default (a lossless mode is available); the PSNR of the result is measured. Samples deeper than 8-bit are reduced to 8-bit. Largest side: 16383 pixels. MatrixTests.everyPath —
psd apng — image supported lossless colorProfile, dimensions, transparency Layers are not read: ImageIO returns the flattened composite stored in the file. A still source gives a single-frame PNG. MatrixTests.everyPath —
psd avif — image supported lossy colorProfile, dimensions, transparency Layers are not read: ImageIO returns the flattened composite stored in the file. Lossy compression; the PSNR of the result is measured. At most 10 bits per component. MatrixTests.everyPath —
psd bmp — image supported lossy dimensions, transparency Layers are not read: ImageIO returns the flattened composite stored in the file. Samples deeper than 8-bit are reduced to 8-bit. No embedded colour profile: pixels are converted to sRGB. Transparency is stored in a 32-bit BMP; some programs ignore BMP alpha. MatrixTests.everyPath —
psd exr — image supported lossy dimensions, transparency Layers are not read: ImageIO returns the flattened composite stored in the file. Pixels are stored as linear values with sRGB / Rec. 709 primaries. 8-bit sources are stored as 16-bit half floats, deeper sources as 32-bit floats. MatrixTests.everyPath —
psd gif — image supported lossy dimensions, transparency Layers are not read: ImageIO returns the flattened composite stored in the file. GIF stores at most 256 colours per frame and on/off transparency. Samples deeper than 8-bit are reduced to 8-bit. No embedded colour profile: pixels are converted to sRGB. MatrixTests.everyPath —
psd hdr — image supported lossy dimensions Layers are not read: ImageIO returns the flattened composite stored in the file. HDR has no transparency: transparent areas are filled with the background colour. Pixels are stored as linear values with sRGB / Rec. 709 primaries. RGBE keeps an 8-bit mantissa per channel. MatrixTests.everyPath —
psd heic — image supported lossy colorProfile, dimensions, transparency Layers are not read: ImageIO returns the flattened composite stored in the file. Lossy compression; the PSNR of the result is measured. At most 10 bits per component. MatrixTests.everyPath —
psd heif — image supported lossy colorProfile, dimensions, transparency Layers are not read: ImageIO returns the flattened composite stored in the file. Lossy compression; the PSNR of the result is measured. At most 10 bits per component. Written as an HEVC-coded HEIF file (the same encoding as HEIC). MatrixTests.everyPath —
psd icns — image supported lossy transparency Layers are not read: ImageIO returns the flattened composite stored in the file. Samples deeper than 8-bit are reduced to 8-bit. No embedded colour profile: pixels are converted to sRGB. The image is resampled to the standard icon sizes, fitted in a transparent square. MatrixTests.everyPath —
psd ico — image supported lossy transparency Layers are not read: ImageIO returns the flattened composite stored in the file. Samples deeper than 8-bit are reduced to 8-bit. No embedded colour profile: pixels are converted to sRGB. The image is resampled to the standard icon sizes, fitted in a transparent square. MatrixTests.everyPath —
psd jp2 — image supported lossy dimensions, transparency Layers are not read: ImageIO returns the flattened composite stored in the file. Lossy by default (a lossless mode is available); the PSNR of the result is measured. Samples deeper than 8-bit are reduced to 8-bit. No embedded colour profile: pixels are converted to sRGB. MatrixTests.everyPath —
psd jpeg — image supported lossy colorProfile, dimensions Layers are not read: ImageIO returns the flattened composite stored in the file. Lossy compression; the PSNR of the result is measured. JPEG has no transparency: transparent areas are filled with the background colour. Samples deeper than 8-bit are reduced to 8-bit. Largest side: 65535 pixels. MatrixTests.everyPath —
psd pbm — image supported lossy dimensions Layers are not read: ImageIO returns the flattened composite stored in the file. PBM has no transparency: transparent areas are filled with the background colour. Samples deeper than 8-bit are reduced to 8-bit. No embedded colour profile: pixels are converted to sRGB. Every pixel becomes black or white (50 % threshold). MatrixTests.everyPath —
psd pdf image.vectorize vector supported lossy dimensions, transparency The drawing is an approximation made of flat colour shapes: gradients, textures and photographs become bands of colour. Partial transparency is not kept: every shape is opaque. Only the first frame of an animated or multi-page image is traced. Text becomes outlines; it is not recognised as text. Metadata and colour profiles are not carried over (colours are converted to sRGB). tracesEverySource("psd") —
psd pgm — image supported lossy dimensions Layers are not read: ImageIO returns the flattened composite stored in the file. PGM has no transparency: transparent areas are filled with the background colour. No embedded colour profile: pixels are converted to sRGB. Colour is reduced to gray (luma). MatrixTests.everyPath —
psd png — image supported lossless colorProfile, dimensions, transparency Layers are not read: ImageIO returns the flattened composite stored in the file. MatrixTests.everyPath —
psd ppm — image supported lossy dimensions Layers are not read: ImageIO returns the flattened composite stored in the file. PPM has no transparency: transparent areas are filled with the background colour. No embedded colour profile: pixels are converted to sRGB. MatrixTests.everyPath —
psd psd — image supported lossy colorProfile, dimensions, transparency Layers are not read: ImageIO returns the flattened composite stored in the file. A flat, single-layer PSD is written; partially transparent pixels are stored premultiplied. MatrixTests.everyPath —
psd svg image.vectorize vector supported lossy dimensions, transparency The drawing is an approximation made of flat colour shapes: gradients, textures and photographs become bands of colour. Partial transparency is not kept: every shape is opaque. Only the first frame of an animated or multi-page image is traced. Text becomes outlines; it is not recognised as text. Metadata and colour profiles are not carried over (colours are converted to sRGB). tracesEverySource("psd") —
psd tga — image supported lossy dimensions, transparency Layers are not read: ImageIO returns the flattened composite stored in the file. Samples deeper than 8-bit are reduced to 8-bit. No embedded colour profile: pixels are converted to sRGB. MatrixTests.everyPath —
psd tiff — image supported lossless colorProfile, dimensions, transparency Layers are not read: ImageIO returns the flattened composite stored in the file. MatrixTests.everyPath —
psd webp — image supported lossy colorProfile, dimensions, transparency Layers are not read: ImageIO returns the flattened composite stored in the file. Lossy by default (a lossless mode is available); the PSNR of the result is measured. Samples deeper than 8-bit are reduced to 8-bit. Largest side: 16383 pixels. MatrixTests.everyPath —
raw apng — image experimental lossy colorProfile, dimensions RAW development is Apple's default rendering (ImageIO): no exposure, white-balance or noise controls. No manufacturer RAW file was available to test on the build machine; decoding relies on the camera support of macOS. EXIF, IPTC and XMP metadata are not written. A still source gives a single-frame PNG. MatrixTests.everyPath —
raw avif — image experimental lossy colorProfile, dimensions, metadata RAW development is Apple's default rendering (ImageIO): no exposure, white-balance or noise controls. No manufacturer RAW file was available to test on the build machine; decoding relies on the camera support of macOS. Lossy compression; the PSNR of the result is measured. At most 10 bits per component. MatrixTests.everyPath —
raw bmp — image experimental lossy dimensions RAW development is Apple's default rendering (ImageIO): no exposure, white-balance or noise controls. No manufacturer RAW file was available to test on the build machine; decoding relies on the camera support of macOS. Samples deeper than 8-bit are reduced to 8-bit. No embedded colour profile: pixels are converted to sRGB. EXIF, IPTC and XMP metadata are not written. MatrixTests.everyPath —
raw exr — image experimental lossy dimensions RAW development is Apple's default rendering (ImageIO): no exposure, white-balance or noise controls. No manufacturer RAW file was available to test on the build machine; decoding relies on the camera support of macOS. Pixels are stored as linear values with sRGB / Rec. 709 primaries. EXIF, IPTC and XMP metadata are not written. 8-bit sources are stored as 16-bit half floats, deeper sources as 32-bit floats. MatrixTests.everyPath —
raw gif — image experimental lossy dimensions RAW development is Apple's default rendering (ImageIO): no exposure, white-balance or noise controls. No manufacturer RAW file was available to test on the build machine; decoding relies on the camera support of macOS. GIF stores at most 256 colours per frame and on/off transparency. Samples deeper than 8-bit are reduced to 8-bit. No embedded colour profile: pixels are converted to sRGB. EXIF, IPTC and XMP metadata are not written. MatrixTests.everyPath —
raw hdr — image experimental lossy dimensions RAW development is Apple's default rendering (ImageIO): no exposure, white-balance or noise controls. No manufacturer RAW file was available to test on the build machine; decoding relies on the camera support of macOS. Pixels are stored as linear values with sRGB / Rec. 709 primaries. EXIF, IPTC and XMP metadata are not written. RGBE keeps an 8-bit mantissa per channel. MatrixTests.everyPath —
raw heic — image experimental lossy colorProfile, dimensions, metadata RAW development is Apple's default rendering (ImageIO): no exposure, white-balance or noise controls. No manufacturer RAW file was available to test on the build machine; decoding relies on the camera support of macOS. Lossy compression; the PSNR of the result is measured. At most 10 bits per component. MatrixTests.everyPath —
raw heif — image experimental lossy colorProfile, dimensions, metadata RAW development is Apple's default rendering (ImageIO): no exposure, white-balance or noise controls. No manufacturer RAW file was available to test on the build machine; decoding relies on the camera support of macOS. Lossy compression; the PSNR of the result is measured. At most 10 bits per component. Written as an HEVC-coded HEIF file (the same encoding as HEIC). MatrixTests.everyPath —
raw icns — image experimental lossy — RAW development is Apple's default rendering (ImageIO): no exposure, white-balance or noise controls. No manufacturer RAW file was available to test on the build machine; decoding relies on the camera support of macOS. Samples deeper than 8-bit are reduced to 8-bit. No embedded colour profile: pixels are converted to sRGB. EXIF, IPTC and XMP metadata are not written. The image is resampled to the standard icon sizes, fitted in a transparent square. MatrixTests.everyPath —
raw ico — image experimental lossy — RAW development is Apple's default rendering (ImageIO): no exposure, white-balance or noise controls. No manufacturer RAW file was available to test on the build machine; decoding relies on the camera support of macOS. Samples deeper than 8-bit are reduced to 8-bit. No embedded colour profile: pixels are converted to sRGB. EXIF, IPTC and XMP metadata are not written. The image is resampled to the standard icon sizes, fitted in a transparent square. MatrixTests.everyPath —
raw jp2 — image experimental lossy dimensions RAW development is Apple's default rendering (ImageIO): no exposure, white-balance or noise controls. No manufacturer RAW file was available to test on the build machine; decoding relies on the camera support of macOS. Lossy by default (a lossless mode is available); the PSNR of the result is measured. Samples deeper than 8-bit are reduced to 8-bit. No embedded colour profile: pixels are converted to sRGB. EXIF, IPTC and XMP metadata are not written. MatrixTests.everyPath —
raw jpeg — image experimental lossy colorProfile, dimensions, metadata RAW development is Apple's default rendering (ImageIO): no exposure, white-balance or noise controls. No manufacturer RAW file was available to test on the build machine; decoding relies on the camera support of macOS. Lossy compression; the PSNR of the result is measured. Samples deeper than 8-bit are reduced to 8-bit. Largest side: 65535 pixels. MatrixTests.everyPath —
raw pbm — image experimental lossy dimensions RAW development is Apple's default rendering (ImageIO): no exposure, white-balance or noise controls. No manufacturer RAW file was available to test on the build machine; decoding relies on the camera support of macOS. Samples deeper than 8-bit are reduced to 8-bit. No embedded colour profile: pixels are converted to sRGB. EXIF, IPTC and XMP metadata are not written. Every pixel becomes black or white (50 % threshold). MatrixTests.everyPath —
raw pgm — image experimental lossy dimensions RAW development is Apple's default rendering (ImageIO): no exposure, white-balance or noise controls. No manufacturer RAW file was available to test on the build machine; decoding relies on the camera support of macOS. No embedded colour profile: pixels are converted to sRGB. EXIF, IPTC and XMP metadata are not written. Colour is reduced to gray (luma). MatrixTests.everyPath —
raw png — image experimental lossy colorProfile, dimensions, metadata RAW development is Apple's default rendering (ImageIO): no exposure, white-balance or noise controls. No manufacturer RAW file was available to test on the build machine; decoding relies on the camera support of macOS. MatrixTests.everyPath —
raw ppm — image experimental lossy dimensions RAW development is Apple's default rendering (ImageIO): no exposure, white-balance or noise controls. No manufacturer RAW file was available to test on the build machine; decoding relies on the camera support of macOS. No embedded colour profile: pixels are converted to sRGB. EXIF, IPTC and XMP metadata are not written. MatrixTests.everyPath —
raw psd — image experimental lossy colorProfile, dimensions RAW development is Apple's default rendering (ImageIO): no exposure, white-balance or noise controls. No manufacturer RAW file was available to test on the build machine; decoding relies on the camera support of macOS. EXIF, IPTC and XMP metadata are not written. A flat, single-layer PSD is written; partially transparent pixels are stored premultiplied. MatrixTests.everyPath —
raw tga — image experimental lossy dimensions RAW development is Apple's default rendering (ImageIO): no exposure, white-balance or noise controls. No manufacturer RAW file was available to test on the build machine; decoding relies on the camera support of macOS. Samples deeper than 8-bit are reduced to 8-bit. No embedded colour profile: pixels are converted to sRGB. EXIF, IPTC and XMP metadata are not written. MatrixTests.everyPath —
raw tiff — image experimental lossy colorProfile, dimensions, metadata RAW development is Apple's default rendering (ImageIO): no exposure, white-balance or noise controls. No manufacturer RAW file was available to test on the build machine; decoding relies on the camera support of macOS. MatrixTests.everyPath —
raw webp — image experimental lossy colorProfile, dimensions, metadata RAW development is Apple's default rendering (ImageIO): no exposure, white-balance or noise controls. No manufacturer RAW file was available to test on the build machine; decoding relies on the camera support of macOS. Lossy by default (a lossless mode is available); the PSNR of the result is measured. Samples deeper than 8-bit are reduced to 8-bit. Largest side: 16383 pixels. MatrixTests.everyPath —
tga apng — image supported lossless colorProfile, dimensions, transparency A still source gives a single-frame PNG. MatrixTests.everyPath —
tga avif — image supported lossy colorProfile, dimensions, transparency Lossy compression; the PSNR of the result is measured. At most 10 bits per component. MatrixTests.everyPath —
tga bmp — image supported lossless dimensions, transparency No embedded colour profile: pixels are converted to sRGB. Transparency is stored in a 32-bit BMP; some programs ignore BMP alpha. MatrixTests.everyPath —
tga exr — image supported lossy dimensions, transparency Pixels are stored as linear values with sRGB / Rec. 709 primaries. 8-bit sources are stored as 16-bit half floats, deeper sources as 32-bit floats. MatrixTests.everyPath —
tga gif — image supported lossy dimensions, transparency GIF stores at most 256 colours per frame and on/off transparency. No embedded colour profile: pixels are converted to sRGB. MatrixTests.everyPath —
tga hdr — image supported lossy dimensions HDR has no transparency: transparent areas are filled with the background colour. Pixels are stored as linear values with sRGB / Rec. 709 primaries. RGBE keeps an 8-bit mantissa per channel. MatrixTests.everyPath —
tga heic — image supported lossy colorProfile, dimensions, transparency Lossy compression; the PSNR of the result is measured. At most 10 bits per component. MatrixTests.everyPath —
tga heif — image supported lossy colorProfile, dimensions, transparency Lossy compression; the PSNR of the result is measured. At most 10 bits per component. Written as an HEVC-coded HEIF file (the same encoding as HEIC). MatrixTests.everyPath —
tga icns — image supported lossy transparency No embedded colour profile: pixels are converted to sRGB. The image is resampled to the standard icon sizes, fitted in a transparent square. MatrixTests.everyPath —
tga ico — image supported lossy transparency No embedded colour profile: pixels are converted to sRGB. The image is resampled to the standard icon sizes, fitted in a transparent square. MatrixTests.everyPath —
tga jp2 — image supported lossy dimensions, transparency Lossy by default (a lossless mode is available); the PSNR of the result is measured. No embedded colour profile: pixels are converted to sRGB. MatrixTests.everyPath —
tga jpeg — image supported lossy colorProfile, dimensions Lossy compression; the PSNR of the result is measured. JPEG has no transparency: transparent areas are filled with the background colour. Largest side: 65535 pixels. MatrixTests.everyPath —
tga pbm — image supported lossy dimensions PBM has no transparency: transparent areas are filled with the background colour. No embedded colour profile: pixels are converted to sRGB. Every pixel becomes black or white (50 % threshold). MatrixTests.everyPath —
tga pdf image.vectorize vector supported lossy dimensions, transparency The drawing is an approximation made of flat colour shapes: gradients, textures and photographs become bands of colour. Partial transparency is not kept: every shape is opaque. Only the first frame of an animated or multi-page image is traced. Text becomes outlines; it is not recognised as text. Metadata and colour profiles are not carried over (colours are converted to sRGB). tracesEverySource("tga") —
tga pgm — image supported lossy dimensions PGM has no transparency: transparent areas are filled with the background colour. No embedded colour profile: pixels are converted to sRGB. Colour is reduced to gray (luma). MatrixTests.everyPath —
tga png — image supported lossless colorProfile, dimensions, transparency — MatrixTests.everyPath —
tga ppm — image supported lossy dimensions PPM has no transparency: transparent areas are filled with the background colour. No embedded colour profile: pixels are converted to sRGB. MatrixTests.everyPath —
tga psd — image supported lossy colorProfile, dimensions, transparency A flat, single-layer PSD is written; partially transparent pixels are stored premultiplied. MatrixTests.everyPath —
tga svg image.vectorize vector supported lossy dimensions, transparency The drawing is an approximation made of flat colour shapes: gradients, textures and photographs become bands of colour. Partial transparency is not kept: every shape is opaque. Only the first frame of an animated or multi-page image is traced. Text becomes outlines; it is not recognised as text. Metadata and colour profiles are not carried over (colours are converted to sRGB). tracesEverySource("tga") —
tga tga — image supported lossless dimensions, transparency No embedded colour profile: pixels are converted to sRGB. MatrixTests.everyPath —
tga tiff — image supported lossless colorProfile, dimensions, transparency — MatrixTests.everyPath —
tga webp — image supported lossy colorProfile, dimensions, transparency Lossy by default (a lossless mode is available); the PSNR of the result is measured. Largest side: 16383 pixels. MatrixTests.everyPath —
tiff apng — image supported lossless colorProfile, dimensions, transparency Multi-page TIFF: a single page is converted (the first unless another is chosen). Floating-point (HDR) sources are tone-mapped into the SDR range. Samples deeper than 16-bit are reduced to 16-bit. EXIF, IPTC and XMP metadata are not written. A still source gives a single-frame PNG. MatrixTests.everyPath —
tiff avif — image supported lossy colorProfile, dimensions, metadata, transparency Multi-page TIFF: a single page is converted (the first unless another is chosen). Lossy compression; the PSNR of the result is measured. Floating-point (HDR) sources are tone-mapped into the SDR range. Samples deeper than 16-bit are reduced to 16-bit. At most 10 bits per component. MatrixTests.everyPath —
tiff bmp — image supported lossy dimensions, transparency Multi-page TIFF: a single page is converted (the first unless another is chosen). Floating-point (HDR) sources are tone-mapped into the SDR range. Samples deeper than 8-bit are reduced to 8-bit. No embedded colour profile: pixels are converted to sRGB. EXIF, IPTC and XMP metadata are not written. Transparency is stored in a 32-bit BMP; some programs ignore BMP alpha. MatrixTests.everyPath —
tiff exr — image supported lossy dimensions, hdr, transparency Multi-page TIFF: a single page is converted (the first unless another is chosen). Pixels are stored as linear values with sRGB / Rec. 709 primaries. EXIF, IPTC and XMP metadata are not written. 8-bit sources are stored as 16-bit half floats, deeper sources as 32-bit floats. MatrixTests.everyPath —
tiff gif — image supported lossy dimensions, transparency Multi-page TIFF: a single page is converted (the first unless another is chosen). GIF stores at most 256 colours per frame and on/off transparency. Floating-point (HDR) sources are tone-mapped into the SDR range. Samples deeper than 8-bit are reduced to 8-bit. No embedded colour profile: pixels are converted to sRGB. EXIF, IPTC and XMP metadata are not written. MatrixTests.everyPath —
tiff hdr — image supported lossy dimensions, hdr Multi-page TIFF: a single page is converted (the first unless another is chosen). HDR has no transparency: transparent areas are filled with the background colour. Pixels are stored as linear values with sRGB / Rec. 709 primaries. EXIF, IPTC and XMP metadata are not written. RGBE keeps an 8-bit mantissa per channel. MatrixTests.everyPath —
tiff heic — image supported lossy colorProfile, dimensions, metadata, transparency Multi-page TIFF: a single page is converted (the first unless another is chosen). Lossy compression; the PSNR of the result is measured. Floating-point (HDR) sources are tone-mapped into the SDR range. Samples deeper than 16-bit are reduced to 16-bit. At most 10 bits per component. MatrixTests.everyPath —
tiff heif — image supported lossy colorProfile, dimensions, metadata, transparency Multi-page TIFF: a single page is converted (the first unless another is chosen). Lossy compression; the PSNR of the result is measured. Floating-point (HDR) sources are tone-mapped into the SDR range. Samples deeper than 16-bit are reduced to 16-bit. At most 10 bits per component. Written as an HEVC-coded HEIF file (the same encoding as HEIC). MatrixTests.everyPath —
tiff icns — image supported lossy transparency Multi-page TIFF: a single page is converted (the first unless another is chosen). Floating-point (HDR) sources are tone-mapped into the SDR range. Samples deeper than 8-bit are reduced to 8-bit. No embedded colour profile: pixels are converted to sRGB. EXIF, IPTC and XMP metadata are not written. The image is resampled to the standard icon sizes, fitted in a transparent square. MatrixTests.everyPath —
tiff ico — image supported lossy transparency Multi-page TIFF: a single page is converted (the first unless another is chosen). Floating-point (HDR) sources are tone-mapped into the SDR range. Samples deeper than 8-bit are reduced to 8-bit. No embedded colour profile: pixels are converted to sRGB. EXIF, IPTC and XMP metadata are not written. The image is resampled to the standard icon sizes, fitted in a transparent square. MatrixTests.everyPath —
tiff jp2 — image supported lossy dimensions, transparency Multi-page TIFF: a single page is converted (the first unless another is chosen). Lossy by default (a lossless mode is available); the PSNR of the result is measured. Floating-point (HDR) sources are tone-mapped into the SDR range. Samples deeper than 8-bit are reduced to 8-bit. No embedded colour profile: pixels are converted to sRGB. EXIF, IPTC and XMP metadata are not written. MatrixTests.everyPath —
tiff jpeg — image supported lossy colorProfile, dimensions, metadata Multi-page TIFF: a single page is converted (the first unless another is chosen). Lossy compression; the PSNR of the result is measured. JPEG has no transparency: transparent areas are filled with the background colour. Floating-point (HDR) sources are tone-mapped into the SDR range. Samples deeper than 8-bit are reduced to 8-bit. Largest side: 65535 pixels. MatrixTests.everyPath —
tiff pbm — image supported lossy dimensions Multi-page TIFF: a single page is converted (the first unless another is chosen). PBM has no transparency: transparent areas are filled with the background colour. Floating-point (HDR) sources are tone-mapped into the SDR range. Samples deeper than 8-bit are reduced to 8-bit. No embedded colour profile: pixels are converted to sRGB. EXIF, IPTC and XMP metadata are not written. Every pixel becomes black or white (50 % threshold). MatrixTests.everyPath —
tiff pdf — pdf supported lossless dimensions, transparency One image per page; EXIF orientation is applied. Other metadata (EXIF, GPS, captions) is not copied into the PDF. With “Recompress as JPEG”, opaque images become lossy JPEG; images with transparency always stay lossless. Every frame of a multi-page TIFF becomes a page. ImagesIntoPDF/everySourceFormat (7 cases), jpegIsEmbeddedWithoutRecompression, losslessByDefaultAndJPEGOnRequest, exifOrientationIsApplied (4 cases), pageSizesMarginsOrderAndPasswords, optionalTextLayer, rejectsWhatIsNotAnImage —
tiff pdf image.vectorize vector supported lossy dimensions, transparency The drawing is an approximation made of flat colour shapes: gradients, textures and photographs become bands of colour. Partial transparency is not kept: every shape is opaque. Only the first frame of an animated or multi-page image is traced. Text becomes outlines; it is not recognised as text. Metadata and colour profiles are not carried over (colours are converted to sRGB). tracesEverySource("tiff") —
tiff pgm — image supported lossy dimensions Multi-page TIFF: a single page is converted (the first unless another is chosen). PGM has no transparency: transparent areas are filled with the background colour. Floating-point (HDR) sources are tone-mapped into the SDR range. Samples deeper than 16-bit are reduced to 16-bit. No embedded colour profile: pixels are converted to sRGB. EXIF, IPTC and XMP metadata are not written. Colour is reduced to gray (luma). MatrixTests.everyPath —
tiff png — image supported lossless colorProfile, dimensions, metadata, transparency Multi-page TIFF: a single page is converted (the first unless another is chosen). Floating-point (HDR) sources are tone-mapped into the SDR range. Samples deeper than 16-bit are reduced to 16-bit. MatrixTests.everyPath —
tiff ppm — image supported lossy dimensions Multi-page TIFF: a single page is converted (the first unless another is chosen). PPM has no transparency: transparent areas are filled with the background colour. Floating-point (HDR) sources are tone-mapped into the SDR range. Samples deeper than 16-bit are reduced to 16-bit. No embedded colour profile: pixels are converted to sRGB. EXIF, IPTC and XMP metadata are not written. MatrixTests.everyPath —
tiff psd — image supported lossy colorProfile, dimensions, transparency Multi-page TIFF: a single page is converted (the first unless another is chosen). Floating-point (HDR) sources are tone-mapped into the SDR range. Samples deeper than 16-bit are reduced to 16-bit. EXIF, IPTC and XMP metadata are not written. A flat, single-layer PSD is written; partially transparent pixels are stored premultiplied. MatrixTests.everyPath —
tiff svg image.vectorize vector supported lossy dimensions, transparency The drawing is an approximation made of flat colour shapes: gradients, textures and photographs become bands of colour. Partial transparency is not kept: every shape is opaque. Only the first frame of an animated or multi-page image is traced. Text becomes outlines; it is not recognised as text. Metadata and colour profiles are not carried over (colours are converted to sRGB). tracesEverySource("tiff") —
tiff tga — image supported lossy dimensions, transparency Multi-page TIFF: a single page is converted (the first unless another is chosen). Floating-point (HDR) sources are tone-mapped into the SDR range. Samples deeper than 8-bit are reduced to 8-bit. No embedded colour profile: pixels are converted to sRGB. EXIF, IPTC and XMP metadata are not written. MatrixTests.everyPath —
tiff tiff — image supported lossless colorProfile, dimensions, hdr, metadata, transparency — MatrixTests.everyPath —
tiff webp — image supported lossy colorProfile, dimensions, metadata, transparency Multi-page TIFF: a single page is converted (the first unless another is chosen). Lossy by default (a lossless mode is available); the PSNR of the result is measured. Floating-point (HDR) sources are tone-mapped into the SDR range. Samples deeper than 8-bit are reduced to 8-bit. Largest side: 16383 pixels. MatrixTests.everyPath —
webp apng — image supported lossless animation, colorProfile, dimensions, transparency EXIF, IPTC and XMP metadata are not written. MatrixTests.everyPath —
webp avif — image supported lossy colorProfile, dimensions, metadata, transparency Animated sources: only the first frame is kept. Lossy compression; the PSNR of the result is measured. At most 10 bits per component. MatrixTests.everyPath —
webp bmp — image supported lossless dimensions, transparency Animated sources: only the first frame is kept. No embedded colour profile: pixels are converted to sRGB. EXIF, IPTC and XMP metadata are not written. Transparency is stored in a 32-bit BMP; some programs ignore BMP alpha. MatrixTests.everyPath —
webp exr — image supported lossy dimensions, transparency Animated sources: only the first frame is kept. Pixels are stored as linear values with sRGB / Rec. 709 primaries. EXIF, IPTC and XMP metadata are not written. 8-bit sources are stored as 16-bit half floats, deeper sources as 32-bit floats. MatrixTests.everyPath —
webp gif — image supported lossy animation, dimensions, transparency GIF stores at most 256 colours per frame and on/off transparency. No embedded colour profile: pixels are converted to sRGB. EXIF, IPTC and XMP metadata are not written. MatrixTests.everyPath —
webp hdr — image supported lossy dimensions Animated sources: only the first frame is kept. HDR has no transparency: transparent areas are filled with the background colour. Pixels are stored as linear values with sRGB / Rec. 709 primaries. EXIF, IPTC and XMP metadata are not written. RGBE keeps an 8-bit mantissa per channel. MatrixTests.everyPath —
webp heic — image supported lossy colorProfile, dimensions, metadata, transparency Animated sources: only the first frame is kept. Lossy compression; the PSNR of the result is measured. At most 10 bits per component. MatrixTests.everyPath —
webp heif — image supported lossy colorProfile, dimensions, metadata, transparency Animated sources: only the first frame is kept. Lossy compression; the PSNR of the result is measured. At most 10 bits per component. Written as an HEVC-coded HEIF file (the same encoding as HEIC). MatrixTests.everyPath —
webp icns — image supported lossy transparency Animated sources: only the first frame is kept. No embedded colour profile: pixels are converted to sRGB. EXIF, IPTC and XMP metadata are not written. The image is resampled to the standard icon sizes, fitted in a transparent square. MatrixTests.everyPath —
webp ico — image supported lossy transparency Animated sources: only the first frame is kept. No embedded colour profile: pixels are converted to sRGB. EXIF, IPTC and XMP metadata are not written. The image is resampled to the standard icon sizes, fitted in a transparent square. MatrixTests.everyPath —
webp jp2 — image supported lossy dimensions, transparency Animated sources: only the first frame is kept. Lossy by default (a lossless mode is available); the PSNR of the result is measured. No embedded colour profile: pixels are converted to sRGB. EXIF, IPTC and XMP metadata are not written. MatrixTests.everyPath —
webp jpeg — image supported lossy colorProfile, dimensions, metadata Animated sources: only the first frame is kept. Lossy compression; the PSNR of the result is measured. JPEG has no transparency: transparent areas are filled with the background colour. Largest side: 65535 pixels. MatrixTests.everyPath —
webp pbm — image supported lossy dimensions Animated sources: only the first frame is kept. PBM has no transparency: transparent areas are filled with the background colour. No embedded colour profile: pixels are converted to sRGB. EXIF, IPTC and XMP metadata are not written. Every pixel becomes black or white (50 % threshold). MatrixTests.everyPath —
webp pdf — pdf supported lossless dimensions One image per page; EXIF orientation is applied. Other metadata (EXIF, GPS, captions) is not copied into the PDF. With “Recompress as JPEG”, opaque images become lossy JPEG; images with transparency always stay lossless. Animated or multi-image files contribute their first frame only. ImagesIntoPDF/everySourceFormat (7 cases), jpegIsEmbeddedWithoutRecompression, losslessByDefaultAndJPEGOnRequest, exifOrientationIsApplied (4 cases), pageSizesMarginsOrderAndPasswords, optionalTextLayer, rejectsWhatIsNotAnImage —
webp pdf image.vectorize vector supported lossy dimensions, transparency The drawing is an approximation made of flat colour shapes: gradients, textures and photographs become bands of colour. Partial transparency is not kept: every shape is opaque. Only the first frame of an animated or multi-page image is traced. Text becomes outlines; it is not recognised as text. Metadata and colour profiles are not carried over (colours are converted to sRGB). tracesEverySource("webp") —
webp pgm — image supported lossy dimensions Animated sources: only the first frame is kept. PGM has no transparency: transparent areas are filled with the background colour. No embedded colour profile: pixels are converted to sRGB. EXIF, IPTC and XMP metadata are not written. Colour is reduced to gray (luma). MatrixTests.everyPath —
webp png — image supported lossless colorProfile, dimensions, metadata, transparency Animated sources: only the first frame is kept. MatrixTests.everyPath —
webp ppm — image supported lossy dimensions Animated sources: only the first frame is kept. PPM has no transparency: transparent areas are filled with the background colour. No embedded colour profile: pixels are converted to sRGB. EXIF, IPTC and XMP metadata are not written. MatrixTests.everyPath —
webp psd — image supported lossy colorProfile, dimensions, transparency Animated sources: only the first frame is kept. EXIF, IPTC and XMP metadata are not written. A flat, single-layer PSD is written; partially transparent pixels are stored premultiplied. MatrixTests.everyPath —
webp svg image.vectorize vector supported lossy dimensions, transparency The drawing is an approximation made of flat colour shapes: gradients, textures and photographs become bands of colour. Partial transparency is not kept: every shape is opaque. Only the first frame of an animated or multi-page image is traced. Text becomes outlines; it is not recognised as text. Metadata and colour profiles are not carried over (colours are converted to sRGB). tracesEverySource("webp") —
webp tga — image supported lossless dimensions, transparency Animated sources: only the first frame is kept. No embedded colour profile: pixels are converted to sRGB. EXIF, IPTC and XMP metadata are not written. MatrixTests.everyPath —
webp tiff — image supported lossless colorProfile, dimensions, metadata, transparency Animated sources: only the first frame is kept. MatrixTests.everyPath —
webp webp — image supported lossy animation, colorProfile, dimensions, metadata, transparency Lossy by default (a lossless mode is available); the PSNR of the result is measured. Largest side: 16383 pixels. MatrixTests.everyPath —
webp webp image.optimize image supported lossy animation, colorProfile, dimensions, metadata, transparency The image is decoded and encoded again; lossy unless “Lossless” is chosen. MatrixTests.operations —

# vector sources (4 paths)

Source Destination Operation Engine Status Fidelity Preserves Limitations Tests Performance
svg jpeg — vector supported lossy dimensions Scripts and event handlers in the SVG are not executed. Animations are not played: the drawing is rendered in its initial state. External references (linked images, fonts, style sheets, other files) are not loaded. Text uses the fonts installed on this Mac; a missing font is replaced. The drawing becomes pixels (at most 8192 per side). JPEG has no transparency: the drawing is placed on a background colour. svgToBitmapsWithScaleAndAlpha —
svg pdf — vector supported structural dimensions, textStyles, vectorGeometry Scripts and event handlers in the SVG are not executed. Animations are not played: the drawing is rendered in its initial state. External references (linked images, fonts, style sheets, other files) are not loaded. Text uses the fonts installed on this Mac; a missing font is replaced. Filter effects, masks and embedded bitmaps can be stored as images inside the PDF. The PDF page is opaque: the drawing is placed on white, or on the background colour you choose. svgToVectorPDF, scriptsAreNotExecutedAndRemoteReferencesAreNeverFetched —
svg png — vector supported lossy dimensions, transparency Scripts and event handlers in the SVG are not executed. Animations are not played: the drawing is rendered in its initial state. External references (linked images, fonts, style sheets, other files) are not loaded. Text uses the fonts installed on this Mac; a missing font is replaced. The drawing becomes pixels (at most 8192 per side). svgToBitmapsWithScaleAndAlpha, tracedSVGRendersTheSameInWebKit —
svg tiff — vector supported lossy dimensions, transparency Scripts and event handlers in the SVG are not executed. Animations are not played: the drawing is rendered in its initial state. External references (linked images, fonts, style sheets, other files) are not loaded. Text uses the fonts installed on this Mac; a missing font is replaced. The drawing becomes pixels (at most 8192 per side). svgToBitmapsWithScaleAndAlpha —

# video sources (302 paths)

Source Destination Operation Engine Status Fidelity Preserves Limitations Tests Performance
avi aac video.extract-audio video supported partial — Only the sound of one audio track is kept; the picture is not. The track is copied bit for bit when it already uses this codec; otherwise it is re-encoded. When re-encoded, the audio goes through a lossy codec again. VideoPaths.extractAudio —
avi aiff video.extract-audio video supported partial metadata Only the sound of one audio track is kept; the picture is not. VideoPaths.extractAudio —
avi apng — video supported lossy animation No sound. Full-colour frames without inter-frame video compression: files are large. The frame rate and the size are reduced (12 fps and 480 px wide unless changed). HDR video is tone-mapped to SDR. VideoPaths.animatedImage —
avi avi — video supported lossy audioTracks, dimensions, metadata, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). Chapters are dropped. Subtitle tracks are dropped: AVI cannot carry them here. HDR video is tone-mapped to standard dynamic range; 10-bit pictures become 8-bit. Only the first video track is kept; data and attachment streams are dropped. MPEG-4 Part 2 video; AVI cannot hold HEVC or AV1 here. VideoPaths.conversion —
avi flac video.extract-audio video supported partial metadata Only the sound of one audio track is kept; the picture is not. The track is copied bit for bit when it already uses this codec; otherwise it is re-encoded. VideoPaths.extractAudio —
avi flv — video supported lossy audioTracks, dimensions, metadata, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). Chapters are dropped. Subtitle tracks are dropped: Flash Video cannot carry them here. HDR video is tone-mapped to standard dynamic range; 10-bit pictures become 8-bit. Only the first video track is kept; data and attachment streams are dropped. One audio track only. VideoPaths.conversion —
avi gif — video supported lossy animation No sound. 256 colours per frame, from a palette computed for this video, with dithering. The frame rate and the size are reduced (12 fps and 480 px wide unless changed). HDR video is tone-mapped to SDR. VideoPaths.animatedImage —
avi m4a video.extract-audio video supported partial metadata Only the sound of one audio track is kept; the picture is not. The track is copied bit for bit when it already uses this codec; otherwise it is re-encoded. When re-encoded, the audio goes through a lossy codec again. VideoPaths.extractAudio —
avi m4v — video supported lossy audioTracks, chapters, dimensions, hdr, metadata, subtitleTracks, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). Text subtitles are kept as plain timed text (styling is lost); image-based subtitles are dropped. HDR10 and HLG are kept with HEVC, AV1, VP9 or ProRes; other codecs get a tone-mapped SDR picture. Dolby Vision metadata is not carried. Only the first video track is kept; data and attachment streams are dropped. VideoPaths.conversion —
avi m4v video.remux video supported lossless audioTracks, chapters, colorProfile, dimensions, hdr, metadata, subtitleTracks, timing Works only when the codecs inside the file are allowed in M4V; otherwise it stops and says which stream is in the way. With trimming, the result starts on the keyframe at or before the requested start (not frame-accurate). Text subtitles are rewritten in the container's own text format; image-based subtitles are dropped. VideoPaths.remuxIsLossless —
avi mkv — video supported lossy audioTracks, chapters, dimensions, hdr, metadata, subtitleTracks, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). HDR10 and HLG are kept with HEVC, AV1, VP9 or ProRes; other codecs get a tone-mapped SDR picture. Dolby Vision metadata is not carried. Only the first video track is kept; data and attachment streams are dropped. VideoPaths.conversion —
avi mkv video.remux video supported lossless audioTracks, chapters, colorProfile, dimensions, hdr, metadata, subtitleTracks, timing Works only when the codecs inside the file are allowed in Matroska; otherwise it stops and says which stream is in the way. With trimming, the result starts on the keyframe at or before the requested start (not frame-accurate). VideoPaths.remuxIsLossless —
avi mov — video supported lossy audioTracks, chapters, dimensions, hdr, metadata, subtitleTracks, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). Text subtitles are kept as plain timed text (styling is lost); image-based subtitles are dropped. HDR10 and HLG are kept with HEVC, AV1, VP9 or ProRes; other codecs get a tone-mapped SDR picture. Dolby Vision metadata is not carried. Only the first video track is kept; data and attachment streams are dropped. VideoPaths.conversion —
avi mov video.remux video supported lossless audioTracks, chapters, colorProfile, dimensions, hdr, metadata, subtitleTracks, timing Works only when the codecs inside the file are allowed in QuickTime Movie; otherwise it stops and says which stream is in the way. With trimming, the result starts on the keyframe at or before the requested start (not frame-accurate). Text subtitles are rewritten in the container's own text format; image-based subtitles are dropped. VideoPaths.remuxIsLossless —
avi mp3 video.extract-audio video supported partial metadata Only the sound of one audio track is kept; the picture is not. The track is copied bit for bit when it already uses this codec; otherwise it is re-encoded. When re-encoded, the audio goes through a lossy codec again. Stereo at most: surround sound is mixed down. VideoPaths.extractAudio —
avi mp4 — video supported lossy audioTracks, chapters, dimensions, hdr, metadata, subtitleTracks, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). Text subtitles are kept as plain timed text (styling is lost); image-based subtitles are dropped. HDR10 and HLG are kept with HEVC, AV1, VP9 or ProRes; other codecs get a tone-mapped SDR picture. Dolby Vision metadata is not carried. Only the first video track is kept; data and attachment streams are dropped. VideoPaths.conversion —
avi mp4 video.remux video supported lossless audioTracks, chapters, colorProfile, dimensions, hdr, metadata, subtitleTracks, timing Works only when the codecs inside the file are allowed in MP4; otherwise it stops and says which stream is in the way. With trimming, the result starts on the keyframe at or before the requested start (not frame-accurate). Text subtitles are rewritten in the container's own text format; image-based subtitles are dropped. VideoPaths.remuxIsLossless —
avi mpeg — video supported lossy audioTracks, dimensions, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). MPEG Program Stream carries no descriptive tags here. Chapters are dropped. Subtitle tracks are dropped: MPEG Program Stream cannot carry them here. HDR video is tone-mapped to standard dynamic range; 10-bit pictures become 8-bit. Only the first video track is kept; data and attachment streams are dropped. MPEG-2 (or MPEG-1) video at standard frame rates only, with MP2 or AC-3 sound. VideoPaths.conversion —
avi mts — video supported lossy audioTracks, dimensions, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). AVCHD carries no descriptive tags here. Chapters are dropped. Subtitle tracks are dropped: AVCHD cannot carry them here. HDR video is tone-mapped to standard dynamic range; 10-bit pictures become 8-bit. Only the first video track is kept; data and attachment streams are dropped. An AVCHD-style transport stream (H.264 with AC-3 or AAC), not a full AVCHD folder structure. VideoPaths.conversion —
avi ogg video.extract-audio video supported partial metadata Only the sound of one audio track is kept; the picture is not. The track is copied bit for bit when it already uses this codec; otherwise it is re-encoded. When re-encoded, the audio goes through a lossy codec again. VideoPaths.extractAudio —
avi opus video.extract-audio video supported partial metadata Only the sound of one audio track is kept; the picture is not. The track is copied bit for bit when it already uses this codec; otherwise it is re-encoded. When re-encoded, the audio goes through a lossy codec again. VideoPaths.extractAudio —
avi ts — video supported lossy audioTracks, dimensions, hdr, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). MPEG Transport Stream carries no descriptive tags here. Chapters are dropped. Subtitle tracks are dropped: MPEG Transport Stream cannot carry them here. HDR10 and HLG are kept with HEVC, AV1, VP9 or ProRes; other codecs get a tone-mapped SDR picture. Dolby Vision metadata is not carried. Only the first video track is kept; data and attachment streams are dropped. VideoPaths.conversion —
avi wav video.extract-audio video supported partial metadata Only the sound of one audio track is kept; the picture is not. VideoPaths.extractAudio —
avi webm — video supported lossy audioTracks, chapters, dimensions, hdr, metadata, subtitleTracks, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). Text subtitles are kept as plain timed text (styling is lost); image-based subtitles are dropped. HDR10 and HLG are kept with HEVC, AV1, VP9 or ProRes; other codecs get a tone-mapped SDR picture. Dolby Vision metadata is not carried. Only the first video track is kept; data and attachment streams are dropped. VideoPaths.conversion —
avi wmv — video supported lossy audioTracks, dimensions, metadata, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). Chapters are dropped. Subtitle tracks are dropped: Windows Media Video cannot carry them here. HDR video is tone-mapped to standard dynamic range; 10-bit pictures become 8-bit. Only the first video track is kept; data and attachment streams are dropped. Written as WMV8 video with WMA2 audio (stereo), the Windows Media codecs this build can encode — not WMV9/VC-1. VideoPaths.conversion —
flv aac video.extract-audio video supported partial — Only the sound of one audio track is kept; the picture is not. The track is copied bit for bit when it already uses this codec; otherwise it is re-encoded. When re-encoded, the audio goes through a lossy codec again. VideoPaths.extractAudio —
flv aiff video.extract-audio video supported partial metadata Only the sound of one audio track is kept; the picture is not. VideoPaths.extractAudio —
flv apng — video supported lossy animation No sound. Full-colour frames without inter-frame video compression: files are large. The frame rate and the size are reduced (12 fps and 480 px wide unless changed). HDR video is tone-mapped to SDR. VideoPaths.animatedImage —
flv avi — video supported lossy audioTracks, dimensions, metadata, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). Chapters are dropped. Subtitle tracks are dropped: AVI cannot carry them here. HDR video is tone-mapped to standard dynamic range; 10-bit pictures become 8-bit. Only the first video track is kept; data and attachment streams are dropped. MPEG-4 Part 2 video; AVI cannot hold HEVC or AV1 here. VideoPaths.conversion —
flv flac video.extract-audio video supported partial metadata Only the sound of one audio track is kept; the picture is not. The track is copied bit for bit when it already uses this codec; otherwise it is re-encoded. VideoPaths.extractAudio —
flv flv — video supported lossy audioTracks, dimensions, metadata, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). Chapters are dropped. Subtitle tracks are dropped: Flash Video cannot carry them here. HDR video is tone-mapped to standard dynamic range; 10-bit pictures become 8-bit. Only the first video track is kept; data and attachment streams are dropped. One audio track only. VideoPaths.conversion —
flv gif — video supported lossy animation No sound. 256 colours per frame, from a palette computed for this video, with dithering. The frame rate and the size are reduced (12 fps and 480 px wide unless changed). HDR video is tone-mapped to SDR. VideoPaths.animatedImage —
flv m4a video.extract-audio video supported partial metadata Only the sound of one audio track is kept; the picture is not. The track is copied bit for bit when it already uses this codec; otherwise it is re-encoded. When re-encoded, the audio goes through a lossy codec again. VideoPaths.extractAudio —
flv m4v — video supported lossy audioTracks, chapters, dimensions, hdr, metadata, subtitleTracks, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). Text subtitles are kept as plain timed text (styling is lost); image-based subtitles are dropped. HDR10 and HLG are kept with HEVC, AV1, VP9 or ProRes; other codecs get a tone-mapped SDR picture. Dolby Vision metadata is not carried. Only the first video track is kept; data and attachment streams are dropped. VideoPaths.conversion —
flv m4v video.remux video supported lossless audioTracks, chapters, colorProfile, dimensions, hdr, metadata, subtitleTracks, timing Works only when the codecs inside the file are allowed in M4V; otherwise it stops and says which stream is in the way. With trimming, the result starts on the keyframe at or before the requested start (not frame-accurate). Text subtitles are rewritten in the container's own text format; image-based subtitles are dropped. VideoPaths.remuxIsLossless —
flv mkv — video supported lossy audioTracks, chapters, dimensions, hdr, metadata, subtitleTracks, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). HDR10 and HLG are kept with HEVC, AV1, VP9 or ProRes; other codecs get a tone-mapped SDR picture. Dolby Vision metadata is not carried. Only the first video track is kept; data and attachment streams are dropped. VideoPaths.conversion —
flv mkv video.remux video supported lossless audioTracks, chapters, colorProfile, dimensions, hdr, metadata, subtitleTracks, timing Works only when the codecs inside the file are allowed in Matroska; otherwise it stops and says which stream is in the way. With trimming, the result starts on the keyframe at or before the requested start (not frame-accurate). VideoPaths.remuxIsLossless —
flv mov — video supported lossy audioTracks, chapters, dimensions, hdr, metadata, subtitleTracks, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). Text subtitles are kept as plain timed text (styling is lost); image-based subtitles are dropped. HDR10 and HLG are kept with HEVC, AV1, VP9 or ProRes; other codecs get a tone-mapped SDR picture. Dolby Vision metadata is not carried. Only the first video track is kept; data and attachment streams are dropped. VideoPaths.conversion —
flv mov video.remux video supported lossless audioTracks, chapters, colorProfile, dimensions, hdr, metadata, subtitleTracks, timing Works only when the codecs inside the file are allowed in QuickTime Movie; otherwise it stops and says which stream is in the way. With trimming, the result starts on the keyframe at or before the requested start (not frame-accurate). Text subtitles are rewritten in the container's own text format; image-based subtitles are dropped. VideoPaths.remuxIsLossless —
flv mp3 video.extract-audio video supported partial metadata Only the sound of one audio track is kept; the picture is not. The track is copied bit for bit when it already uses this codec; otherwise it is re-encoded. When re-encoded, the audio goes through a lossy codec again. Stereo at most: surround sound is mixed down. VideoPaths.extractAudio —
flv mp4 — video supported lossy audioTracks, chapters, dimensions, hdr, metadata, subtitleTracks, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). Text subtitles are kept as plain timed text (styling is lost); image-based subtitles are dropped. HDR10 and HLG are kept with HEVC, AV1, VP9 or ProRes; other codecs get a tone-mapped SDR picture. Dolby Vision metadata is not carried. Only the first video track is kept; data and attachment streams are dropped. VideoPaths.conversion —
flv mp4 video.remux video supported lossless audioTracks, chapters, colorProfile, dimensions, hdr, metadata, subtitleTracks, timing Works only when the codecs inside the file are allowed in MP4; otherwise it stops and says which stream is in the way. With trimming, the result starts on the keyframe at or before the requested start (not frame-accurate). Text subtitles are rewritten in the container's own text format; image-based subtitles are dropped. VideoPaths.remuxIsLossless —
flv mpeg — video supported lossy audioTracks, dimensions, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). MPEG Program Stream carries no descriptive tags here. Chapters are dropped. Subtitle tracks are dropped: MPEG Program Stream cannot carry them here. HDR video is tone-mapped to standard dynamic range; 10-bit pictures become 8-bit. Only the first video track is kept; data and attachment streams are dropped. MPEG-2 (or MPEG-1) video at standard frame rates only, with MP2 or AC-3 sound. VideoPaths.conversion —
flv mts — video supported lossy audioTracks, dimensions, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). AVCHD carries no descriptive tags here. Chapters are dropped. Subtitle tracks are dropped: AVCHD cannot carry them here. HDR video is tone-mapped to standard dynamic range; 10-bit pictures become 8-bit. Only the first video track is kept; data and attachment streams are dropped. An AVCHD-style transport stream (H.264 with AC-3 or AAC), not a full AVCHD folder structure. VideoPaths.conversion —
flv mts video.remux video supported lossless audioTracks, colorProfile, dimensions, hdr, timing Works only when the codecs inside the file are allowed in AVCHD; otherwise it stops and says which stream is in the way. With trimming, the result starts on the keyframe at or before the requested start (not frame-accurate). Subtitle tracks are dropped: AVCHD cannot carry them here. Chapters are dropped. VideoPaths.remuxIsLossless —
flv ogg video.extract-audio video supported partial metadata Only the sound of one audio track is kept; the picture is not. The track is copied bit for bit when it already uses this codec; otherwise it is re-encoded. When re-encoded, the audio goes through a lossy codec again. VideoPaths.extractAudio —
flv opus video.extract-audio video supported partial metadata Only the sound of one audio track is kept; the picture is not. The track is copied bit for bit when it already uses this codec; otherwise it is re-encoded. When re-encoded, the audio goes through a lossy codec again. VideoPaths.extractAudio —
flv ts — video supported lossy audioTracks, dimensions, hdr, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). MPEG Transport Stream carries no descriptive tags here. Chapters are dropped. Subtitle tracks are dropped: MPEG Transport Stream cannot carry them here. HDR10 and HLG are kept with HEVC, AV1, VP9 or ProRes; other codecs get a tone-mapped SDR picture. Dolby Vision metadata is not carried. Only the first video track is kept; data and attachment streams are dropped. VideoPaths.conversion —
flv ts video.remux video supported lossless audioTracks, colorProfile, dimensions, hdr, timing Works only when the codecs inside the file are allowed in MPEG Transport Stream; otherwise it stops and says which stream is in the way. With trimming, the result starts on the keyframe at or before the requested start (not frame-accurate). Subtitle tracks are dropped: MPEG Transport Stream cannot carry them here. Chapters are dropped. VideoPaths.remuxIsLossless —
flv wav video.extract-audio video supported partial metadata Only the sound of one audio track is kept; the picture is not. VideoPaths.extractAudio —
flv webm — video supported lossy audioTracks, chapters, dimensions, hdr, metadata, subtitleTracks, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). Text subtitles are kept as plain timed text (styling is lost); image-based subtitles are dropped. HDR10 and HLG are kept with HEVC, AV1, VP9 or ProRes; other codecs get a tone-mapped SDR picture. Dolby Vision metadata is not carried. Only the first video track is kept; data and attachment streams are dropped. VideoPaths.conversion —
flv wmv — video supported lossy audioTracks, dimensions, metadata, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). Chapters are dropped. Subtitle tracks are dropped: Windows Media Video cannot carry them here. HDR video is tone-mapped to standard dynamic range; 10-bit pictures become 8-bit. Only the first video track is kept; data and attachment streams are dropped. Written as WMV8 video with WMA2 audio (stereo), the Windows Media codecs this build can encode — not WMV9/VC-1. VideoPaths.conversion —
m4v aac video.extract-audio video supported partial — Only the sound of one audio track is kept; the picture is not. The track is copied bit for bit when it already uses this codec; otherwise it is re-encoded. When re-encoded, the audio goes through a lossy codec again. VideoPaths.extractAudio —
m4v aiff video.extract-audio video supported partial metadata Only the sound of one audio track is kept; the picture is not. VideoPaths.extractAudio —
m4v apng — video supported lossy animation No sound. Full-colour frames without inter-frame video compression: files are large. The frame rate and the size are reduced (12 fps and 480 px wide unless changed). HDR video is tone-mapped to SDR. VideoPaths.animatedImage —
m4v ass video.extract-subtitles video supported partial timing One text subtitle track is written; picture and sound are not. Image-based subtitles (DVD, Blu-ray PGS) cannot be converted to text: there is no OCR. Plain subtitles get the default ASS style. VideoPaths.extractSubtitles —
m4v avi — video supported lossy audioTracks, dimensions, metadata, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). Chapters are dropped. Subtitle tracks are dropped: AVI cannot carry them here. HDR video is tone-mapped to standard dynamic range; 10-bit pictures become 8-bit. Only the first video track is kept; data and attachment streams are dropped. MPEG-4 Part 2 video; AVI cannot hold HEVC or AV1 here. VideoPaths.conversion —
m4v flac video.extract-audio video supported partial metadata Only the sound of one audio track is kept; the picture is not. The track is copied bit for bit when it already uses this codec; otherwise it is re-encoded. VideoPaths.extractAudio —
m4v flv — video supported lossy audioTracks, dimensions, metadata, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). Chapters are dropped. Subtitle tracks are dropped: Flash Video cannot carry them here. HDR video is tone-mapped to standard dynamic range; 10-bit pictures become 8-bit. Only the first video track is kept; data and attachment streams are dropped. One audio track only. VideoPaths.conversion —
m4v flv video.remux video supported lossless audioTracks, colorProfile, dimensions, hdr, metadata, timing Works only when the codecs inside the file are allowed in Flash Video; otherwise it stops and says which stream is in the way. With trimming, the result starts on the keyframe at or before the requested start (not frame-accurate). Subtitle tracks are dropped: Flash Video cannot carry them here. Chapters are dropped. VideoPaths.remuxIsLossless —
m4v gif — video supported lossy animation No sound. 256 colours per frame, from a palette computed for this video, with dithering. The frame rate and the size are reduced (12 fps and 480 px wide unless changed). HDR video is tone-mapped to SDR. VideoPaths.animatedImage —
m4v m4a video.extract-audio video supported partial metadata Only the sound of one audio track is kept; the picture is not. The track is copied bit for bit when it already uses this codec; otherwise it is re-encoded. When re-encoded, the audio goes through a lossy codec again. VideoPaths.extractAudio —
m4v m4v — video supported lossy audioTracks, chapters, dimensions, hdr, metadata, subtitleTracks, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). Text subtitles are kept as plain timed text (styling is lost); image-based subtitles are dropped. HDR10 and HLG are kept with HEVC, AV1, VP9 or ProRes; other codecs get a tone-mapped SDR picture. Dolby Vision metadata is not carried. Only the first video track is kept; data and attachment streams are dropped. VideoPaths.conversion —
m4v mkv — video supported lossy audioTracks, chapters, dimensions, hdr, metadata, subtitleTracks, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). HDR10 and HLG are kept with HEVC, AV1, VP9 or ProRes; other codecs get a tone-mapped SDR picture. Dolby Vision metadata is not carried. Only the first video track is kept; data and attachment streams are dropped. VideoPaths.conversion —
m4v mkv video.remux video supported lossless audioTracks, chapters, colorProfile, dimensions, hdr, metadata, subtitleTracks, timing Works only when the codecs inside the file are allowed in Matroska; otherwise it stops and says which stream is in the way. With trimming, the result starts on the keyframe at or before the requested start (not frame-accurate). VideoPaths.remuxIsLossless —
m4v mov — video supported lossy audioTracks, chapters, dimensions, hdr, metadata, subtitleTracks, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). Text subtitles are kept as plain timed text (styling is lost); image-based subtitles are dropped. HDR10 and HLG are kept with HEVC, AV1, VP9 or ProRes; other codecs get a tone-mapped SDR picture. Dolby Vision metadata is not carried. Only the first video track is kept; data and attachment streams are dropped. VideoPaths.conversion —
m4v mov video.remux video supported lossless audioTracks, chapters, colorProfile, dimensions, hdr, metadata, subtitleTracks, timing Works only when the codecs inside the file are allowed in QuickTime Movie; otherwise it stops and says which stream is in the way. With trimming, the result starts on the keyframe at or before the requested start (not frame-accurate). Text subtitles are rewritten in the container's own text format; image-based subtitles are dropped. VideoPaths.remuxIsLossless —
m4v mp3 video.extract-audio video supported partial metadata Only the sound of one audio track is kept; the picture is not. The track is copied bit for bit when it already uses this codec; otherwise it is re-encoded. When re-encoded, the audio goes through a lossy codec again. Stereo at most: surround sound is mixed down. VideoPaths.extractAudio —
m4v mp4 — video supported lossy audioTracks, chapters, dimensions, hdr, metadata, subtitleTracks, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). Text subtitles are kept as plain timed text (styling is lost); image-based subtitles are dropped. HDR10 and HLG are kept with HEVC, AV1, VP9 or ProRes; other codecs get a tone-mapped SDR picture. Dolby Vision metadata is not carried. Only the first video track is kept; data and attachment streams are dropped. VideoPaths.conversion —
m4v mp4 video.remux video supported lossless audioTracks, chapters, colorProfile, dimensions, hdr, metadata, subtitleTracks, timing Works only when the codecs inside the file are allowed in MP4; otherwise it stops and says which stream is in the way. With trimming, the result starts on the keyframe at or before the requested start (not frame-accurate). Text subtitles are rewritten in the container's own text format; image-based subtitles are dropped. VideoPaths.remuxIsLossless —
m4v mpeg — video supported lossy audioTracks, dimensions, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). MPEG Program Stream carries no descriptive tags here. Chapters are dropped. Subtitle tracks are dropped: MPEG Program Stream cannot carry them here. HDR video is tone-mapped to standard dynamic range; 10-bit pictures become 8-bit. Only the first video track is kept; data and attachment streams are dropped. MPEG-2 (or MPEG-1) video at standard frame rates only, with MP2 or AC-3 sound. VideoPaths.conversion —
m4v mts — video supported lossy audioTracks, dimensions, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). AVCHD carries no descriptive tags here. Chapters are dropped. Subtitle tracks are dropped: AVCHD cannot carry them here. HDR video is tone-mapped to standard dynamic range; 10-bit pictures become 8-bit. Only the first video track is kept; data and attachment streams are dropped. An AVCHD-style transport stream (H.264 with AC-3 or AAC), not a full AVCHD folder structure. VideoPaths.conversion —
m4v mts video.remux video supported lossless audioTracks, colorProfile, dimensions, hdr, timing Works only when the codecs inside the file are allowed in AVCHD; otherwise it stops and says which stream is in the way. With trimming, the result starts on the keyframe at or before the requested start (not frame-accurate). Subtitle tracks are dropped: AVCHD cannot carry them here. Chapters are dropped. VideoPaths.remuxIsLossless —
m4v ogg video.extract-audio video supported partial metadata Only the sound of one audio track is kept; the picture is not. The track is copied bit for bit when it already uses this codec; otherwise it is re-encoded. When re-encoded, the audio goes through a lossy codec again. VideoPaths.extractAudio —
m4v opus video.extract-audio video supported partial metadata Only the sound of one audio track is kept; the picture is not. The track is copied bit for bit when it already uses this codec; otherwise it is re-encoded. When re-encoded, the audio goes through a lossy codec again. VideoPaths.extractAudio —
m4v srt video.extract-subtitles video supported partial timing One text subtitle track is written; picture and sound are not. Image-based subtitles (DVD, Blu-ray PGS) cannot be converted to text: there is no OCR. Styling of ASS/SSA subtitles (fonts, colours, positions) is lost. VideoPaths.extractSubtitles —
m4v ts — video supported lossy audioTracks, dimensions, hdr, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). MPEG Transport Stream carries no descriptive tags here. Chapters are dropped. Subtitle tracks are dropped: MPEG Transport Stream cannot carry them here. HDR10 and HLG are kept with HEVC, AV1, VP9 or ProRes; other codecs get a tone-mapped SDR picture. Dolby Vision metadata is not carried. Only the first video track is kept; data and attachment streams are dropped. VideoPaths.conversion —
m4v ts video.remux video supported lossless audioTracks, colorProfile, dimensions, hdr, timing Works only when the codecs inside the file are allowed in MPEG Transport Stream; otherwise it stops and says which stream is in the way. With trimming, the result starts on the keyframe at or before the requested start (not frame-accurate). Subtitle tracks are dropped: MPEG Transport Stream cannot carry them here. Chapters are dropped. VideoPaths.remuxIsLossless —
m4v vtt video.extract-subtitles video supported partial timing One text subtitle track is written; picture and sound are not. Image-based subtitles (DVD, Blu-ray PGS) cannot be converted to text: there is no OCR. Styling of ASS/SSA subtitles (fonts, colours, positions) is lost. VideoPaths.extractSubtitles —
m4v wav video.extract-audio video supported partial metadata Only the sound of one audio track is kept; the picture is not. VideoPaths.extractAudio —
m4v webm — video supported lossy audioTracks, chapters, dimensions, hdr, metadata, subtitleTracks, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). Text subtitles are kept as plain timed text (styling is lost); image-based subtitles are dropped. HDR10 and HLG are kept with HEVC, AV1, VP9 or ProRes; other codecs get a tone-mapped SDR picture. Dolby Vision metadata is not carried. Only the first video track is kept; data and attachment streams are dropped. VideoPaths.conversion —
m4v wmv — video supported lossy audioTracks, dimensions, metadata, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). Chapters are dropped. Subtitle tracks are dropped: Windows Media Video cannot carry them here. HDR video is tone-mapped to standard dynamic range; 10-bit pictures become 8-bit. Only the first video track is kept; data and attachment streams are dropped. Written as WMV8 video with WMA2 audio (stereo), the Windows Media codecs this build can encode — not WMV9/VC-1. VideoPaths.conversion —
mkv aac video.extract-audio video supported partial — Only the sound of one audio track is kept; the picture is not. The track is copied bit for bit when it already uses this codec; otherwise it is re-encoded. When re-encoded, the audio goes through a lossy codec again. VideoPaths.extractAudio —
mkv aiff video.extract-audio video supported partial metadata Only the sound of one audio track is kept; the picture is not. VideoPaths.extractAudio —
mkv apng — video supported lossy animation No sound. Full-colour frames without inter-frame video compression: files are large. The frame rate and the size are reduced (12 fps and 480 px wide unless changed). HDR video is tone-mapped to SDR. VideoPaths.animatedImage —
mkv ass video.extract-subtitles video supported partial timing One text subtitle track is written; picture and sound are not. Image-based subtitles (DVD, Blu-ray PGS) cannot be converted to text: there is no OCR. Plain subtitles get the default ASS style. VideoPaths.extractSubtitles —
mkv avi — video supported lossy audioTracks, dimensions, metadata, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). Chapters are dropped. Subtitle tracks are dropped: AVI cannot carry them here. HDR video is tone-mapped to standard dynamic range; 10-bit pictures become 8-bit. Only the first video track is kept; data and attachment streams are dropped. MPEG-4 Part 2 video; AVI cannot hold HEVC or AV1 here. VideoPaths.conversion —
mkv flac video.extract-audio video supported partial metadata Only the sound of one audio track is kept; the picture is not. The track is copied bit for bit when it already uses this codec; otherwise it is re-encoded. VideoPaths.extractAudio —
mkv flv — video supported lossy audioTracks, dimensions, metadata, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). Chapters are dropped. Subtitle tracks are dropped: Flash Video cannot carry them here. HDR video is tone-mapped to standard dynamic range; 10-bit pictures become 8-bit. Only the first video track is kept; data and attachment streams are dropped. One audio track only. VideoPaths.conversion —
mkv flv video.remux video supported lossless audioTracks, colorProfile, dimensions, hdr, metadata, timing Works only when the codecs inside the file are allowed in Flash Video; otherwise it stops and says which stream is in the way. With trimming, the result starts on the keyframe at or before the requested start (not frame-accurate). Subtitle tracks are dropped: Flash Video cannot carry them here. Chapters are dropped. VideoPaths.remuxIsLossless —
mkv gif — video supported lossy animation No sound. 256 colours per frame, from a palette computed for this video, with dithering. The frame rate and the size are reduced (12 fps and 480 px wide unless changed). HDR video is tone-mapped to SDR. VideoPaths.animatedImage —
mkv m4a video.extract-audio video supported partial metadata Only the sound of one audio track is kept; the picture is not. The track is copied bit for bit when it already uses this codec; otherwise it is re-encoded. When re-encoded, the audio goes through a lossy codec again. VideoPaths.extractAudio —
mkv m4v — video supported lossy audioTracks, chapters, dimensions, hdr, metadata, subtitleTracks, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). Text subtitles are kept as plain timed text (styling is lost); image-based subtitles are dropped. HDR10 and HLG are kept with HEVC, AV1, VP9 or ProRes; other codecs get a tone-mapped SDR picture. Dolby Vision metadata is not carried. Only the first video track is kept; data and attachment streams are dropped. VideoPaths.conversion —
mkv m4v video.remux video supported lossless audioTracks, chapters, colorProfile, dimensions, hdr, metadata, subtitleTracks, timing Works only when the codecs inside the file are allowed in M4V; otherwise it stops and says which stream is in the way. With trimming, the result starts on the keyframe at or before the requested start (not frame-accurate). Text subtitles are rewritten in the container's own text format; image-based subtitles are dropped. VideoPaths.remuxIsLossless —
mkv mkv — video supported lossy audioTracks, chapters, dimensions, hdr, metadata, subtitleTracks, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). HDR10 and HLG are kept with HEVC, AV1, VP9 or ProRes; other codecs get a tone-mapped SDR picture. Dolby Vision metadata is not carried. Only the first video track is kept; data and attachment streams are dropped. VideoPaths.conversion —
mkv mov — video supported lossy audioTracks, chapters, dimensions, hdr, metadata, subtitleTracks, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). Text subtitles are kept as plain timed text (styling is lost); image-based subtitles are dropped. HDR10 and HLG are kept with HEVC, AV1, VP9 or ProRes; other codecs get a tone-mapped SDR picture. Dolby Vision metadata is not carried. Only the first video track is kept; data and attachment streams are dropped. VideoPaths.conversion —
mkv mov video.remux video supported lossless audioTracks, chapters, colorProfile, dimensions, hdr, metadata, subtitleTracks, timing Works only when the codecs inside the file are allowed in QuickTime Movie; otherwise it stops and says which stream is in the way. With trimming, the result starts on the keyframe at or before the requested start (not frame-accurate). Text subtitles are rewritten in the container's own text format; image-based subtitles are dropped. VideoPaths.remuxIsLossless —
mkv mp3 video.extract-audio video supported partial metadata Only the sound of one audio track is kept; the picture is not. The track is copied bit for bit when it already uses this codec; otherwise it is re-encoded. When re-encoded, the audio goes through a lossy codec again. Stereo at most: surround sound is mixed down. VideoPaths.extractAudio —
mkv mp4 — video supported lossy audioTracks, chapters, dimensions, hdr, metadata, subtitleTracks, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). Text subtitles are kept as plain timed text (styling is lost); image-based subtitles are dropped. HDR10 and HLG are kept with HEVC, AV1, VP9 or ProRes; other codecs get a tone-mapped SDR picture. Dolby Vision metadata is not carried. Only the first video track is kept; data and attachment streams are dropped. VideoPaths.conversion —
mkv mp4 video.remux video supported lossless audioTracks, chapters, colorProfile, dimensions, hdr, metadata, subtitleTracks, timing Works only when the codecs inside the file are allowed in MP4; otherwise it stops and says which stream is in the way. With trimming, the result starts on the keyframe at or before the requested start (not frame-accurate). Text subtitles are rewritten in the container's own text format; image-based subtitles are dropped. VideoPaths.remuxIsLossless —
mkv mpeg — video supported lossy audioTracks, dimensions, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). MPEG Program Stream carries no descriptive tags here. Chapters are dropped. Subtitle tracks are dropped: MPEG Program Stream cannot carry them here. HDR video is tone-mapped to standard dynamic range; 10-bit pictures become 8-bit. Only the first video track is kept; data and attachment streams are dropped. MPEG-2 (or MPEG-1) video at standard frame rates only, with MP2 or AC-3 sound. VideoPaths.conversion —
mkv mpeg video.remux video supported lossless audioTracks, colorProfile, dimensions, hdr, timing Works only when the codecs inside the file are allowed in MPEG Program Stream; otherwise it stops and says which stream is in the way. With trimming, the result starts on the keyframe at or before the requested start (not frame-accurate). Subtitle tracks are dropped: MPEG Program Stream cannot carry them here. Chapters are dropped. VideoPaths.remuxIsLossless —
mkv mts — video supported lossy audioTracks, dimensions, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). AVCHD carries no descriptive tags here. Chapters are dropped. Subtitle tracks are dropped: AVCHD cannot carry them here. HDR video is tone-mapped to standard dynamic range; 10-bit pictures become 8-bit. Only the first video track is kept; data and attachment streams are dropped. An AVCHD-style transport stream (H.264 with AC-3 or AAC), not a full AVCHD folder structure. VideoPaths.conversion —
mkv mts video.remux video supported lossless audioTracks, colorProfile, dimensions, hdr, timing Works only when the codecs inside the file are allowed in AVCHD; otherwise it stops and says which stream is in the way. With trimming, the result starts on the keyframe at or before the requested start (not frame-accurate). Subtitle tracks are dropped: AVCHD cannot carry them here. Chapters are dropped. VideoPaths.remuxIsLossless —
mkv ogg video.extract-audio video supported partial metadata Only the sound of one audio track is kept; the picture is not. The track is copied bit for bit when it already uses this codec; otherwise it is re-encoded. When re-encoded, the audio goes through a lossy codec again. VideoPaths.extractAudio —
mkv opus video.extract-audio video supported partial metadata Only the sound of one audio track is kept; the picture is not. The track is copied bit for bit when it already uses this codec; otherwise it is re-encoded. When re-encoded, the audio goes through a lossy codec again. VideoPaths.extractAudio —
mkv srt video.extract-subtitles video supported partial timing One text subtitle track is written; picture and sound are not. Image-based subtitles (DVD, Blu-ray PGS) cannot be converted to text: there is no OCR. Styling of ASS/SSA subtitles (fonts, colours, positions) is lost. VideoPaths.extractSubtitles —
mkv ts — video supported lossy audioTracks, dimensions, hdr, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). MPEG Transport Stream carries no descriptive tags here. Chapters are dropped. Subtitle tracks are dropped: MPEG Transport Stream cannot carry them here. HDR10 and HLG are kept with HEVC, AV1, VP9 or ProRes; other codecs get a tone-mapped SDR picture. Dolby Vision metadata is not carried. Only the first video track is kept; data and attachment streams are dropped. VideoPaths.conversion —
mkv ts video.remux video supported lossless audioTracks, colorProfile, dimensions, hdr, timing Works only when the codecs inside the file are allowed in MPEG Transport Stream; otherwise it stops and says which stream is in the way. With trimming, the result starts on the keyframe at or before the requested start (not frame-accurate). Subtitle tracks are dropped: MPEG Transport Stream cannot carry them here. Chapters are dropped. VideoPaths.remuxIsLossless —
mkv vtt video.extract-subtitles video supported partial timing One text subtitle track is written; picture and sound are not. Image-based subtitles (DVD, Blu-ray PGS) cannot be converted to text: there is no OCR. Styling of ASS/SSA subtitles (fonts, colours, positions) is lost. VideoPaths.extractSubtitles —
mkv wav video.extract-audio video supported partial metadata Only the sound of one audio track is kept; the picture is not. VideoPaths.extractAudio —
mkv webm — video supported lossy audioTracks, chapters, dimensions, hdr, metadata, subtitleTracks, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). Text subtitles are kept as plain timed text (styling is lost); image-based subtitles are dropped. HDR10 and HLG are kept with HEVC, AV1, VP9 or ProRes; other codecs get a tone-mapped SDR picture. Dolby Vision metadata is not carried. Only the first video track is kept; data and attachment streams are dropped. VideoPaths.conversion —
mkv webm video.remux video supported lossless audioTracks, chapters, colorProfile, dimensions, hdr, metadata, subtitleTracks, timing Works only when the codecs inside the file are allowed in WebM; otherwise it stops and says which stream is in the way. With trimming, the result starts on the keyframe at or before the requested start (not frame-accurate). Text subtitles are rewritten in the container's own text format; image-based subtitles are dropped. VideoPaths.remuxIsLossless —
mkv wmv — video supported lossy audioTracks, dimensions, metadata, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). Chapters are dropped. Subtitle tracks are dropped: Windows Media Video cannot carry them here. HDR video is tone-mapped to standard dynamic range; 10-bit pictures become 8-bit. Only the first video track is kept; data and attachment streams are dropped. Written as WMV8 video with WMA2 audio (stereo), the Windows Media codecs this build can encode — not WMV9/VC-1. VideoPaths.conversion —
mov aac video.extract-audio video supported partial — Only the sound of one audio track is kept; the picture is not. The track is copied bit for bit when it already uses this codec; otherwise it is re-encoded. When re-encoded, the audio goes through a lossy codec again. VideoPaths.extractAudio —
mov aiff video.extract-audio video supported partial metadata Only the sound of one audio track is kept; the picture is not. VideoPaths.extractAudio —
mov apng — video supported lossy animation No sound. Full-colour frames without inter-frame video compression: files are large. The frame rate and the size are reduced (12 fps and 480 px wide unless changed). HDR video is tone-mapped to SDR. VideoPaths.animatedImage —
mov ass video.extract-subtitles video supported partial timing One text subtitle track is written; picture and sound are not. Image-based subtitles (DVD, Blu-ray PGS) cannot be converted to text: there is no OCR. Plain subtitles get the default ASS style. VideoPaths.extractSubtitles —
mov avi — video supported lossy audioTracks, dimensions, metadata, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). Chapters are dropped. Subtitle tracks are dropped: AVI cannot carry them here. HDR video is tone-mapped to standard dynamic range; 10-bit pictures become 8-bit. Only the first video track is kept; data and attachment streams are dropped. MPEG-4 Part 2 video; AVI cannot hold HEVC or AV1 here. VideoPaths.conversion —
mov flac video.extract-audio video supported partial metadata Only the sound of one audio track is kept; the picture is not. The track is copied bit for bit when it already uses this codec; otherwise it is re-encoded. VideoPaths.extractAudio —
mov flv — video supported lossy audioTracks, dimensions, metadata, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). Chapters are dropped. Subtitle tracks are dropped: Flash Video cannot carry them here. HDR video is tone-mapped to standard dynamic range; 10-bit pictures become 8-bit. Only the first video track is kept; data and attachment streams are dropped. One audio track only. VideoPaths.conversion —
mov flv video.remux video supported lossless audioTracks, colorProfile, dimensions, hdr, metadata, timing Works only when the codecs inside the file are allowed in Flash Video; otherwise it stops and says which stream is in the way. With trimming, the result starts on the keyframe at or before the requested start (not frame-accurate). Subtitle tracks are dropped: Flash Video cannot carry them here. Chapters are dropped. VideoPaths.remuxIsLossless —
mov gif — video supported lossy animation No sound. 256 colours per frame, from a palette computed for this video, with dithering. The frame rate and the size are reduced (12 fps and 480 px wide unless changed). HDR video is tone-mapped to SDR. VideoPaths.animatedImage —
mov m4a video.extract-audio video supported partial metadata Only the sound of one audio track is kept; the picture is not. The track is copied bit for bit when it already uses this codec; otherwise it is re-encoded. When re-encoded, the audio goes through a lossy codec again. VideoPaths.extractAudio —
mov m4v — video supported lossy audioTracks, chapters, dimensions, hdr, metadata, subtitleTracks, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). Text subtitles are kept as plain timed text (styling is lost); image-based subtitles are dropped. HDR10 and HLG are kept with HEVC, AV1, VP9 or ProRes; other codecs get a tone-mapped SDR picture. Dolby Vision metadata is not carried. Only the first video track is kept; data and attachment streams are dropped. VideoPaths.conversion —
mov m4v video.remux video supported lossless audioTracks, chapters, colorProfile, dimensions, hdr, metadata, subtitleTracks, timing Works only when the codecs inside the file are allowed in M4V; otherwise it stops and says which stream is in the way. With trimming, the result starts on the keyframe at or before the requested start (not frame-accurate). Text subtitles are rewritten in the container's own text format; image-based subtitles are dropped. VideoPaths.remuxIsLossless —
mov mkv — video supported lossy audioTracks, chapters, dimensions, hdr, metadata, subtitleTracks, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). HDR10 and HLG are kept with HEVC, AV1, VP9 or ProRes; other codecs get a tone-mapped SDR picture. Dolby Vision metadata is not carried. Only the first video track is kept; data and attachment streams are dropped. VideoPaths.conversion —
mov mkv video.remux video supported lossless audioTracks, chapters, colorProfile, dimensions, hdr, metadata, subtitleTracks, timing Works only when the codecs inside the file are allowed in Matroska; otherwise it stops and says which stream is in the way. With trimming, the result starts on the keyframe at or before the requested start (not frame-accurate). VideoPaths.remuxIsLossless —
mov mov — video supported lossy audioTracks, chapters, dimensions, hdr, metadata, subtitleTracks, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). Text subtitles are kept as plain timed text (styling is lost); image-based subtitles are dropped. HDR10 and HLG are kept with HEVC, AV1, VP9 or ProRes; other codecs get a tone-mapped SDR picture. Dolby Vision metadata is not carried. Only the first video track is kept; data and attachment streams are dropped. VideoPaths.conversion —
mov mp3 video.extract-audio video supported partial metadata Only the sound of one audio track is kept; the picture is not. The track is copied bit for bit when it already uses this codec; otherwise it is re-encoded. When re-encoded, the audio goes through a lossy codec again. Stereo at most: surround sound is mixed down. VideoPaths.extractAudio —
mov mp4 — video supported lossy audioTracks, chapters, dimensions, hdr, metadata, subtitleTracks, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). Text subtitles are kept as plain timed text (styling is lost); image-based subtitles are dropped. HDR10 and HLG are kept with HEVC, AV1, VP9 or ProRes; other codecs get a tone-mapped SDR picture. Dolby Vision metadata is not carried. Only the first video track is kept; data and attachment streams are dropped. VideoPaths.conversion —
mov mp4 video.remux video supported lossless audioTracks, chapters, colorProfile, dimensions, hdr, metadata, subtitleTracks, timing Works only when the codecs inside the file are allowed in MP4; otherwise it stops and says which stream is in the way. With trimming, the result starts on the keyframe at or before the requested start (not frame-accurate). Text subtitles are rewritten in the container's own text format; image-based subtitles are dropped. VideoPaths.remuxIsLossless —
mov mpeg — video supported lossy audioTracks, dimensions, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). MPEG Program Stream carries no descriptive tags here. Chapters are dropped. Subtitle tracks are dropped: MPEG Program Stream cannot carry them here. HDR video is tone-mapped to standard dynamic range; 10-bit pictures become 8-bit. Only the first video track is kept; data and attachment streams are dropped. MPEG-2 (or MPEG-1) video at standard frame rates only, with MP2 or AC-3 sound. VideoPaths.conversion —
mov mts — video supported lossy audioTracks, dimensions, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). AVCHD carries no descriptive tags here. Chapters are dropped. Subtitle tracks are dropped: AVCHD cannot carry them here. HDR video is tone-mapped to standard dynamic range; 10-bit pictures become 8-bit. Only the first video track is kept; data and attachment streams are dropped. An AVCHD-style transport stream (H.264 with AC-3 or AAC), not a full AVCHD folder structure. VideoPaths.conversion —
mov mts video.remux video supported lossless audioTracks, colorProfile, dimensions, hdr, timing Works only when the codecs inside the file are allowed in AVCHD; otherwise it stops and says which stream is in the way. With trimming, the result starts on the keyframe at or before the requested start (not frame-accurate). Subtitle tracks are dropped: AVCHD cannot carry them here. Chapters are dropped. VideoPaths.remuxIsLossless —
mov ogg video.extract-audio video supported partial metadata Only the sound of one audio track is kept; the picture is not. The track is copied bit for bit when it already uses this codec; otherwise it is re-encoded. When re-encoded, the audio goes through a lossy codec again. VideoPaths.extractAudio —
mov opus video.extract-audio video supported partial metadata Only the sound of one audio track is kept; the picture is not. The track is copied bit for bit when it already uses this codec; otherwise it is re-encoded. When re-encoded, the audio goes through a lossy codec again. VideoPaths.extractAudio —
mov srt video.extract-subtitles video supported partial timing One text subtitle track is written; picture and sound are not. Image-based subtitles (DVD, Blu-ray PGS) cannot be converted to text: there is no OCR. Styling of ASS/SSA subtitles (fonts, colours, positions) is lost. VideoPaths.extractSubtitles —
mov ts — video supported lossy audioTracks, dimensions, hdr, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). MPEG Transport Stream carries no descriptive tags here. Chapters are dropped. Subtitle tracks are dropped: MPEG Transport Stream cannot carry them here. HDR10 and HLG are kept with HEVC, AV1, VP9 or ProRes; other codecs get a tone-mapped SDR picture. Dolby Vision metadata is not carried. Only the first video track is kept; data and attachment streams are dropped. VideoPaths.conversion —
mov ts video.remux video supported lossless audioTracks, colorProfile, dimensions, hdr, timing Works only when the codecs inside the file are allowed in MPEG Transport Stream; otherwise it stops and says which stream is in the way. With trimming, the result starts on the keyframe at or before the requested start (not frame-accurate). Subtitle tracks are dropped: MPEG Transport Stream cannot carry them here. Chapters are dropped. VideoPaths.remuxIsLossless —
mov vtt video.extract-subtitles video supported partial timing One text subtitle track is written; picture and sound are not. Image-based subtitles (DVD, Blu-ray PGS) cannot be converted to text: there is no OCR. Styling of ASS/SSA subtitles (fonts, colours, positions) is lost. VideoPaths.extractSubtitles —
mov wav video.extract-audio video supported partial metadata Only the sound of one audio track is kept; the picture is not. VideoPaths.extractAudio —
mov webm — video supported lossy audioTracks, chapters, dimensions, hdr, metadata, subtitleTracks, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). Text subtitles are kept as plain timed text (styling is lost); image-based subtitles are dropped. HDR10 and HLG are kept with HEVC, AV1, VP9 or ProRes; other codecs get a tone-mapped SDR picture. Dolby Vision metadata is not carried. Only the first video track is kept; data and attachment streams are dropped. VideoPaths.conversion —
mov wmv — video supported lossy audioTracks, dimensions, metadata, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). Chapters are dropped. Subtitle tracks are dropped: Windows Media Video cannot carry them here. HDR video is tone-mapped to standard dynamic range; 10-bit pictures become 8-bit. Only the first video track is kept; data and attachment streams are dropped. Written as WMV8 video with WMA2 audio (stereo), the Windows Media codecs this build can encode — not WMV9/VC-1. VideoPaths.conversion —
mp4 aac video.extract-audio video supported partial — Only the sound of one audio track is kept; the picture is not. The track is copied bit for bit when it already uses this codec; otherwise it is re-encoded. When re-encoded, the audio goes through a lossy codec again. VideoPaths.extractAudio —
mp4 aiff video.extract-audio video supported partial metadata Only the sound of one audio track is kept; the picture is not. VideoPaths.extractAudio —
mp4 apng — video supported lossy animation No sound. Full-colour frames without inter-frame video compression: files are large. The frame rate and the size are reduced (12 fps and 480 px wide unless changed). HDR video is tone-mapped to SDR. VideoPaths.animatedImage —
mp4 ass video.extract-subtitles video supported partial timing One text subtitle track is written; picture and sound are not. Image-based subtitles (DVD, Blu-ray PGS) cannot be converted to text: there is no OCR. Plain subtitles get the default ASS style. VideoPaths.extractSubtitles —
mp4 avi — video supported lossy audioTracks, dimensions, metadata, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). Chapters are dropped. Subtitle tracks are dropped: AVI cannot carry them here. HDR video is tone-mapped to standard dynamic range; 10-bit pictures become 8-bit. Only the first video track is kept; data and attachment streams are dropped. MPEG-4 Part 2 video; AVI cannot hold HEVC or AV1 here. VideoPaths.conversion —
mp4 flac video.extract-audio video supported partial metadata Only the sound of one audio track is kept; the picture is not. The track is copied bit for bit when it already uses this codec; otherwise it is re-encoded. VideoPaths.extractAudio —
mp4 flv — video supported lossy audioTracks, dimensions, metadata, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). Chapters are dropped. Subtitle tracks are dropped: Flash Video cannot carry them here. HDR video is tone-mapped to standard dynamic range; 10-bit pictures become 8-bit. Only the first video track is kept; data and attachment streams are dropped. One audio track only. VideoPaths.conversion —
mp4 flv video.remux video supported lossless audioTracks, colorProfile, dimensions, hdr, metadata, timing Works only when the codecs inside the file are allowed in Flash Video; otherwise it stops and says which stream is in the way. With trimming, the result starts on the keyframe at or before the requested start (not frame-accurate). Subtitle tracks are dropped: Flash Video cannot carry them here. Chapters are dropped. VideoPaths.remuxIsLossless —
mp4 gif — video supported lossy animation No sound. 256 colours per frame, from a palette computed for this video, with dithering. The frame rate and the size are reduced (12 fps and 480 px wide unless changed). HDR video is tone-mapped to SDR. VideoPaths.animatedImage —
mp4 m4a video.extract-audio video supported partial metadata Only the sound of one audio track is kept; the picture is not. The track is copied bit for bit when it already uses this codec; otherwise it is re-encoded. When re-encoded, the audio goes through a lossy codec again. VideoPaths.extractAudio —
mp4 m4v — video supported lossy audioTracks, chapters, dimensions, hdr, metadata, subtitleTracks, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). Text subtitles are kept as plain timed text (styling is lost); image-based subtitles are dropped. HDR10 and HLG are kept with HEVC, AV1, VP9 or ProRes; other codecs get a tone-mapped SDR picture. Dolby Vision metadata is not carried. Only the first video track is kept; data and attachment streams are dropped. VideoPaths.conversion —
mp4 m4v video.remux video supported lossless audioTracks, chapters, colorProfile, dimensions, hdr, metadata, subtitleTracks, timing Works only when the codecs inside the file are allowed in M4V; otherwise it stops and says which stream is in the way. With trimming, the result starts on the keyframe at or before the requested start (not frame-accurate). Text subtitles are rewritten in the container's own text format; image-based subtitles are dropped. VideoPaths.remuxIsLossless —
mp4 mkv — video supported lossy audioTracks, chapters, dimensions, hdr, metadata, subtitleTracks, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). HDR10 and HLG are kept with HEVC, AV1, VP9 or ProRes; other codecs get a tone-mapped SDR picture. Dolby Vision metadata is not carried. Only the first video track is kept; data and attachment streams are dropped. VideoPaths.conversion —
mp4 mkv video.remux video supported lossless audioTracks, chapters, colorProfile, dimensions, hdr, metadata, subtitleTracks, timing Works only when the codecs inside the file are allowed in Matroska; otherwise it stops and says which stream is in the way. With trimming, the result starts on the keyframe at or before the requested start (not frame-accurate). VideoPaths.remuxIsLossless —
mp4 mov — video supported lossy audioTracks, chapters, dimensions, hdr, metadata, subtitleTracks, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). Text subtitles are kept as plain timed text (styling is lost); image-based subtitles are dropped. HDR10 and HLG are kept with HEVC, AV1, VP9 or ProRes; other codecs get a tone-mapped SDR picture. Dolby Vision metadata is not carried. Only the first video track is kept; data and attachment streams are dropped. VideoPaths.conversion —
mp4 mov video.remux video supported lossless audioTracks, chapters, colorProfile, dimensions, hdr, metadata, subtitleTracks, timing Works only when the codecs inside the file are allowed in QuickTime Movie; otherwise it stops and says which stream is in the way. With trimming, the result starts on the keyframe at or before the requested start (not frame-accurate). Text subtitles are rewritten in the container's own text format; image-based subtitles are dropped. VideoPaths.remuxIsLossless —
mp4 mp3 video.extract-audio video supported partial metadata Only the sound of one audio track is kept; the picture is not. The track is copied bit for bit when it already uses this codec; otherwise it is re-encoded. When re-encoded, the audio goes through a lossy codec again. Stereo at most: surround sound is mixed down. VideoPaths.extractAudio —
mp4 mp4 — video supported lossy audioTracks, chapters, dimensions, hdr, metadata, subtitleTracks, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). Text subtitles are kept as plain timed text (styling is lost); image-based subtitles are dropped. HDR10 and HLG are kept with HEVC, AV1, VP9 or ProRes; other codecs get a tone-mapped SDR picture. Dolby Vision metadata is not carried. Only the first video track is kept; data and attachment streams are dropped. VideoPaths.conversion —
mp4 mpeg — video supported lossy audioTracks, dimensions, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). MPEG Program Stream carries no descriptive tags here. Chapters are dropped. Subtitle tracks are dropped: MPEG Program Stream cannot carry them here. HDR video is tone-mapped to standard dynamic range; 10-bit pictures become 8-bit. Only the first video track is kept; data and attachment streams are dropped. MPEG-2 (or MPEG-1) video at standard frame rates only, with MP2 or AC-3 sound. VideoPaths.conversion —
mp4 mts — video supported lossy audioTracks, dimensions, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). AVCHD carries no descriptive tags here. Chapters are dropped. Subtitle tracks are dropped: AVCHD cannot carry them here. HDR video is tone-mapped to standard dynamic range; 10-bit pictures become 8-bit. Only the first video track is kept; data and attachment streams are dropped. An AVCHD-style transport stream (H.264 with AC-3 or AAC), not a full AVCHD folder structure. VideoPaths.conversion —
mp4 mts video.remux video supported lossless audioTracks, colorProfile, dimensions, hdr, timing Works only when the codecs inside the file are allowed in AVCHD; otherwise it stops and says which stream is in the way. With trimming, the result starts on the keyframe at or before the requested start (not frame-accurate). Subtitle tracks are dropped: AVCHD cannot carry them here. Chapters are dropped. VideoPaths.remuxIsLossless —
mp4 ogg video.extract-audio video supported partial metadata Only the sound of one audio track is kept; the picture is not. The track is copied bit for bit when it already uses this codec; otherwise it is re-encoded. When re-encoded, the audio goes through a lossy codec again. VideoPaths.extractAudio —
mp4 opus video.extract-audio video supported partial metadata Only the sound of one audio track is kept; the picture is not. The track is copied bit for bit when it already uses this codec; otherwise it is re-encoded. When re-encoded, the audio goes through a lossy codec again. VideoPaths.extractAudio —
mp4 srt video.extract-subtitles video supported partial timing One text subtitle track is written; picture and sound are not. Image-based subtitles (DVD, Blu-ray PGS) cannot be converted to text: there is no OCR. Styling of ASS/SSA subtitles (fonts, colours, positions) is lost. VideoPaths.extractSubtitles —
mp4 ts — video supported lossy audioTracks, dimensions, hdr, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). MPEG Transport Stream carries no descriptive tags here. Chapters are dropped. Subtitle tracks are dropped: MPEG Transport Stream cannot carry them here. HDR10 and HLG are kept with HEVC, AV1, VP9 or ProRes; other codecs get a tone-mapped SDR picture. Dolby Vision metadata is not carried. Only the first video track is kept; data and attachment streams are dropped. VideoPaths.conversion —
mp4 ts video.remux video supported lossless audioTracks, colorProfile, dimensions, hdr, timing Works only when the codecs inside the file are allowed in MPEG Transport Stream; otherwise it stops and says which stream is in the way. With trimming, the result starts on the keyframe at or before the requested start (not frame-accurate). Subtitle tracks are dropped: MPEG Transport Stream cannot carry them here. Chapters are dropped. VideoPaths.remuxIsLossless —
mp4 vtt video.extract-subtitles video supported partial timing One text subtitle track is written; picture and sound are not. Image-based subtitles (DVD, Blu-ray PGS) cannot be converted to text: there is no OCR. Styling of ASS/SSA subtitles (fonts, colours, positions) is lost. VideoPaths.extractSubtitles —
mp4 wav video.extract-audio video supported partial metadata Only the sound of one audio track is kept; the picture is not. VideoPaths.extractAudio —
mp4 webm — video supported lossy audioTracks, chapters, dimensions, hdr, metadata, subtitleTracks, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). Text subtitles are kept as plain timed text (styling is lost); image-based subtitles are dropped. HDR10 and HLG are kept with HEVC, AV1, VP9 or ProRes; other codecs get a tone-mapped SDR picture. Dolby Vision metadata is not carried. Only the first video track is kept; data and attachment streams are dropped. VideoPaths.conversion —
mp4 webm video.remux video supported lossless audioTracks, chapters, colorProfile, dimensions, hdr, metadata, subtitleTracks, timing Works only when the codecs inside the file are allowed in WebM; otherwise it stops and says which stream is in the way. With trimming, the result starts on the keyframe at or before the requested start (not frame-accurate). Text subtitles are rewritten in the container's own text format; image-based subtitles are dropped. VideoPaths.remuxIsLossless —
mp4 wmv — video supported lossy audioTracks, dimensions, metadata, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). Chapters are dropped. Subtitle tracks are dropped: Windows Media Video cannot carry them here. HDR video is tone-mapped to standard dynamic range; 10-bit pictures become 8-bit. Only the first video track is kept; data and attachment streams are dropped. Written as WMV8 video with WMA2 audio (stereo), the Windows Media codecs this build can encode — not WMV9/VC-1. VideoPaths.conversion —
mpeg aac video.extract-audio video supported partial — Only the sound of one audio track is kept; the picture is not. The track is copied bit for bit when it already uses this codec; otherwise it is re-encoded. When re-encoded, the audio goes through a lossy codec again. VideoPaths.extractAudio —
mpeg aiff video.extract-audio video supported partial metadata Only the sound of one audio track is kept; the picture is not. VideoPaths.extractAudio —
mpeg apng — video supported lossy animation No sound. Full-colour frames without inter-frame video compression: files are large. The frame rate and the size are reduced (12 fps and 480 px wide unless changed). HDR video is tone-mapped to SDR. VideoPaths.animatedImage —
mpeg avi — video supported lossy audioTracks, dimensions, metadata, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). Chapters are dropped. Subtitle tracks are dropped: AVI cannot carry them here. HDR video is tone-mapped to standard dynamic range; 10-bit pictures become 8-bit. Only the first video track is kept; data and attachment streams are dropped. MPEG-4 Part 2 video; AVI cannot hold HEVC or AV1 here. VideoPaths.conversion —
mpeg flac video.extract-audio video supported partial metadata Only the sound of one audio track is kept; the picture is not. The track is copied bit for bit when it already uses this codec; otherwise it is re-encoded. VideoPaths.extractAudio —
mpeg flv — video supported lossy audioTracks, dimensions, metadata, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). Chapters are dropped. Subtitle tracks are dropped: Flash Video cannot carry them here. HDR video is tone-mapped to standard dynamic range; 10-bit pictures become 8-bit. Only the first video track is kept; data and attachment streams are dropped. One audio track only. VideoPaths.conversion —
mpeg gif — video supported lossy animation No sound. 256 colours per frame, from a palette computed for this video, with dithering. The frame rate and the size are reduced (12 fps and 480 px wide unless changed). HDR video is tone-mapped to SDR. VideoPaths.animatedImage —
mpeg m4a video.extract-audio video supported partial metadata Only the sound of one audio track is kept; the picture is not. The track is copied bit for bit when it already uses this codec; otherwise it is re-encoded. When re-encoded, the audio goes through a lossy codec again. VideoPaths.extractAudio —
mpeg m4v — video supported lossy audioTracks, chapters, dimensions, hdr, metadata, subtitleTracks, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). Text subtitles are kept as plain timed text (styling is lost); image-based subtitles are dropped. HDR10 and HLG are kept with HEVC, AV1, VP9 or ProRes; other codecs get a tone-mapped SDR picture. Dolby Vision metadata is not carried. Only the first video track is kept; data and attachment streams are dropped. VideoPaths.conversion —
mpeg mkv — video supported lossy audioTracks, chapters, dimensions, hdr, metadata, subtitleTracks, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). HDR10 and HLG are kept with HEVC, AV1, VP9 or ProRes; other codecs get a tone-mapped SDR picture. Dolby Vision metadata is not carried. Only the first video track is kept; data and attachment streams are dropped. VideoPaths.conversion —
mpeg mkv video.remux video supported lossless audioTracks, chapters, colorProfile, dimensions, hdr, metadata, subtitleTracks, timing Works only when the codecs inside the file are allowed in Matroska; otherwise it stops and says which stream is in the way. With trimming, the result starts on the keyframe at or before the requested start (not frame-accurate). VideoPaths.remuxIsLossless —
mpeg mov — video supported lossy audioTracks, chapters, dimensions, hdr, metadata, subtitleTracks, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). Text subtitles are kept as plain timed text (styling is lost); image-based subtitles are dropped. HDR10 and HLG are kept with HEVC, AV1, VP9 or ProRes; other codecs get a tone-mapped SDR picture. Dolby Vision metadata is not carried. Only the first video track is kept; data and attachment streams are dropped. VideoPaths.conversion —
mpeg mp3 video.extract-audio video supported partial metadata Only the sound of one audio track is kept; the picture is not. The track is copied bit for bit when it already uses this codec; otherwise it is re-encoded. When re-encoded, the audio goes through a lossy codec again. Stereo at most: surround sound is mixed down. VideoPaths.extractAudio —
mpeg mp4 — video supported lossy audioTracks, chapters, dimensions, hdr, metadata, subtitleTracks, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). Text subtitles are kept as plain timed text (styling is lost); image-based subtitles are dropped. HDR10 and HLG are kept with HEVC, AV1, VP9 or ProRes; other codecs get a tone-mapped SDR picture. Dolby Vision metadata is not carried. Only the first video track is kept; data and attachment streams are dropped. VideoPaths.conversion —
mpeg mp4 video.remux video supported lossless audioTracks, chapters, colorProfile, dimensions, hdr, metadata, subtitleTracks, timing Works only when the codecs inside the file are allowed in MP4; otherwise it stops and says which stream is in the way. With trimming, the result starts on the keyframe at or before the requested start (not frame-accurate). Text subtitles are rewritten in the container's own text format; image-based subtitles are dropped. VideoPaths.remuxIsLossless —
mpeg mpeg — video supported lossy audioTracks, dimensions, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). MPEG Program Stream carries no descriptive tags here. Chapters are dropped. Subtitle tracks are dropped: MPEG Program Stream cannot carry them here. HDR video is tone-mapped to standard dynamic range; 10-bit pictures become 8-bit. Only the first video track is kept; data and attachment streams are dropped. MPEG-2 (or MPEG-1) video at standard frame rates only, with MP2 or AC-3 sound. VideoPaths.conversion —
mpeg mts — video supported lossy audioTracks, dimensions, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). AVCHD carries no descriptive tags here. Chapters are dropped. Subtitle tracks are dropped: AVCHD cannot carry them here. HDR video is tone-mapped to standard dynamic range; 10-bit pictures become 8-bit. Only the first video track is kept; data and attachment streams are dropped. An AVCHD-style transport stream (H.264 with AC-3 or AAC), not a full AVCHD folder structure. VideoPaths.conversion —
mpeg ogg video.extract-audio video supported partial metadata Only the sound of one audio track is kept; the picture is not. The track is copied bit for bit when it already uses this codec; otherwise it is re-encoded. When re-encoded, the audio goes through a lossy codec again. VideoPaths.extractAudio —
mpeg opus video.extract-audio video supported partial metadata Only the sound of one audio track is kept; the picture is not. The track is copied bit for bit when it already uses this codec; otherwise it is re-encoded. When re-encoded, the audio goes through a lossy codec again. VideoPaths.extractAudio —
mpeg ts — video supported lossy audioTracks, dimensions, hdr, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). MPEG Transport Stream carries no descriptive tags here. Chapters are dropped. Subtitle tracks are dropped: MPEG Transport Stream cannot carry them here. HDR10 and HLG are kept with HEVC, AV1, VP9 or ProRes; other codecs get a tone-mapped SDR picture. Dolby Vision metadata is not carried. Only the first video track is kept; data and attachment streams are dropped. VideoPaths.conversion —
mpeg ts video.remux video supported lossless audioTracks, colorProfile, dimensions, hdr, timing Works only when the codecs inside the file are allowed in MPEG Transport Stream; otherwise it stops and says which stream is in the way. With trimming, the result starts on the keyframe at or before the requested start (not frame-accurate). Subtitle tracks are dropped: MPEG Transport Stream cannot carry them here. Chapters are dropped. VideoPaths.remuxIsLossless —
mpeg wav video.extract-audio video supported partial metadata Only the sound of one audio track is kept; the picture is not. VideoPaths.extractAudio —
mpeg webm — video supported lossy audioTracks, chapters, dimensions, hdr, metadata, subtitleTracks, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). Text subtitles are kept as plain timed text (styling is lost); image-based subtitles are dropped. HDR10 and HLG are kept with HEVC, AV1, VP9 or ProRes; other codecs get a tone-mapped SDR picture. Dolby Vision metadata is not carried. Only the first video track is kept; data and attachment streams are dropped. VideoPaths.conversion —
mpeg wmv — video supported lossy audioTracks, dimensions, metadata, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). Chapters are dropped. Subtitle tracks are dropped: Windows Media Video cannot carry them here. HDR video is tone-mapped to standard dynamic range; 10-bit pictures become 8-bit. Only the first video track is kept; data and attachment streams are dropped. Written as WMV8 video with WMA2 audio (stereo), the Windows Media codecs this build can encode — not WMV9/VC-1. VideoPaths.conversion —
mts aac video.extract-audio video supported partial — Only the sound of one audio track is kept; the picture is not. The track is copied bit for bit when it already uses this codec; otherwise it is re-encoded. When re-encoded, the audio goes through a lossy codec again. VideoPaths.extractAudio —
mts aiff video.extract-audio video supported partial metadata Only the sound of one audio track is kept; the picture is not. VideoPaths.extractAudio —
mts apng — video supported lossy animation No sound. Full-colour frames without inter-frame video compression: files are large. The frame rate and the size are reduced (12 fps and 480 px wide unless changed). HDR video is tone-mapped to SDR. VideoPaths.animatedImage —
mts avi — video supported lossy audioTracks, dimensions, metadata, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). Chapters are dropped. Subtitle tracks are dropped: AVI cannot carry them here. HDR video is tone-mapped to standard dynamic range; 10-bit pictures become 8-bit. Only the first video track is kept; data and attachment streams are dropped. MPEG-4 Part 2 video; AVI cannot hold HEVC or AV1 here. VideoPaths.conversion —
mts flac video.extract-audio video supported partial metadata Only the sound of one audio track is kept; the picture is not. The track is copied bit for bit when it already uses this codec; otherwise it is re-encoded. VideoPaths.extractAudio —
mts flv — video supported lossy audioTracks, dimensions, metadata, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). Chapters are dropped. Subtitle tracks are dropped: Flash Video cannot carry them here. HDR video is tone-mapped to standard dynamic range; 10-bit pictures become 8-bit. Only the first video track is kept; data and attachment streams are dropped. One audio track only. VideoPaths.conversion —
mts flv video.remux video supported lossless audioTracks, colorProfile, dimensions, hdr, metadata, timing Works only when the codecs inside the file are allowed in Flash Video; otherwise it stops and says which stream is in the way. With trimming, the result starts on the keyframe at or before the requested start (not frame-accurate). Subtitle tracks are dropped: Flash Video cannot carry them here. Chapters are dropped. VideoPaths.remuxIsLossless —
mts gif — video supported lossy animation No sound. 256 colours per frame, from a palette computed for this video, with dithering. The frame rate and the size are reduced (12 fps and 480 px wide unless changed). HDR video is tone-mapped to SDR. VideoPaths.animatedImage —
mts m4a video.extract-audio video supported partial metadata Only the sound of one audio track is kept; the picture is not. The track is copied bit for bit when it already uses this codec; otherwise it is re-encoded. When re-encoded, the audio goes through a lossy codec again. VideoPaths.extractAudio —
mts m4v — video supported lossy audioTracks, chapters, dimensions, hdr, metadata, subtitleTracks, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). Text subtitles are kept as plain timed text (styling is lost); image-based subtitles are dropped. HDR10 and HLG are kept with HEVC, AV1, VP9 or ProRes; other codecs get a tone-mapped SDR picture. Dolby Vision metadata is not carried. Only the first video track is kept; data and attachment streams are dropped. VideoPaths.conversion —
mts m4v video.remux video supported lossless audioTracks, chapters, colorProfile, dimensions, hdr, metadata, subtitleTracks, timing Works only when the codecs inside the file are allowed in M4V; otherwise it stops and says which stream is in the way. With trimming, the result starts on the keyframe at or before the requested start (not frame-accurate). Text subtitles are rewritten in the container's own text format; image-based subtitles are dropped. VideoPaths.remuxIsLossless —
mts mkv — video supported lossy audioTracks, chapters, dimensions, hdr, metadata, subtitleTracks, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). HDR10 and HLG are kept with HEVC, AV1, VP9 or ProRes; other codecs get a tone-mapped SDR picture. Dolby Vision metadata is not carried. Only the first video track is kept; data and attachment streams are dropped. VideoPaths.conversion —
mts mkv video.remux video supported lossless audioTracks, chapters, colorProfile, dimensions, hdr, metadata, subtitleTracks, timing Works only when the codecs inside the file are allowed in Matroska; otherwise it stops and says which stream is in the way. With trimming, the result starts on the keyframe at or before the requested start (not frame-accurate). VideoPaths.remuxIsLossless —
mts mov — video supported lossy audioTracks, chapters, dimensions, hdr, metadata, subtitleTracks, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). Text subtitles are kept as plain timed text (styling is lost); image-based subtitles are dropped. HDR10 and HLG are kept with HEVC, AV1, VP9 or ProRes; other codecs get a tone-mapped SDR picture. Dolby Vision metadata is not carried. Only the first video track is kept; data and attachment streams are dropped. VideoPaths.conversion —
mts mov video.remux video supported lossless audioTracks, chapters, colorProfile, dimensions, hdr, metadata, subtitleTracks, timing Works only when the codecs inside the file are allowed in QuickTime Movie; otherwise it stops and says which stream is in the way. With trimming, the result starts on the keyframe at or before the requested start (not frame-accurate). Text subtitles are rewritten in the container's own text format; image-based subtitles are dropped. VideoPaths.remuxIsLossless —
mts mp3 video.extract-audio video supported partial metadata Only the sound of one audio track is kept; the picture is not. The track is copied bit for bit when it already uses this codec; otherwise it is re-encoded. When re-encoded, the audio goes through a lossy codec again. Stereo at most: surround sound is mixed down. VideoPaths.extractAudio —
mts mp4 — video supported lossy audioTracks, chapters, dimensions, hdr, metadata, subtitleTracks, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). Text subtitles are kept as plain timed text (styling is lost); image-based subtitles are dropped. HDR10 and HLG are kept with HEVC, AV1, VP9 or ProRes; other codecs get a tone-mapped SDR picture. Dolby Vision metadata is not carried. Only the first video track is kept; data and attachment streams are dropped. VideoPaths.conversion —
mts mp4 video.remux video supported lossless audioTracks, chapters, colorProfile, dimensions, hdr, metadata, subtitleTracks, timing Works only when the codecs inside the file are allowed in MP4; otherwise it stops and says which stream is in the way. With trimming, the result starts on the keyframe at or before the requested start (not frame-accurate). Text subtitles are rewritten in the container's own text format; image-based subtitles are dropped. VideoPaths.remuxIsLossless —
mts mpeg — video supported lossy audioTracks, dimensions, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). MPEG Program Stream carries no descriptive tags here. Chapters are dropped. Subtitle tracks are dropped: MPEG Program Stream cannot carry them here. HDR video is tone-mapped to standard dynamic range; 10-bit pictures become 8-bit. Only the first video track is kept; data and attachment streams are dropped. MPEG-2 (or MPEG-1) video at standard frame rates only, with MP2 or AC-3 sound. VideoPaths.conversion —
mts mts — video supported lossy audioTracks, dimensions, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). AVCHD carries no descriptive tags here. Chapters are dropped. Subtitle tracks are dropped: AVCHD cannot carry them here. HDR video is tone-mapped to standard dynamic range; 10-bit pictures become 8-bit. Only the first video track is kept; data and attachment streams are dropped. An AVCHD-style transport stream (H.264 with AC-3 or AAC), not a full AVCHD folder structure. VideoPaths.conversion —
mts ogg video.extract-audio video supported partial metadata Only the sound of one audio track is kept; the picture is not. The track is copied bit for bit when it already uses this codec; otherwise it is re-encoded. When re-encoded, the audio goes through a lossy codec again. VideoPaths.extractAudio —
mts opus video.extract-audio video supported partial metadata Only the sound of one audio track is kept; the picture is not. The track is copied bit for bit when it already uses this codec; otherwise it is re-encoded. When re-encoded, the audio goes through a lossy codec again. VideoPaths.extractAudio —
mts ts — video supported lossy audioTracks, dimensions, hdr, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). MPEG Transport Stream carries no descriptive tags here. Chapters are dropped. Subtitle tracks are dropped: MPEG Transport Stream cannot carry them here. HDR10 and HLG are kept with HEVC, AV1, VP9 or ProRes; other codecs get a tone-mapped SDR picture. Dolby Vision metadata is not carried. Only the first video track is kept; data and attachment streams are dropped. VideoPaths.conversion —
mts ts video.remux video supported lossless audioTracks, colorProfile, dimensions, hdr, timing Works only when the codecs inside the file are allowed in MPEG Transport Stream; otherwise it stops and says which stream is in the way. With trimming, the result starts on the keyframe at or before the requested start (not frame-accurate). Subtitle tracks are dropped: MPEG Transport Stream cannot carry them here. Chapters are dropped. VideoPaths.remuxIsLossless —
mts wav video.extract-audio video supported partial metadata Only the sound of one audio track is kept; the picture is not. VideoPaths.extractAudio —
mts webm — video supported lossy audioTracks, chapters, dimensions, hdr, metadata, subtitleTracks, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). Text subtitles are kept as plain timed text (styling is lost); image-based subtitles are dropped. HDR10 and HLG are kept with HEVC, AV1, VP9 or ProRes; other codecs get a tone-mapped SDR picture. Dolby Vision metadata is not carried. Only the first video track is kept; data and attachment streams are dropped. VideoPaths.conversion —
mts wmv — video supported lossy audioTracks, dimensions, metadata, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). Chapters are dropped. Subtitle tracks are dropped: Windows Media Video cannot carry them here. HDR video is tone-mapped to standard dynamic range; 10-bit pictures become 8-bit. Only the first video track is kept; data and attachment streams are dropped. Written as WMV8 video with WMA2 audio (stereo), the Windows Media codecs this build can encode — not WMV9/VC-1. VideoPaths.conversion —
ts aac video.extract-audio video supported partial — Only the sound of one audio track is kept; the picture is not. The track is copied bit for bit when it already uses this codec; otherwise it is re-encoded. When re-encoded, the audio goes through a lossy codec again. VideoPaths.extractAudio —
ts aiff video.extract-audio video supported partial metadata Only the sound of one audio track is kept; the picture is not. VideoPaths.extractAudio —
ts apng — video supported lossy animation No sound. Full-colour frames without inter-frame video compression: files are large. The frame rate and the size are reduced (12 fps and 480 px wide unless changed). HDR video is tone-mapped to SDR. VideoPaths.animatedImage —
ts avi — video supported lossy audioTracks, dimensions, metadata, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). Chapters are dropped. Subtitle tracks are dropped: AVI cannot carry them here. HDR video is tone-mapped to standard dynamic range; 10-bit pictures become 8-bit. Only the first video track is kept; data and attachment streams are dropped. MPEG-4 Part 2 video; AVI cannot hold HEVC or AV1 here. VideoPaths.conversion —
ts flac video.extract-audio video supported partial metadata Only the sound of one audio track is kept; the picture is not. The track is copied bit for bit when it already uses this codec; otherwise it is re-encoded. VideoPaths.extractAudio —
ts flv — video supported lossy audioTracks, dimensions, metadata, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). Chapters are dropped. Subtitle tracks are dropped: Flash Video cannot carry them here. HDR video is tone-mapped to standard dynamic range; 10-bit pictures become 8-bit. Only the first video track is kept; data and attachment streams are dropped. One audio track only. VideoPaths.conversion —
ts flv video.remux video supported lossless audioTracks, colorProfile, dimensions, hdr, metadata, timing Works only when the codecs inside the file are allowed in Flash Video; otherwise it stops and says which stream is in the way. With trimming, the result starts on the keyframe at or before the requested start (not frame-accurate). Subtitle tracks are dropped: Flash Video cannot carry them here. Chapters are dropped. VideoPaths.remuxIsLossless —
ts gif — video supported lossy animation No sound. 256 colours per frame, from a palette computed for this video, with dithering. The frame rate and the size are reduced (12 fps and 480 px wide unless changed). HDR video is tone-mapped to SDR. VideoPaths.animatedImage —
ts m4a video.extract-audio video supported partial metadata Only the sound of one audio track is kept; the picture is not. The track is copied bit for bit when it already uses this codec; otherwise it is re-encoded. When re-encoded, the audio goes through a lossy codec again. VideoPaths.extractAudio —
ts m4v — video supported lossy audioTracks, chapters, dimensions, hdr, metadata, subtitleTracks, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). Text subtitles are kept as plain timed text (styling is lost); image-based subtitles are dropped. HDR10 and HLG are kept with HEVC, AV1, VP9 or ProRes; other codecs get a tone-mapped SDR picture. Dolby Vision metadata is not carried. Only the first video track is kept; data and attachment streams are dropped. VideoPaths.conversion —
ts m4v video.remux video supported lossless audioTracks, chapters, colorProfile, dimensions, hdr, metadata, subtitleTracks, timing Works only when the codecs inside the file are allowed in M4V; otherwise it stops and says which stream is in the way. With trimming, the result starts on the keyframe at or before the requested start (not frame-accurate). Text subtitles are rewritten in the container's own text format; image-based subtitles are dropped. VideoPaths.remuxIsLossless —
ts mkv — video supported lossy audioTracks, chapters, dimensions, hdr, metadata, subtitleTracks, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). HDR10 and HLG are kept with HEVC, AV1, VP9 or ProRes; other codecs get a tone-mapped SDR picture. Dolby Vision metadata is not carried. Only the first video track is kept; data and attachment streams are dropped. VideoPaths.conversion —
ts mkv video.remux video supported lossless audioTracks, chapters, colorProfile, dimensions, hdr, metadata, subtitleTracks, timing Works only when the codecs inside the file are allowed in Matroska; otherwise it stops and says which stream is in the way. With trimming, the result starts on the keyframe at or before the requested start (not frame-accurate). VideoPaths.remuxIsLossless —
ts mov — video supported lossy audioTracks, chapters, dimensions, hdr, metadata, subtitleTracks, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). Text subtitles are kept as plain timed text (styling is lost); image-based subtitles are dropped. HDR10 and HLG are kept with HEVC, AV1, VP9 or ProRes; other codecs get a tone-mapped SDR picture. Dolby Vision metadata is not carried. Only the first video track is kept; data and attachment streams are dropped. VideoPaths.conversion —
ts mov video.remux video supported lossless audioTracks, chapters, colorProfile, dimensions, hdr, metadata, subtitleTracks, timing Works only when the codecs inside the file are allowed in QuickTime Movie; otherwise it stops and says which stream is in the way. With trimming, the result starts on the keyframe at or before the requested start (not frame-accurate). Text subtitles are rewritten in the container's own text format; image-based subtitles are dropped. VideoPaths.remuxIsLossless —
ts mp3 video.extract-audio video supported partial metadata Only the sound of one audio track is kept; the picture is not. The track is copied bit for bit when it already uses this codec; otherwise it is re-encoded. When re-encoded, the audio goes through a lossy codec again. Stereo at most: surround sound is mixed down. VideoPaths.extractAudio —
ts mp4 — video supported lossy audioTracks, chapters, dimensions, hdr, metadata, subtitleTracks, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). Text subtitles are kept as plain timed text (styling is lost); image-based subtitles are dropped. HDR10 and HLG are kept with HEVC, AV1, VP9 or ProRes; other codecs get a tone-mapped SDR picture. Dolby Vision metadata is not carried. Only the first video track is kept; data and attachment streams are dropped. VideoPaths.conversion —
ts mp4 video.remux video supported lossless audioTracks, chapters, colorProfile, dimensions, hdr, metadata, subtitleTracks, timing Works only when the codecs inside the file are allowed in MP4; otherwise it stops and says which stream is in the way. With trimming, the result starts on the keyframe at or before the requested start (not frame-accurate). Text subtitles are rewritten in the container's own text format; image-based subtitles are dropped. VideoPaths.remuxIsLossless —
ts mpeg — video supported lossy audioTracks, dimensions, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). MPEG Program Stream carries no descriptive tags here. Chapters are dropped. Subtitle tracks are dropped: MPEG Program Stream cannot carry them here. HDR video is tone-mapped to standard dynamic range; 10-bit pictures become 8-bit. Only the first video track is kept; data and attachment streams are dropped. MPEG-2 (or MPEG-1) video at standard frame rates only, with MP2 or AC-3 sound. VideoPaths.conversion —
ts mpeg video.remux video supported lossless audioTracks, colorProfile, dimensions, hdr, timing Works only when the codecs inside the file are allowed in MPEG Program Stream; otherwise it stops and says which stream is in the way. With trimming, the result starts on the keyframe at or before the requested start (not frame-accurate). Subtitle tracks are dropped: MPEG Program Stream cannot carry them here. Chapters are dropped. VideoPaths.remuxIsLossless —
ts mts — video supported lossy audioTracks, dimensions, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). AVCHD carries no descriptive tags here. Chapters are dropped. Subtitle tracks are dropped: AVCHD cannot carry them here. HDR video is tone-mapped to standard dynamic range; 10-bit pictures become 8-bit. Only the first video track is kept; data and attachment streams are dropped. An AVCHD-style transport stream (H.264 with AC-3 or AAC), not a full AVCHD folder structure. VideoPaths.conversion —
ts mts video.remux video supported lossless audioTracks, colorProfile, dimensions, hdr, timing Works only when the codecs inside the file are allowed in AVCHD; otherwise it stops and says which stream is in the way. With trimming, the result starts on the keyframe at or before the requested start (not frame-accurate). Subtitle tracks are dropped: AVCHD cannot carry them here. Chapters are dropped. VideoPaths.remuxIsLossless —
ts ogg video.extract-audio video supported partial metadata Only the sound of one audio track is kept; the picture is not. The track is copied bit for bit when it already uses this codec; otherwise it is re-encoded. When re-encoded, the audio goes through a lossy codec again. VideoPaths.extractAudio —
ts opus video.extract-audio video supported partial metadata Only the sound of one audio track is kept; the picture is not. The track is copied bit for bit when it already uses this codec; otherwise it is re-encoded. When re-encoded, the audio goes through a lossy codec again. VideoPaths.extractAudio —
ts ts — video supported lossy audioTracks, dimensions, hdr, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). MPEG Transport Stream carries no descriptive tags here. Chapters are dropped. Subtitle tracks are dropped: MPEG Transport Stream cannot carry them here. HDR10 and HLG are kept with HEVC, AV1, VP9 or ProRes; other codecs get a tone-mapped SDR picture. Dolby Vision metadata is not carried. Only the first video track is kept; data and attachment streams are dropped. VideoPaths.conversion —
ts wav video.extract-audio video supported partial metadata Only the sound of one audio track is kept; the picture is not. VideoPaths.extractAudio —
ts webm — video supported lossy audioTracks, chapters, dimensions, hdr, metadata, subtitleTracks, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). Text subtitles are kept as plain timed text (styling is lost); image-based subtitles are dropped. HDR10 and HLG are kept with HEVC, AV1, VP9 or ProRes; other codecs get a tone-mapped SDR picture. Dolby Vision metadata is not carried. Only the first video track is kept; data and attachment streams are dropped. VideoPaths.conversion —
ts wmv — video supported lossy audioTracks, dimensions, metadata, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). Chapters are dropped. Subtitle tracks are dropped: Windows Media Video cannot carry them here. HDR video is tone-mapped to standard dynamic range; 10-bit pictures become 8-bit. Only the first video track is kept; data and attachment streams are dropped. Written as WMV8 video with WMA2 audio (stereo), the Windows Media codecs this build can encode — not WMV9/VC-1. VideoPaths.conversion —
webm aac video.extract-audio video supported partial — Only the sound of one audio track is kept; the picture is not. The track is copied bit for bit when it already uses this codec; otherwise it is re-encoded. When re-encoded, the audio goes through a lossy codec again. VideoPaths.extractAudio —
webm aiff video.extract-audio video supported partial metadata Only the sound of one audio track is kept; the picture is not. VideoPaths.extractAudio —
webm apng — video supported lossy animation No sound. Full-colour frames without inter-frame video compression: files are large. The frame rate and the size are reduced (12 fps and 480 px wide unless changed). HDR video is tone-mapped to SDR. VideoPaths.animatedImage —
webm ass video.extract-subtitles video supported partial timing One text subtitle track is written; picture and sound are not. Image-based subtitles (DVD, Blu-ray PGS) cannot be converted to text: there is no OCR. Plain subtitles get the default ASS style. VideoPaths.extractSubtitles —
webm avi — video supported lossy audioTracks, dimensions, metadata, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). Chapters are dropped. Subtitle tracks are dropped: AVI cannot carry them here. HDR video is tone-mapped to standard dynamic range; 10-bit pictures become 8-bit. Only the first video track is kept; data and attachment streams are dropped. MPEG-4 Part 2 video; AVI cannot hold HEVC or AV1 here. VideoPaths.conversion —
webm flac video.extract-audio video supported partial metadata Only the sound of one audio track is kept; the picture is not. The track is copied bit for bit when it already uses this codec; otherwise it is re-encoded. VideoPaths.extractAudio —
webm flv — video supported lossy audioTracks, dimensions, metadata, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). Chapters are dropped. Subtitle tracks are dropped: Flash Video cannot carry them here. HDR video is tone-mapped to standard dynamic range; 10-bit pictures become 8-bit. Only the first video track is kept; data and attachment streams are dropped. One audio track only. VideoPaths.conversion —
webm gif — video supported lossy animation No sound. 256 colours per frame, from a palette computed for this video, with dithering. The frame rate and the size are reduced (12 fps and 480 px wide unless changed). HDR video is tone-mapped to SDR. VideoPaths.animatedImage —
webm m4a video.extract-audio video supported partial metadata Only the sound of one audio track is kept; the picture is not. The track is copied bit for bit when it already uses this codec; otherwise it is re-encoded. When re-encoded, the audio goes through a lossy codec again. VideoPaths.extractAudio —
webm m4v — video supported lossy audioTracks, chapters, dimensions, hdr, metadata, subtitleTracks, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). Text subtitles are kept as plain timed text (styling is lost); image-based subtitles are dropped. HDR10 and HLG are kept with HEVC, AV1, VP9 or ProRes; other codecs get a tone-mapped SDR picture. Dolby Vision metadata is not carried. Only the first video track is kept; data and attachment streams are dropped. VideoPaths.conversion —
webm mkv — video supported lossy audioTracks, chapters, dimensions, hdr, metadata, subtitleTracks, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). HDR10 and HLG are kept with HEVC, AV1, VP9 or ProRes; other codecs get a tone-mapped SDR picture. Dolby Vision metadata is not carried. Only the first video track is kept; data and attachment streams are dropped. VideoPaths.conversion —
webm mkv video.remux video supported lossless audioTracks, chapters, colorProfile, dimensions, hdr, metadata, subtitleTracks, timing Works only when the codecs inside the file are allowed in Matroska; otherwise it stops and says which stream is in the way. With trimming, the result starts on the keyframe at or before the requested start (not frame-accurate). VideoPaths.remuxIsLossless —
webm mov — video supported lossy audioTracks, chapters, dimensions, hdr, metadata, subtitleTracks, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). Text subtitles are kept as plain timed text (styling is lost); image-based subtitles are dropped. HDR10 and HLG are kept with HEVC, AV1, VP9 or ProRes; other codecs get a tone-mapped SDR picture. Dolby Vision metadata is not carried. Only the first video track is kept; data and attachment streams are dropped. VideoPaths.conversion —
webm mp3 video.extract-audio video supported partial metadata Only the sound of one audio track is kept; the picture is not. The track is copied bit for bit when it already uses this codec; otherwise it is re-encoded. When re-encoded, the audio goes through a lossy codec again. Stereo at most: surround sound is mixed down. VideoPaths.extractAudio —
webm mp4 — video supported lossy audioTracks, chapters, dimensions, hdr, metadata, subtitleTracks, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). Text subtitles are kept as plain timed text (styling is lost); image-based subtitles are dropped. HDR10 and HLG are kept with HEVC, AV1, VP9 or ProRes; other codecs get a tone-mapped SDR picture. Dolby Vision metadata is not carried. Only the first video track is kept; data and attachment streams are dropped. VideoPaths.conversion —
webm mp4 video.remux video supported lossless audioTracks, chapters, colorProfile, dimensions, hdr, metadata, subtitleTracks, timing Works only when the codecs inside the file are allowed in MP4; otherwise it stops and says which stream is in the way. With trimming, the result starts on the keyframe at or before the requested start (not frame-accurate). Text subtitles are rewritten in the container's own text format; image-based subtitles are dropped. VideoPaths.remuxIsLossless —
webm mpeg — video supported lossy audioTracks, dimensions, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). MPEG Program Stream carries no descriptive tags here. Chapters are dropped. Subtitle tracks are dropped: MPEG Program Stream cannot carry them here. HDR video is tone-mapped to standard dynamic range; 10-bit pictures become 8-bit. Only the first video track is kept; data and attachment streams are dropped. MPEG-2 (or MPEG-1) video at standard frame rates only, with MP2 or AC-3 sound. VideoPaths.conversion —
webm mts — video supported lossy audioTracks, dimensions, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). AVCHD carries no descriptive tags here. Chapters are dropped. Subtitle tracks are dropped: AVCHD cannot carry them here. HDR video is tone-mapped to standard dynamic range; 10-bit pictures become 8-bit. Only the first video track is kept; data and attachment streams are dropped. An AVCHD-style transport stream (H.264 with AC-3 or AAC), not a full AVCHD folder structure. VideoPaths.conversion —
webm ogg video.extract-audio video supported partial metadata Only the sound of one audio track is kept; the picture is not. The track is copied bit for bit when it already uses this codec; otherwise it is re-encoded. When re-encoded, the audio goes through a lossy codec again. VideoPaths.extractAudio —
webm opus video.extract-audio video supported partial metadata Only the sound of one audio track is kept; the picture is not. The track is copied bit for bit when it already uses this codec; otherwise it is re-encoded. When re-encoded, the audio goes through a lossy codec again. VideoPaths.extractAudio —
webm srt video.extract-subtitles video supported partial timing One text subtitle track is written; picture and sound are not. Image-based subtitles (DVD, Blu-ray PGS) cannot be converted to text: there is no OCR. Styling of ASS/SSA subtitles (fonts, colours, positions) is lost. VideoPaths.extractSubtitles —
webm ts — video supported lossy audioTracks, dimensions, hdr, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). MPEG Transport Stream carries no descriptive tags here. Chapters are dropped. Subtitle tracks are dropped: MPEG Transport Stream cannot carry them here. HDR10 and HLG are kept with HEVC, AV1, VP9 or ProRes; other codecs get a tone-mapped SDR picture. Dolby Vision metadata is not carried. Only the first video track is kept; data and attachment streams are dropped. VideoPaths.conversion —
webm vtt video.extract-subtitles video supported partial timing One text subtitle track is written; picture and sound are not. Image-based subtitles (DVD, Blu-ray PGS) cannot be converted to text: there is no OCR. Styling of ASS/SSA subtitles (fonts, colours, positions) is lost. VideoPaths.extractSubtitles —
webm wav video.extract-audio video supported partial metadata Only the sound of one audio track is kept; the picture is not. VideoPaths.extractAudio —
webm webm — video supported lossy audioTracks, chapters, dimensions, hdr, metadata, subtitleTracks, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). Text subtitles are kept as plain timed text (styling is lost); image-based subtitles are dropped. HDR10 and HLG are kept with HEVC, AV1, VP9 or ProRes; other codecs get a tone-mapped SDR picture. Dolby Vision metadata is not carried. Only the first video track is kept; data and attachment streams are dropped. VideoPaths.conversion —
webm wmv — video supported lossy audioTracks, dimensions, metadata, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). Chapters are dropped. Subtitle tracks are dropped: Windows Media Video cannot carry them here. HDR video is tone-mapped to standard dynamic range; 10-bit pictures become 8-bit. Only the first video track is kept; data and attachment streams are dropped. Written as WMV8 video with WMA2 audio (stereo), the Windows Media codecs this build can encode — not WMV9/VC-1. VideoPaths.conversion —
wmv aac video.extract-audio video supported partial — Only the sound of one audio track is kept; the picture is not. The track is copied bit for bit when it already uses this codec; otherwise it is re-encoded. When re-encoded, the audio goes through a lossy codec again. VideoPaths.extractAudio —
wmv aiff video.extract-audio video supported partial metadata Only the sound of one audio track is kept; the picture is not. VideoPaths.extractAudio —
wmv apng — video supported lossy animation No sound. Full-colour frames without inter-frame video compression: files are large. The frame rate and the size are reduced (12 fps and 480 px wide unless changed). HDR video is tone-mapped to SDR. VideoPaths.animatedImage —
wmv avi — video supported lossy audioTracks, dimensions, metadata, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). Chapters are dropped. Subtitle tracks are dropped: AVI cannot carry them here. HDR video is tone-mapped to standard dynamic range; 10-bit pictures become 8-bit. Only the first video track is kept; data and attachment streams are dropped. MPEG-4 Part 2 video; AVI cannot hold HEVC or AV1 here. VideoPaths.conversion —
wmv flac video.extract-audio video supported partial metadata Only the sound of one audio track is kept; the picture is not. The track is copied bit for bit when it already uses this codec; otherwise it is re-encoded. VideoPaths.extractAudio —
wmv flv — video supported lossy audioTracks, dimensions, metadata, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). Chapters are dropped. Subtitle tracks are dropped: Flash Video cannot carry them here. HDR video is tone-mapped to standard dynamic range; 10-bit pictures become 8-bit. Only the first video track is kept; data and attachment streams are dropped. One audio track only. VideoPaths.conversion —
wmv gif — video supported lossy animation No sound. 256 colours per frame, from a palette computed for this video, with dithering. The frame rate and the size are reduced (12 fps and 480 px wide unless changed). HDR video is tone-mapped to SDR. VideoPaths.animatedImage —
wmv m4a video.extract-audio video supported partial metadata Only the sound of one audio track is kept; the picture is not. The track is copied bit for bit when it already uses this codec; otherwise it is re-encoded. When re-encoded, the audio goes through a lossy codec again. VideoPaths.extractAudio —
wmv m4v — video supported lossy audioTracks, chapters, dimensions, hdr, metadata, subtitleTracks, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). Text subtitles are kept as plain timed text (styling is lost); image-based subtitles are dropped. HDR10 and HLG are kept with HEVC, AV1, VP9 or ProRes; other codecs get a tone-mapped SDR picture. Dolby Vision metadata is not carried. Only the first video track is kept; data and attachment streams are dropped. VideoPaths.conversion —
wmv mkv — video supported lossy audioTracks, chapters, dimensions, hdr, metadata, subtitleTracks, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). HDR10 and HLG are kept with HEVC, AV1, VP9 or ProRes; other codecs get a tone-mapped SDR picture. Dolby Vision metadata is not carried. Only the first video track is kept; data and attachment streams are dropped. VideoPaths.conversion —
wmv mkv video.remux video supported lossless audioTracks, chapters, colorProfile, dimensions, hdr, metadata, subtitleTracks, timing Works only when the codecs inside the file are allowed in Matroska; otherwise it stops and says which stream is in the way. With trimming, the result starts on the keyframe at or before the requested start (not frame-accurate). VideoPaths.remuxIsLossless —
wmv mov — video supported lossy audioTracks, chapters, dimensions, hdr, metadata, subtitleTracks, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). Text subtitles are kept as plain timed text (styling is lost); image-based subtitles are dropped. HDR10 and HLG are kept with HEVC, AV1, VP9 or ProRes; other codecs get a tone-mapped SDR picture. Dolby Vision metadata is not carried. Only the first video track is kept; data and attachment streams are dropped. VideoPaths.conversion —
wmv mp3 video.extract-audio video supported partial metadata Only the sound of one audio track is kept; the picture is not. The track is copied bit for bit when it already uses this codec; otherwise it is re-encoded. When re-encoded, the audio goes through a lossy codec again. Stereo at most: surround sound is mixed down. VideoPaths.extractAudio —
wmv mp4 — video supported lossy audioTracks, chapters, dimensions, hdr, metadata, subtitleTracks, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). Text subtitles are kept as plain timed text (styling is lost); image-based subtitles are dropped. HDR10 and HLG are kept with HEVC, AV1, VP9 or ProRes; other codecs get a tone-mapped SDR picture. Dolby Vision metadata is not carried. Only the first video track is kept; data and attachment streams are dropped. VideoPaths.conversion —
wmv mpeg — video supported lossy audioTracks, dimensions, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). MPEG Program Stream carries no descriptive tags here. Chapters are dropped. Subtitle tracks are dropped: MPEG Program Stream cannot carry them here. HDR video is tone-mapped to standard dynamic range; 10-bit pictures become 8-bit. Only the first video track is kept; data and attachment streams are dropped. MPEG-2 (or MPEG-1) video at standard frame rates only, with MP2 or AC-3 sound. VideoPaths.conversion —
wmv mts — video supported lossy audioTracks, dimensions, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). AVCHD carries no descriptive tags here. Chapters are dropped. Subtitle tracks are dropped: AVCHD cannot carry them here. HDR video is tone-mapped to standard dynamic range; 10-bit pictures become 8-bit. Only the first video track is kept; data and attachment streams are dropped. An AVCHD-style transport stream (H.264 with AC-3 or AAC), not a full AVCHD folder structure. VideoPaths.conversion —
wmv ogg video.extract-audio video supported partial metadata Only the sound of one audio track is kept; the picture is not. The track is copied bit for bit when it already uses this codec; otherwise it is re-encoded. When re-encoded, the audio goes through a lossy codec again. VideoPaths.extractAudio —
wmv opus video.extract-audio video supported partial metadata Only the sound of one audio track is kept; the picture is not. The track is copied bit for bit when it already uses this codec; otherwise it is re-encoded. When re-encoded, the audio goes through a lossy codec again. VideoPaths.extractAudio —
wmv ts — video supported lossy audioTracks, dimensions, hdr, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). MPEG Transport Stream carries no descriptive tags here. Chapters are dropped. Subtitle tracks are dropped: MPEG Transport Stream cannot carry them here. HDR10 and HLG are kept with HEVC, AV1, VP9 or ProRes; other codecs get a tone-mapped SDR picture. Dolby Vision metadata is not carried. Only the first video track is kept; data and attachment streams are dropped. VideoPaths.conversion —
wmv wav video.extract-audio video supported partial metadata Only the sound of one audio track is kept; the picture is not. VideoPaths.extractAudio —
wmv webm — video supported lossy audioTracks, chapters, dimensions, hdr, metadata, subtitleTracks, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). Text subtitles are kept as plain timed text (styling is lost); image-based subtitles are dropped. HDR10 and HLG are kept with HEVC, AV1, VP9 or ProRes; other codecs get a tone-mapped SDR picture. Dolby Vision metadata is not carried. Only the first video track is kept; data and attachment streams are dropped. VideoPaths.conversion —
wmv wmv — video supported lossy audioTracks, dimensions, metadata, timing The picture is re-encoded: some detail is lost (how much depends on the quality preset). Chapters are dropped. Subtitle tracks are dropped: Windows Media Video cannot carry them here. HDR video is tone-mapped to standard dynamic range; 10-bit pictures become 8-bit. Only the first video track is kept; data and attachment streams are dropped. Written as WMV8 video with WMA2 audio (stereo), the Windows Media codecs this build can encode — not WMV9/VC-1. VideoPaths.conversion —

# audio sources (132 paths)

Source Destination Operation Engine Status Fidelity Preserves Limitations Tests Performance
aac aac — audio supported lossy — Both formats are lossy: re-encoding loses a little more quality (generation loss). AAC (ADTS) carries no tags here: title, artist and the other tags are dropped. Album artwork is not carried to AAC (ADTS). AudioPaths.conversion —
aac aiff — audio supported lossy metadata The source is lossy: the file gets bigger, and quality lost earlier is not restored. AudioPaths.conversion —
aac alac — audio supported lossy metadata The source is lossy: the file gets bigger, and quality lost earlier is not restored. AudioPaths.conversion —
aac caf — audio supported lossy metadata The source is lossy: the file gets bigger, and quality lost earlier is not restored. Album artwork is not carried to Core Audio Format. AudioPaths.conversion —
aac flac — audio supported lossy metadata The source is lossy: the file gets bigger, and quality lost earlier is not restored. AudioPaths.conversion —
aac m4a — audio supported lossy metadata Both formats are lossy: re-encoding loses a little more quality (generation loss). AudioPaths.conversion —
aac mp3 — audio supported lossy metadata Both formats are lossy: re-encoding loses a little more quality (generation loss). Stereo at most: surround sound is mixed down. AudioPaths.conversion —
aac ogg — audio supported lossy metadata Both formats are lossy: re-encoding loses a little more quality (generation loss). Album artwork is not carried to Ogg Vorbis. AudioPaths.conversion —
aac opus — audio supported lossy metadata Both formats are lossy: re-encoding loses a little more quality (generation loss). Album artwork is not carried to Opus. Opus always works at 48 kHz. AudioPaths.conversion —
aac wav — audio supported lossy metadata The source is lossy: the file gets bigger, and quality lost earlier is not restored. Album artwork is not carried to WAV. WAV stores only a few tags (title, artist, album, date, genre, comment, track). AudioPaths.conversion —
aac wma — audio supported lossy metadata Both formats are lossy: re-encoding loses a little more quality (generation loss). Album artwork is not carried to Windows Media Audio. Stereo at most: surround sound is mixed down. AudioPaths.conversion —
aiff aac — audio supported lossy — Lossy encoding: some audio detail is discarded. AAC (ADTS) carries no tags here: title, artist and the other tags are dropped. Album artwork is not carried to AAC (ADTS). AudioPaths.conversion —
aiff aiff — audio supported lossless metadata Sample-exact when the sample rate, bit depth and channels are kept and loudness normalisation is off. AudioPaths.conversion, losslessPathsAreSampleExact —
aiff alac — audio supported lossless metadata Sample-exact when the sample rate, bit depth and channels are kept and loudness normalisation is off. 32-bit and floating-point sources are stored as 24-bit. AudioPaths.conversion, losslessPathsAreSampleExact —
aiff caf — audio supported lossless metadata Sample-exact when the sample rate, bit depth and channels are kept and loudness normalisation is off. Album artwork is not carried to Core Audio Format. AudioPaths.conversion, losslessPathsAreSampleExact —
aiff flac — audio supported lossless metadata Sample-exact when the sample rate, bit depth and channels are kept and loudness normalisation is off. 32-bit and floating-point sources are stored as 24-bit. AudioPaths.conversion, losslessPathsAreSampleExact —
aiff m4a — audio supported lossy metadata Lossy encoding: some audio detail is discarded. AudioPaths.conversion —
aiff mp3 — audio supported lossy metadata Lossy encoding: some audio detail is discarded. Stereo at most: surround sound is mixed down. AudioPaths.conversion —
aiff ogg — audio supported lossy metadata Lossy encoding: some audio detail is discarded. Album artwork is not carried to Ogg Vorbis. AudioPaths.conversion —
aiff opus — audio supported lossy metadata Lossy encoding: some audio detail is discarded. Album artwork is not carried to Opus. Opus always works at 48 kHz. AudioPaths.conversion —
aiff wav — audio supported lossless metadata Sample-exact when the sample rate, bit depth and channels are kept and loudness normalisation is off. Album artwork is not carried to WAV. WAV stores only a few tags (title, artist, album, date, genre, comment, track). AudioPaths.conversion, losslessPathsAreSampleExact —
aiff wma — audio supported lossy metadata Lossy encoding: some audio detail is discarded. Album artwork is not carried to Windows Media Audio. Stereo at most: surround sound is mixed down. AudioPaths.conversion —
alac aac — audio supported lossy — Lossy encoding: some audio detail is discarded. AAC (ADTS) carries no tags here: title, artist and the other tags are dropped. Album artwork is not carried to AAC (ADTS). AudioPaths.conversion —
alac aiff — audio supported lossless metadata Sample-exact when the sample rate, bit depth and channels are kept and loudness normalisation is off. AudioPaths.conversion, losslessPathsAreSampleExact —
alac alac — audio supported lossless metadata Sample-exact when the sample rate, bit depth and channels are kept and loudness normalisation is off. 32-bit and floating-point sources are stored as 24-bit. AudioPaths.conversion, losslessPathsAreSampleExact —
alac caf — audio supported lossless metadata Sample-exact when the sample rate, bit depth and channels are kept and loudness normalisation is off. Album artwork is not carried to Core Audio Format. AudioPaths.conversion, losslessPathsAreSampleExact —
alac flac — audio supported lossless metadata Sample-exact when the sample rate, bit depth and channels are kept and loudness normalisation is off. 32-bit and floating-point sources are stored as 24-bit. AudioPaths.conversion, losslessPathsAreSampleExact —
alac m4a — audio supported lossy metadata Lossy encoding: some audio detail is discarded. AudioPaths.conversion —
alac mp3 — audio supported lossy metadata Lossy encoding: some audio detail is discarded. Stereo at most: surround sound is mixed down. AudioPaths.conversion —
alac ogg — audio supported lossy metadata Lossy encoding: some audio detail is discarded. Album artwork is not carried to Ogg Vorbis. AudioPaths.conversion —
alac opus — audio supported lossy metadata Lossy encoding: some audio detail is discarded. Album artwork is not carried to Opus. Opus always works at 48 kHz. AudioPaths.conversion —
alac wav — audio supported lossless metadata Sample-exact when the sample rate, bit depth and channels are kept and loudness normalisation is off. Album artwork is not carried to WAV. WAV stores only a few tags (title, artist, album, date, genre, comment, track). AudioPaths.conversion, losslessPathsAreSampleExact —
alac wma — audio supported lossy metadata Lossy encoding: some audio detail is discarded. Album artwork is not carried to Windows Media Audio. Stereo at most: surround sound is mixed down. AudioPaths.conversion —
caf aac — audio supported lossy — Lossy encoding: some audio detail is discarded. AAC (ADTS) carries no tags here: title, artist and the other tags are dropped. Album artwork is not carried to AAC (ADTS). AudioPaths.conversion —
caf aiff — audio supported lossless metadata Sample-exact when the sample rate, bit depth and channels are kept and loudness normalisation is off. A CAF file that holds lossy audio (AAC…) is decoded first; the result is then not lossless with respect to the original recording. AudioPaths.conversion, losslessPathsAreSampleExact —
caf alac — audio supported lossless metadata Sample-exact when the sample rate, bit depth and channels are kept and loudness normalisation is off. 32-bit and floating-point sources are stored as 24-bit. A CAF file that holds lossy audio (AAC…) is decoded first; the result is then not lossless with respect to the original recording. AudioPaths.conversion, losslessPathsAreSampleExact —
caf caf — audio supported lossless metadata Sample-exact when the sample rate, bit depth and channels are kept and loudness normalisation is off. A CAF file that holds lossy audio (AAC…) is decoded first; the result is then not lossless with respect to the original recording. Album artwork is not carried to Core Audio Format. AudioPaths.conversion, losslessPathsAreSampleExact —
caf flac — audio supported lossless metadata Sample-exact when the sample rate, bit depth and channels are kept and loudness normalisation is off. 32-bit and floating-point sources are stored as 24-bit. A CAF file that holds lossy audio (AAC…) is decoded first; the result is then not lossless with respect to the original recording. AudioPaths.conversion, losslessPathsAreSampleExact —
caf m4a — audio supported lossy metadata Lossy encoding: some audio detail is discarded. AudioPaths.conversion —
caf mp3 — audio supported lossy metadata Lossy encoding: some audio detail is discarded. Stereo at most: surround sound is mixed down. AudioPaths.conversion —
caf ogg — audio supported lossy metadata Lossy encoding: some audio detail is discarded. Album artwork is not carried to Ogg Vorbis. AudioPaths.conversion —
caf opus — audio supported lossy metadata Lossy encoding: some audio detail is discarded. Album artwork is not carried to Opus. Opus always works at 48 kHz. AudioPaths.conversion —
caf wav — audio supported lossless metadata Sample-exact when the sample rate, bit depth and channels are kept and loudness normalisation is off. A CAF file that holds lossy audio (AAC…) is decoded first; the result is then not lossless with respect to the original recording. Album artwork is not carried to WAV. WAV stores only a few tags (title, artist, album, date, genre, comment, track). AudioPaths.conversion, losslessPathsAreSampleExact —
caf wma — audio supported lossy metadata Lossy encoding: some audio detail is discarded. Album artwork is not carried to Windows Media Audio. Stereo at most: surround sound is mixed down. AudioPaths.conversion —
flac aac — audio supported lossy — Lossy encoding: some audio detail is discarded. AAC (ADTS) carries no tags here: title, artist and the other tags are dropped. Album artwork is not carried to AAC (ADTS). AudioPaths.conversion —
flac aiff — audio supported lossless metadata Sample-exact when the sample rate, bit depth and channels are kept and loudness normalisation is off. AudioPaths.conversion, losslessPathsAreSampleExact —
flac alac — audio supported lossless metadata Sample-exact when the sample rate, bit depth and channels are kept and loudness normalisation is off. 32-bit and floating-point sources are stored as 24-bit. AudioPaths.conversion, losslessPathsAreSampleExact —
flac caf — audio supported lossless metadata Sample-exact when the sample rate, bit depth and channels are kept and loudness normalisation is off. Album artwork is not carried to Core Audio Format. AudioPaths.conversion, losslessPathsAreSampleExact —
flac flac — audio supported lossless metadata Sample-exact when the sample rate, bit depth and channels are kept and loudness normalisation is off. 32-bit and floating-point sources are stored as 24-bit. AudioPaths.conversion, losslessPathsAreSampleExact —
flac m4a — audio supported lossy metadata Lossy encoding: some audio detail is discarded. AudioPaths.conversion —
flac mp3 — audio supported lossy metadata Lossy encoding: some audio detail is discarded. Stereo at most: surround sound is mixed down. AudioPaths.conversion —
flac ogg — audio supported lossy metadata Lossy encoding: some audio detail is discarded. Album artwork is not carried to Ogg Vorbis. AudioPaths.conversion —
flac opus — audio supported lossy metadata Lossy encoding: some audio detail is discarded. Album artwork is not carried to Opus. Opus always works at 48 kHz. AudioPaths.conversion —
flac wav — audio supported lossless metadata Sample-exact when the sample rate, bit depth and channels are kept and loudness normalisation is off. Album artwork is not carried to WAV. WAV stores only a few tags (title, artist, album, date, genre, comment, track). AudioPaths.conversion, losslessPathsAreSampleExact —
flac wma — audio supported lossy metadata Lossy encoding: some audio detail is discarded. Album artwork is not carried to Windows Media Audio. Stereo at most: surround sound is mixed down. AudioPaths.conversion —
m4a aac — audio supported lossy — Both formats are lossy: re-encoding loses a little more quality (generation loss). AAC (ADTS) carries no tags here: title, artist and the other tags are dropped. Album artwork is not carried to AAC (ADTS). AudioPaths.conversion —
m4a aiff — audio supported lossy metadata The source is lossy: the file gets bigger, and quality lost earlier is not restored. AudioPaths.conversion —
m4a alac — audio supported lossy metadata The source is lossy: the file gets bigger, and quality lost earlier is not restored. AudioPaths.conversion —
m4a caf — audio supported lossy metadata The source is lossy: the file gets bigger, and quality lost earlier is not restored. Album artwork is not carried to Core Audio Format. AudioPaths.conversion —
m4a flac — audio supported lossy metadata The source is lossy: the file gets bigger, and quality lost earlier is not restored. AudioPaths.conversion —
m4a m4a — audio supported lossy metadata Both formats are lossy: re-encoding loses a little more quality (generation loss). AudioPaths.conversion —
m4a mp3 — audio supported lossy metadata Both formats are lossy: re-encoding loses a little more quality (generation loss). Stereo at most: surround sound is mixed down. AudioPaths.conversion —
m4a ogg — audio supported lossy metadata Both formats are lossy: re-encoding loses a little more quality (generation loss). Album artwork is not carried to Ogg Vorbis. AudioPaths.conversion —
m4a opus — audio supported lossy metadata Both formats are lossy: re-encoding loses a little more quality (generation loss). Album artwork is not carried to Opus. Opus always works at 48 kHz. AudioPaths.conversion —
m4a wav — audio supported lossy metadata The source is lossy: the file gets bigger, and quality lost earlier is not restored. Album artwork is not carried to WAV. WAV stores only a few tags (title, artist, album, date, genre, comment, track). AudioPaths.conversion —
m4a wma — audio supported lossy metadata Both formats are lossy: re-encoding loses a little more quality (generation loss). Album artwork is not carried to Windows Media Audio. Stereo at most: surround sound is mixed down. AudioPaths.conversion —
midi aac — audio supported structural — The notes are played by the macOS General MIDI synthesiser; the sound depends on that sound bank, not on the instruments the file was written for. Lyrics, markers and other MIDI meta events are not carried over. The rendered audio is then encoded with a lossy codec. AudioPaths.midiRendering —
midi aiff — audio supported structural — The notes are played by the macOS General MIDI synthesiser; the sound depends on that sound bank, not on the instruments the file was written for. Lyrics, markers and other MIDI meta events are not carried over. AudioPaths.midiRendering —
midi alac — audio supported structural — The notes are played by the macOS General MIDI synthesiser; the sound depends on that sound bank, not on the instruments the file was written for. Lyrics, markers and other MIDI meta events are not carried over. AudioPaths.midiRendering —
midi caf — audio supported structural — The notes are played by the macOS General MIDI synthesiser; the sound depends on that sound bank, not on the instruments the file was written for. Lyrics, markers and other MIDI meta events are not carried over. AudioPaths.midiRendering —
midi flac — audio supported structural — The notes are played by the macOS General MIDI synthesiser; the sound depends on that sound bank, not on the instruments the file was written for. Lyrics, markers and other MIDI meta events are not carried over. AudioPaths.midiRendering —
midi m4a — audio supported structural — The notes are played by the macOS General MIDI synthesiser; the sound depends on that sound bank, not on the instruments the file was written for. Lyrics, markers and other MIDI meta events are not carried over. The rendered audio is then encoded with a lossy codec. AudioPaths.midiRendering —
midi mp3 — audio supported structural — The notes are played by the macOS General MIDI synthesiser; the sound depends on that sound bank, not on the instruments the file was written for. Lyrics, markers and other MIDI meta events are not carried over. The rendered audio is then encoded with a lossy codec. Stereo at most: surround sound is mixed down. AudioPaths.midiRendering —
midi ogg — audio supported structural — The notes are played by the macOS General MIDI synthesiser; the sound depends on that sound bank, not on the instruments the file was written for. Lyrics, markers and other MIDI meta events are not carried over. The rendered audio is then encoded with a lossy codec. AudioPaths.midiRendering —
midi opus — audio supported structural — The notes are played by the macOS General MIDI synthesiser; the sound depends on that sound bank, not on the instruments the file was written for. Lyrics, markers and other MIDI meta events are not carried over. The rendered audio is then encoded with a lossy codec. Opus always works at 48 kHz. AudioPaths.midiRendering —
midi wav — audio supported structural — The notes are played by the macOS General MIDI synthesiser; the sound depends on that sound bank, not on the instruments the file was written for. Lyrics, markers and other MIDI meta events are not carried over. WAV stores only a few tags (title, artist, album, date, genre, comment, track). AudioPaths.midiRendering —
midi wma — audio supported structural — The notes are played by the macOS General MIDI synthesiser; the sound depends on that sound bank, not on the instruments the file was written for. Lyrics, markers and other MIDI meta events are not carried over. The rendered audio is then encoded with a lossy codec. Stereo at most: surround sound is mixed down. AudioPaths.midiRendering —
mp3 aac — audio supported lossy — Both formats are lossy: re-encoding loses a little more quality (generation loss). AAC (ADTS) carries no tags here: title, artist and the other tags are dropped. Album artwork is not carried to AAC (ADTS). AudioPaths.conversion —
mp3 aiff — audio supported lossy metadata The source is lossy: the file gets bigger, and quality lost earlier is not restored. AudioPaths.conversion —
mp3 alac — audio supported lossy metadata The source is lossy: the file gets bigger, and quality lost earlier is not restored. AudioPaths.conversion —
mp3 caf — audio supported lossy metadata The source is lossy: the file gets bigger, and quality lost earlier is not restored. Album artwork is not carried to Core Audio Format. AudioPaths.conversion —
mp3 flac — audio supported lossy metadata The source is lossy: the file gets bigger, and quality lost earlier is not restored. AudioPaths.conversion —
mp3 m4a — audio supported lossy metadata Both formats are lossy: re-encoding loses a little more quality (generation loss). AudioPaths.conversion —
mp3 mp3 — audio supported lossy metadata Both formats are lossy: re-encoding loses a little more quality (generation loss). Stereo at most: surround sound is mixed down. AudioPaths.conversion —
mp3 ogg — audio supported lossy metadata Both formats are lossy: re-encoding loses a little more quality (generation loss). Album artwork is not carried to Ogg Vorbis. AudioPaths.conversion —
mp3 opus — audio supported lossy metadata Both formats are lossy: re-encoding loses a little more quality (generation loss). Album artwork is not carried to Opus. Opus always works at 48 kHz. AudioPaths.conversion —
mp3 wav — audio supported lossy metadata The source is lossy: the file gets bigger, and quality lost earlier is not restored. Album artwork is not carried to WAV. WAV stores only a few tags (title, artist, album, date, genre, comment, track). AudioPaths.conversion —
mp3 wma — audio supported lossy metadata Both formats are lossy: re-encoding loses a little more quality (generation loss). Album artwork is not carried to Windows Media Audio. Stereo at most: surround sound is mixed down. AudioPaths.conversion —
ogg aac — audio supported lossy — Both formats are lossy: re-encoding loses a little more quality (generation loss). AAC (ADTS) carries no tags here: title, artist and the other tags are dropped. Album artwork is not carried to AAC (ADTS). AudioPaths.conversion —
ogg aiff — audio supported lossy metadata The source is lossy: the file gets bigger, and quality lost earlier is not restored. AudioPaths.conversion —
ogg alac — audio supported lossy metadata The source is lossy: the file gets bigger, and quality lost earlier is not restored. AudioPaths.conversion —
ogg caf — audio supported lossy metadata The source is lossy: the file gets bigger, and quality lost earlier is not restored. Album artwork is not carried to Core Audio Format. AudioPaths.conversion —
ogg flac — audio supported lossy metadata The source is lossy: the file gets bigger, and quality lost earlier is not restored. AudioPaths.conversion —
ogg m4a — audio supported lossy metadata Both formats are lossy: re-encoding loses a little more quality (generation loss). AudioPaths.conversion —
ogg mp3 — audio supported lossy metadata Both formats are lossy: re-encoding loses a little more quality (generation loss). Stereo at most: surround sound is mixed down. AudioPaths.conversion —
ogg ogg — audio supported lossy metadata Both formats are lossy: re-encoding loses a little more quality (generation loss). Album artwork is not carried to Ogg Vorbis. AudioPaths.conversion —
ogg opus — audio supported lossy metadata Both formats are lossy: re-encoding loses a little more quality (generation loss). Album artwork is not carried to Opus. Opus always works at 48 kHz. AudioPaths.conversion —
ogg wav — audio supported lossy metadata The source is lossy: the file gets bigger, and quality lost earlier is not restored. Album artwork is not carried to WAV. WAV stores only a few tags (title, artist, album, date, genre, comment, track). AudioPaths.conversion —
ogg wma — audio supported lossy metadata Both formats are lossy: re-encoding loses a little more quality (generation loss). Album artwork is not carried to Windows Media Audio. Stereo at most: surround sound is mixed down. AudioPaths.conversion —
opus aac — audio supported lossy — Both formats are lossy: re-encoding loses a little more quality (generation loss). AAC (ADTS) carries no tags here: title, artist and the other tags are dropped. Album artwork is not carried to AAC (ADTS). AudioPaths.conversion —
opus aiff — audio supported lossy metadata The source is lossy: the file gets bigger, and quality lost earlier is not restored. AudioPaths.conversion —
opus alac — audio supported lossy metadata The source is lossy: the file gets bigger, and quality lost earlier is not restored. AudioPaths.conversion —
opus caf — audio supported lossy metadata The source is lossy: the file gets bigger, and quality lost earlier is not restored. Album artwork is not carried to Core Audio Format. AudioPaths.conversion —
opus flac — audio supported lossy metadata The source is lossy: the file gets bigger, and quality lost earlier is not restored. AudioPaths.conversion —
opus m4a — audio supported lossy metadata Both formats are lossy: re-encoding loses a little more quality (generation loss). AudioPaths.conversion —
opus mp3 — audio supported lossy metadata Both formats are lossy: re-encoding loses a little more quality (generation loss). Stereo at most: surround sound is mixed down. AudioPaths.conversion —
opus ogg — audio supported lossy metadata Both formats are lossy: re-encoding loses a little more quality (generation loss). Album artwork is not carried to Ogg Vorbis. AudioPaths.conversion —
opus opus — audio supported lossy metadata Both formats are lossy: re-encoding loses a little more quality (generation loss). Album artwork is not carried to Opus. Opus always works at 48 kHz. AudioPaths.conversion —
opus wav — audio supported lossy metadata The source is lossy: the file gets bigger, and quality lost earlier is not restored. Album artwork is not carried to WAV. WAV stores only a few tags (title, artist, album, date, genre, comment, track). AudioPaths.conversion —
opus wma — audio supported lossy metadata Both formats are lossy: re-encoding loses a little more quality (generation loss). Album artwork is not carried to Windows Media Audio. Stereo at most: surround sound is mixed down. AudioPaths.conversion —
wav aac — audio supported lossy — Lossy encoding: some audio detail is discarded. AAC (ADTS) carries no tags here: title, artist and the other tags are dropped. Album artwork is not carried to AAC (ADTS). AudioPaths.conversion 1.39 (engine wall, run 1 (s); WAV → AAC 160 kb/s (M4A))
wav aiff — audio supported lossless metadata Sample-exact when the sample rate, bit depth and channels are kept and loudness normalisation is off. AudioPaths.conversion, losslessPathsAreSampleExact —
wav alac — audio supported lossless metadata Sample-exact when the sample rate, bit depth and channels are kept and loudness normalisation is off. 32-bit and floating-point sources are stored as 24-bit. AudioPaths.conversion, losslessPathsAreSampleExact —
wav caf — audio supported lossless metadata Sample-exact when the sample rate, bit depth and channels are kept and loudness normalisation is off. Album artwork is not carried to Core Audio Format. AudioPaths.conversion, losslessPathsAreSampleExact —
wav flac — audio supported lossless metadata Sample-exact when the sample rate, bit depth and channels are kept and loudness normalisation is off. 32-bit and floating-point sources are stored as 24-bit. AudioPaths.conversion, losslessPathsAreSampleExact 0.32 (engine wall, run 1 (s); WAV → FLAC level 5)
wav m4a — audio supported lossy metadata Lossy encoding: some audio detail is discarded. AudioPaths.conversion —
wav mp3 — audio supported lossy metadata Lossy encoding: some audio detail is discarded. Stereo at most: surround sound is mixed down. AudioPaths.conversion 1.65 (engine wall, run 1 (s); WAV → MP3 192 kb/s)
wav ogg — audio supported lossy metadata Lossy encoding: some audio detail is discarded. Album artwork is not carried to Ogg Vorbis. AudioPaths.conversion —
wav opus — audio supported lossy metadata Lossy encoding: some audio detail is discarded. Album artwork is not carried to Opus. Opus always works at 48 kHz. AudioPaths.conversion 1.45 (engine wall, run 1 (s); WAV → Opus 128 kb/s)
wav wav — audio supported lossless metadata Sample-exact when the sample rate, bit depth and channels are kept and loudness normalisation is off. Album artwork is not carried to WAV. WAV stores only a few tags (title, artist, album, date, genre, comment, track). AudioPaths.conversion, losslessPathsAreSampleExact —
wav wma — audio supported lossy metadata Lossy encoding: some audio detail is discarded. Album artwork is not carried to Windows Media Audio. Stereo at most: surround sound is mixed down. AudioPaths.conversion —
wma aac — audio supported lossy — Both formats are lossy: re-encoding loses a little more quality (generation loss). AAC (ADTS) carries no tags here: title, artist and the other tags are dropped. Album artwork is not carried to AAC (ADTS). AudioPaths.conversion —
wma aiff — audio supported lossy metadata The source is lossy: the file gets bigger, and quality lost earlier is not restored. AudioPaths.conversion —
wma alac — audio supported lossy metadata The source is lossy: the file gets bigger, and quality lost earlier is not restored. AudioPaths.conversion —
wma caf — audio supported lossy metadata The source is lossy: the file gets bigger, and quality lost earlier is not restored. Album artwork is not carried to Core Audio Format. AudioPaths.conversion —
wma flac — audio supported lossy metadata The source is lossy: the file gets bigger, and quality lost earlier is not restored. AudioPaths.conversion —
wma m4a — audio supported lossy metadata Both formats are lossy: re-encoding loses a little more quality (generation loss). AudioPaths.conversion —
wma mp3 — audio supported lossy metadata Both formats are lossy: re-encoding loses a little more quality (generation loss). Stereo at most: surround sound is mixed down. AudioPaths.conversion —
wma ogg — audio supported lossy metadata Both formats are lossy: re-encoding loses a little more quality (generation loss). Album artwork is not carried to Ogg Vorbis. AudioPaths.conversion —
wma opus — audio supported lossy metadata Both formats are lossy: re-encoding loses a little more quality (generation loss). Album artwork is not carried to Opus. Opus always works at 48 kHz. AudioPaths.conversion —
wma wav — audio supported lossy metadata The source is lossy: the file gets bigger, and quality lost earlier is not restored. Album artwork is not carried to WAV. WAV stores only a few tags (title, artist, album, date, genre, comment, track). AudioPaths.conversion —
wma wma — audio supported lossy metadata Both formats are lossy: re-encoding loses a little more quality (generation loss). Album artwork is not carried to Windows Media Audio. Stereo at most: surround sound is mixed down. AudioPaths.conversion —

# document sources (91 paths)

Source Destination Operation Engine Status Fidelity Preserves Limitations Tests Performance
doc docx — documents supported structural images, links, metadata, tables, textStyles Converted by LibreOffice: compatibility with Microsoft Word is good but not complete (SmartArt, some fields, text effects, exact fonts and spacing can differ). Macros are never run and are not carried over. Embedded OLE objects may be reduced to their preview image. Word 97–2004 input is first imported by LibreOffice. DocumentPathTests.declaredPath 1.34 s (time; LibreOffice DOC→DOCX)
doc epub — ebooks supported structural chapters, images, links, metadata, tables, textStyles Reflowable book: page layout, fonts, headers, footers and text boxes are dropped. Chapters are split at top-level headings. Word 97–2004 input is first imported by LibreOffice. DocumentPathTests.declaredPath —
doc html — documents supported structural images, links, metadata, tables, textStyles Page layout, headers and footers are dropped. Images are embedded in the HTML file as data URIs. Word 97–2004 input is first imported by LibreOffice. DocumentPathTests.declaredPath —
doc latex — documents supported structural links, metadata, tables, textStyles Images are kept as references only; the image files are not written (use “Extract media” on the source). Page layout, fonts, colours, headers and footers are dropped; complex or merged-cell tables are simplified. The result is LaTeX source; it is not compiled to PDF. Word 97–2004 input is first imported by LibreOffice. DocumentPathTests.declaredPath —
doc markdown — documents supported structural links, metadata, tables, textStyles Images are kept as references only; the image files are not written (use “Extract media” on the source). Page layout, fonts, colours, headers and footers are dropped; complex or merged-cell tables are simplified. Word 97–2004 input is first imported by LibreOffice. DocumentPathTests.declaredPath 1.20 s (two steps) (time; LibreOffice + Pandoc DOC→Markdown)
doc odt — documents supported structural images, links, metadata, tables, textStyles Converted by LibreOffice: compatibility with Microsoft Word is good but not complete (SmartArt, some fields, text effects, exact fonts and spacing can differ). Macros are never run and are not carried over. Embedded OLE objects may be reduced to their preview image. Word 97–2004 input is first imported by LibreOffice. DocumentPathTests.declaredPath —
doc pdf — documents supported structural images, links, metadata, textStyles Laid out by LibreOffice, not by the application that wrote the file: line and page breaks can differ from Microsoft Word or Pages. Fonts that are not installed are substituted. Macros are never run; comments are not printed. Word 97–2004 input is first imported by LibreOffice. DocumentPathTests.declaredPath —
doc rst — documents supported structural links, tables, textStyles Images are kept as references only; the image files are not written (use “Extract media” on the source). Page layout, fonts, colours, headers and footers are dropped; complex or merged-cell tables are simplified. Word 97–2004 input is first imported by LibreOffice. DocumentPathTests.declaredPath —
doc rtf — documents supported structural images, links, metadata, tables, textStyles Converted by LibreOffice: compatibility with Microsoft Word is good but not complete (SmartArt, some fields, text effects, exact fonts and spacing can differ). Macros are never run and are not carried over. Embedded OLE objects may be reduced to their preview image. Word 97–2004 input is first imported by LibreOffice. DocumentPathTests.declaredPath —
doc txt — documents supported partial — Text only: fonts, emphasis, images, table grids, links and page layout are dropped. Headers, footers and text boxes are not included. Word 97–2004 input is first imported by LibreOffice. DocumentPathTests.declaredPath —
docx doc — documents supported structural images, links, metadata, tables, textStyles Converted by LibreOffice: compatibility with Microsoft Word is good but not complete (SmartArt, some fields, text effects, exact fonts and spacing can differ). Macros are never run and are not carried over. Embedded OLE objects may be reduced to their preview image. Word 97–2004 is a legacy format: newer features (modern text effects, content controls) have no equivalent. DocumentPathTests.declaredPath —
docx epub — ebooks supported structural chapters, images, links, metadata, tables, textStyles Reflowable book: page layout, fonts, headers, footers and text boxes are dropped. Chapters are split at top-level headings. DocumentPathTests.declaredPath 0.08 s (time; Pandoc DOCX→EPUB)
docx folder media.extract documents supported lossless images Only the embedded image and media files are extracted, byte for byte, under their names in the package; linked (external) images are not fetched. EbookTests.coverAndMedia —
docx html — documents supported structural images, links, metadata, tables, textStyles Page layout, headers and footers are dropped. Images are embedded in the HTML file as data URIs. DocumentPathTests.declaredPath —
docx latex — documents supported structural links, metadata, tables, textStyles Images are kept as references only; the image files are not written (use “Extract media” on the source). Page layout, fonts, colours, headers and footers are dropped; complex or merged-cell tables are simplified. The result is LaTeX source; it is not compiled to PDF. DocumentPathTests.declaredPath —
docx markdown — documents supported structural links, metadata, tables, textStyles Images are kept as references only; the image files are not written (use “Extract media” on the source). Page layout, fonts, colours, headers and footers are dropped; complex or merged-cell tables are simplified. DocumentPathTests.declaredPath 0.06 s (time; Pandoc DOCX→Markdown)
docx odt — documents supported structural images, links, metadata, tables, textStyles Converted by LibreOffice: compatibility with Microsoft Word is good but not complete (SmartArt, some fields, text effects, exact fonts and spacing can differ). Macros are never run and are not carried over. Embedded OLE objects may be reduced to their preview image. DocumentPathTests.declaredPath 1.49 s (engine call incl. reading source and output) (time; LibreOffice DOCX→ODT)
docx pdf — documents supported structural images, links, metadata, textStyles Laid out by LibreOffice, not by the application that wrote the file: line and page breaks can differ from Microsoft Word or Pages. Fonts that are not installed are substituted. Macros are never run; comments are not printed. DocumentPathTests.declaredPath 1.40 s (time; LibreOffice DOCX→PDF, 1 page, run 1)
docx rst — documents supported structural links, tables, textStyles Images are kept as references only; the image files are not written (use “Extract media” on the source). Page layout, fonts, colours, headers and footers are dropped; complex or merged-cell tables are simplified. DocumentPathTests.declaredPath —
docx rtf — documents supported structural images, links, metadata, tables, textStyles Converted by LibreOffice: compatibility with Microsoft Word is good but not complete (SmartArt, some fields, text effects, exact fonts and spacing can differ). Macros are never run and are not carried over. Embedded OLE objects may be reduced to their preview image. DocumentPathTests.declaredPath —
docx txt — documents supported partial — Text only: fonts, emphasis, images, table grids, links and page layout are dropped. Headers, footers and text boxes are not included. DocumentPathTests.declaredPath —
html doc — documents supported structural images, links, metadata, tables, textStyles The document is rebuilt from its structure (headings, lists, tables, images, links, notes); the source's visual styling is replaced by default styles. Word 97–2004 is a legacy format: newer features (modern text effects, content controls) have no equivalent. CSS styling and scripts are ignored; remote images are never downloaded. DocumentPathTests.declaredPath —
html docx — documents supported structural images, links, metadata, tables, textStyles The document is rebuilt from its structure (headings, lists, tables, images, links, notes); the source's visual styling is replaced by default styles. CSS styling and scripts are ignored; remote images are never downloaded. DocumentPathTests.declaredPath —
html epub — ebooks supported structural chapters, images, links, metadata, tables, textStyles Reflowable book: page layout, fonts, headers, footers and text boxes are dropped. Chapters are split at top-level headings. CSS styling and scripts are ignored; remote images are never downloaded. DocumentPathTests.declaredPath —
html latex — documents supported structural links, metadata, tables, textStyles Images are kept as references only; the image files are not written (use “Extract media” on the source). Page layout, fonts, colours, headers and footers are dropped; complex or merged-cell tables are simplified. The result is LaTeX source; it is not compiled to PDF. CSS styling and scripts are ignored; remote images are never downloaded. DocumentPathTests.declaredPath —
html markdown — documents supported structural links, metadata, tables, textStyles Images are kept as references only; the image files are not written (use “Extract media” on the source). Page layout, fonts, colours, headers and footers are dropped; complex or merged-cell tables are simplified. CSS styling and scripts are ignored; remote images are never downloaded. DocumentPathTests.declaredPath —
html odt — documents supported structural images, links, metadata, tables, textStyles The document is rebuilt from its structure (headings, lists, tables, images, links, notes); the source's visual styling is replaced by default styles. CSS styling and scripts are ignored; remote images are never downloaded. DocumentPathTests.declaredPath —
html pdf — documents supported structural images, links, metadata, textStyles Rendered by WebKit with JavaScript disabled and without any network access: remote images, fonts and stylesheets are missing. No table structure, bookmarks or tagged-PDF accessibility information. DocumentPathTests.declaredPath 0.14 s (time; WebKit HTML→PDF, 1 page)
html rst — documents supported structural links, tables, textStyles Images are kept as references only; the image files are not written (use “Extract media” on the source). Page layout, fonts, colours, headers and footers are dropped; complex or merged-cell tables are simplified. CSS styling and scripts are ignored; remote images are never downloaded. DocumentPathTests.declaredPath —
html rtf — documents supported structural images, links, tables, textStyles The document is rebuilt from its structure (headings, lists, tables, images, links, notes); the source's visual styling is replaced by default styles. Footnotes may be flattened into the text flow; document properties are not written. CSS styling and scripts are ignored; remote images are never downloaded. DocumentPathTests.declaredPath —
html txt — documents supported partial — Text only: fonts, emphasis, images, table grids, links and page layout are dropped. CSS styling and scripts are ignored; remote images are never downloaded. DocumentPathTests.declaredPath —
latex doc — documents supported structural images, links, metadata, tables, textStyles The document is rebuilt from its structure (headings, lists, tables, images, links, notes); the source's visual styling is replaced by default styles. Word 97–2004 is a legacy format: newer features (modern text effects, content controls) have no equivalent. LaTeX is read for structure only: custom macros, most packages, figures drawn in TeX and bibliographies are not interpreted. Math is kept as TeX or MathML where the target allows. DocumentPathTests.declaredPath —
latex docx — documents supported structural images, links, metadata, tables, textStyles The document is rebuilt from its structure (headings, lists, tables, images, links, notes); the source's visual styling is replaced by default styles. LaTeX is read for structure only: custom macros, most packages, figures drawn in TeX and bibliographies are not interpreted. Math is kept as TeX or MathML where the target allows. DocumentPathTests.declaredPath —
latex epub — ebooks supported structural chapters, images, links, metadata, tables, textStyles Reflowable book: page layout, fonts, headers, footers and text boxes are dropped. Chapters are split at top-level headings. LaTeX is read for structure only: custom macros, most packages, figures drawn in TeX and bibliographies are not interpreted. Math is kept as TeX or MathML where the target allows. DocumentPathTests.declaredPath —
latex html — documents supported structural images, links, metadata, tables, textStyles Page layout, headers and footers are dropped. Images are embedded in the HTML file as data URIs. LaTeX is read for structure only: custom macros, most packages, figures drawn in TeX and bibliographies are not interpreted. Math is kept as TeX or MathML where the target allows. DocumentPathTests.declaredPath —
latex markdown — documents supported structural links, metadata, tables, textStyles Images are kept as references only; the image files are not written (use “Extract media” on the source). Page layout, fonts, colours, headers and footers are dropped; complex or merged-cell tables are simplified. LaTeX is read for structure only: custom macros, most packages, figures drawn in TeX and bibliographies are not interpreted. Math is kept as TeX or MathML where the target allows. DocumentPathTests.declaredPath —
latex odt — documents supported structural images, links, metadata, tables, textStyles The document is rebuilt from its structure (headings, lists, tables, images, links, notes); the source's visual styling is replaced by default styles. LaTeX is read for structure only: custom macros, most packages, figures drawn in TeX and bibliographies are not interpreted. Math is kept as TeX or MathML where the target allows. DocumentPathTests.declaredPath —
latex rst — documents supported structural links, tables, textStyles Images are kept as references only; the image files are not written (use “Extract media” on the source). Page layout, fonts, colours, headers and footers are dropped; complex or merged-cell tables are simplified. LaTeX is read for structure only: custom macros, most packages, figures drawn in TeX and bibliographies are not interpreted. Math is kept as TeX or MathML where the target allows. DocumentPathTests.declaredPath —
latex rtf — documents supported structural images, links, tables, textStyles The document is rebuilt from its structure (headings, lists, tables, images, links, notes); the source's visual styling is replaced by default styles. Footnotes may be flattened into the text flow; document properties are not written. LaTeX is read for structure only: custom macros, most packages, figures drawn in TeX and bibliographies are not interpreted. Math is kept as TeX or MathML where the target allows. DocumentPathTests.declaredPath —
latex txt — documents supported partial — Text only: fonts, emphasis, images, table grids, links and page layout are dropped. LaTeX is read for structure only: custom macros, most packages, figures drawn in TeX and bibliographies are not interpreted. Math is kept as TeX or MathML where the target allows. DocumentPathTests.declaredPath —
markdown doc — documents supported structural images, links, metadata, tables, textStyles The document is rebuilt from its structure (headings, lists, tables, images, links, notes); the source's visual styling is replaced by default styles. Word 97–2004 is a legacy format: newer features (modern text effects, content controls) have no equivalent. DocumentPathTests.declaredPath —
markdown docx — documents supported structural images, links, metadata, tables, textStyles The document is rebuilt from its structure (headings, lists, tables, images, links, notes); the source's visual styling is replaced by default styles. DocumentPathTests.declaredPath 0.22 s (time; Pandoc Markdown→DOCX)
markdown epub — ebooks supported structural chapters, images, links, metadata, tables, textStyles Reflowable book: page layout, fonts, headers, footers and text boxes are dropped. Chapters are split at top-level headings. DocumentPathTests.declaredPath —
markdown html — documents supported structural images, links, metadata, tables, textStyles Page layout, headers and footers are dropped. Images are embedded in the HTML file as data URIs. DocumentPathTests.declaredPath 0.21 s (time; Pandoc Markdown→HTML)
markdown latex — documents supported structural links, metadata, tables, textStyles Images are kept as references only; the image files are not written (use “Extract media” on the source). Page layout, fonts, colours, headers and footers are dropped; complex or merged-cell tables are simplified. The result is LaTeX source; it is not compiled to PDF. DocumentPathTests.declaredPath —
markdown odt — documents supported structural images, links, metadata, tables, textStyles The document is rebuilt from its structure (headings, lists, tables, images, links, notes); the source's visual styling is replaced by default styles. DocumentPathTests.declaredPath —
markdown pdf — documents supported structural images, links, metadata, textStyles Rendered by WebKit with JavaScript disabled and without any network access: remote images, fonts and stylesheets are missing. Typeset with ORIX Convert's built-in print stylesheet; this is a reflowed rendering, not the source's own page design. No table structure, bookmarks or tagged-PDF accessibility information. DocumentPathTests.declaredPath 0.41 s (time; Pandoc + WebKit Markdown→PDF, 1 page)
markdown rst — documents supported structural links, tables, textStyles Images are kept as references only; the image files are not written (use “Extract media” on the source). Page layout, fonts, colours, headers and footers are dropped; complex or merged-cell tables are simplified. DocumentPathTests.declaredPath —
markdown rtf — documents supported structural images, links, tables, textStyles The document is rebuilt from its structure (headings, lists, tables, images, links, notes); the source's visual styling is replaced by default styles. Footnotes may be flattened into the text flow; document properties are not written. DocumentPathTests.declaredPath —
markdown txt — documents supported partial — Text only: fonts, emphasis, images, table grids, links and page layout are dropped. DocumentPathTests.declaredPath —
odt doc — documents supported structural images, links, metadata, tables, textStyles Converted by LibreOffice: compatibility with Microsoft Word is good but not complete (SmartArt, some fields, text effects, exact fonts and spacing can differ). Macros are never run and are not carried over. Embedded OLE objects may be reduced to their preview image. Word 97–2004 is a legacy format: newer features (modern text effects, content controls) have no equivalent. DocumentPathTests.declaredPath —
odt docx — documents supported structural images, links, metadata, tables, textStyles Converted by LibreOffice: compatibility with Microsoft Word is good but not complete (SmartArt, some fields, text effects, exact fonts and spacing can differ). Macros are never run and are not carried over. Embedded OLE objects may be reduced to their preview image. DocumentPathTests.declaredPath —
odt epub — ebooks supported structural chapters, images, links, metadata, tables, textStyles Reflowable book: page layout, fonts, headers, footers and text boxes are dropped. Chapters are split at top-level headings. DocumentPathTests.declaredPath —
odt folder media.extract documents supported lossless images Only the embedded image and media files are extracted, byte for byte, under their names in the package; linked (external) images are not fetched. EbookTests.coverAndMedia —
odt html — documents supported structural images, links, metadata, tables, textStyles Page layout, headers and footers are dropped. Images are embedded in the HTML file as data URIs. DocumentPathTests.declaredPath —
odt latex — documents supported structural links, metadata, tables, textStyles Images are kept as references only; the image files are not written (use “Extract media” on the source). Page layout, fonts, colours, headers and footers are dropped; complex or merged-cell tables are simplified. The result is LaTeX source; it is not compiled to PDF. DocumentPathTests.declaredPath —
odt markdown — documents supported structural links, metadata, tables, textStyles Images are kept as references only; the image files are not written (use “Extract media” on the source). Page layout, fonts, colours, headers and footers are dropped; complex or merged-cell tables are simplified. DocumentPathTests.declaredPath —
odt pdf — documents supported structural images, links, metadata, textStyles Laid out by LibreOffice, not by the application that wrote the file: line and page breaks can differ from Microsoft Word or Pages. Fonts that are not installed are substituted. Macros are never run; comments are not printed. DocumentPathTests.declaredPath —
odt rst — documents supported structural links, tables, textStyles Images are kept as references only; the image files are not written (use “Extract media” on the source). Page layout, fonts, colours, headers and footers are dropped; complex or merged-cell tables are simplified. DocumentPathTests.declaredPath —
odt rtf — documents supported structural images, links, metadata, tables, textStyles Converted by LibreOffice: compatibility with Microsoft Word is good but not complete (SmartArt, some fields, text effects, exact fonts and spacing can differ). Macros are never run and are not carried over. Embedded OLE objects may be reduced to their preview image. DocumentPathTests.declaredPath —
odt txt — documents supported partial — Text only: fonts, emphasis, images, table grids, links and page layout are dropped. Headers, footers and text boxes are not included. DocumentPathTests.declaredPath —
rst doc — documents supported structural images, links, tables, textStyles The document is rebuilt from its structure (headings, lists, tables, images, links, notes); the source's visual styling is replaced by default styles. Word 97–2004 is a legacy format: newer features (modern text effects, content controls) have no equivalent. DocumentPathTests.declaredPath —
rst docx — documents supported structural images, links, tables, textStyles The document is rebuilt from its structure (headings, lists, tables, images, links, notes); the source's visual styling is replaced by default styles. DocumentPathTests.declaredPath —
rst epub — ebooks supported structural chapters, images, links, tables, textStyles Reflowable book: page layout, fonts, headers, footers and text boxes are dropped. Chapters are split at top-level headings. DocumentPathTests.declaredPath —
rst html — documents supported structural images, links, tables, textStyles Page layout, headers and footers are dropped. Images are embedded in the HTML file as data URIs. DocumentPathTests.declaredPath —
rst latex — documents supported structural links, tables, textStyles Images are kept as references only; the image files are not written (use “Extract media” on the source). Page layout, fonts, colours, headers and footers are dropped; complex or merged-cell tables are simplified. The result is LaTeX source; it is not compiled to PDF. DocumentPathTests.declaredPath —
rst markdown — documents supported structural links, tables, textStyles Images are kept as references only; the image files are not written (use “Extract media” on the source). Page layout, fonts, colours, headers and footers are dropped; complex or merged-cell tables are simplified. DocumentPathTests.declaredPath —
rst odt — documents supported structural images, links, tables, textStyles The document is rebuilt from its structure (headings, lists, tables, images, links, notes); the source's visual styling is replaced by default styles. DocumentPathTests.declaredPath —
rst pdf — documents supported structural images, links, textStyles Rendered by WebKit with JavaScript disabled and without any network access: remote images, fonts and stylesheets are missing. Typeset with ORIX Convert's built-in print stylesheet; this is a reflowed rendering, not the source's own page design. No table structure, bookmarks or tagged-PDF accessibility information. DocumentPathTests.declaredPath —
rst rtf — documents supported structural images, links, tables, textStyles The document is rebuilt from its structure (headings, lists, tables, images, links, notes); the source's visual styling is replaced by default styles. Footnotes may be flattened into the text flow; document properties are not written. DocumentPathTests.declaredPath —
rst txt — documents supported partial — Text only: fonts, emphasis, images, table grids, links and page layout are dropped. DocumentPathTests.declaredPath —
rtf doc — documents supported structural images, links, tables, textStyles Converted by LibreOffice: compatibility with Microsoft Word is good but not complete (SmartArt, some fields, text effects, exact fonts and spacing can differ). Macros are never run and are not carried over. Embedded OLE objects may be reduced to their preview image. Word 97–2004 is a legacy format: newer features (modern text effects, content controls) have no equivalent. DocumentPathTests.declaredPath —
rtf docx — documents supported structural images, links, tables, textStyles Converted by LibreOffice: compatibility with Microsoft Word is good but not complete (SmartArt, some fields, text effects, exact fonts and spacing can differ). Macros are never run and are not carried over. Embedded OLE objects may be reduced to their preview image. DocumentPathTests.declaredPath 0.02 s (in process) (time; Apple text system RTF→DOCX)
rtf epub — ebooks supported structural chapters, images, links, tables, textStyles Reflowable book: page layout, fonts, headers, footers and text boxes are dropped. Chapters are split at top-level headings. RTF input is first imported by LibreOffice. DocumentPathTests.declaredPath —
rtf html — documents supported structural images, links, tables, textStyles Page layout, headers and footers are dropped. Images are embedded in the HTML file as data URIs. RTF input is first imported by LibreOffice. DocumentPathTests.declaredPath —
rtf latex — documents supported structural links, tables, textStyles Images are kept as references only; the image files are not written (use “Extract media” on the source). Page layout, fonts, colours, headers and footers are dropped; complex or merged-cell tables are simplified. The result is LaTeX source; it is not compiled to PDF. RTF input is first imported by LibreOffice. DocumentPathTests.declaredPath —
rtf markdown — documents supported structural links, tables, textStyles Images are kept as references only; the image files are not written (use “Extract media” on the source). Page layout, fonts, colours, headers and footers are dropped; complex or merged-cell tables are simplified. RTF input is first imported by LibreOffice. DocumentPathTests.declaredPath —
rtf odt — documents supported structural images, links, tables, textStyles Converted by LibreOffice: compatibility with Microsoft Word is good but not complete (SmartArt, some fields, text effects, exact fonts and spacing can differ). Macros are never run and are not carried over. Embedded OLE objects may be reduced to their preview image. DocumentPathTests.declaredPath —
rtf pdf — documents supported structural images, links, textStyles Laid out by LibreOffice, not by the application that wrote the file: line and page breaks can differ from Microsoft Word or Pages. Fonts that are not installed are substituted. Macros are never run; comments are not printed. DocumentPathTests.declaredPath —
rtf rst — documents supported structural links, tables, textStyles Images are kept as references only; the image files are not written (use “Extract media” on the source). Page layout, fonts, colours, headers and footers are dropped; complex or merged-cell tables are simplified. RTF input is first imported by LibreOffice. DocumentPathTests.declaredPath —
rtf txt — documents supported partial — Text only: fonts, emphasis, images, table grids, links and page layout are dropped. Headers, footers and text boxes are not included. RTF input is first imported by LibreOffice. DocumentPathTests.declaredPath —
txt doc — documents supported structural — Word 97–2004 is a legacy format: newer features (modern text effects, content controls) have no equivalent. Plain text has no structure: paragraphs are split at blank lines and no character is interpreted as markup. DocumentPathTests.declaredPath —
txt docx — documents supported structural — Plain text has no structure: paragraphs are split at blank lines and no character is interpreted as markup. DocumentPathTests.declaredPath —
txt epub — ebooks supported structural — Plain text has no headings: the book is a single chapter. DocumentPathTests.declaredPath —
txt html — documents supported structural — Plain text has no structure: paragraphs are split at blank lines and no character is interpreted as markup. DocumentPathTests.declaredPath —
txt latex — documents supported structural — The result is LaTeX source; it is not compiled to PDF. Plain text has no structure: paragraphs are split at blank lines and no character is interpreted as markup. DocumentPathTests.declaredPath —
txt markdown — documents supported structural — Plain text has no structure: paragraphs are split at blank lines and no character is interpreted as markup. DocumentPathTests.declaredPath —
txt odt — documents supported structural — Plain text has no structure: paragraphs are split at blank lines and no character is interpreted as markup. DocumentPathTests.declaredPath —
txt pdf — documents supported structural — Laid out by LibreOffice, not by the application that wrote the file: line and page breaks can differ from Microsoft Word or Pages. Fonts that are not installed are substituted. Macros are never run; comments are not printed. Plain text is laid out by LibreOffice with a default font and page setup. DocumentPathTests.declaredPath —
txt rst — documents supported structural — Plain text has no structure: paragraphs are split at blank lines and no character is interpreted as markup. DocumentPathTests.declaredPath —
txt rtf — documents supported structural — Plain text has no structure: paragraphs are split at blank lines and no character is interpreted as markup. DocumentPathTests.declaredPath —

# pdf sources (17 paths)

Source Destination Operation Engine Status Fidelity Preserves Limitations Tests Performance
pdf docx — pdf experimental partial textStyles This is a text reconstruction, not a faithful editable twin: page layout, columns, tables, images and vector graphics are not rebuilt. Paragraphs are guessed from line geometry and headings from font size (kept as larger text, not as Word heading styles); both can be wrong. Fonts that are not installed on this Mac are replaced by Helvetica, Times New Roman or Courier New with the same size, bold and italic. Scanned PDFs have no text to rebuild: make them searchable with OCR first. RichText/rebuildsParagraphsAndStyles(docx), scannedDocumentsAreRefusedWithAdvice 102 pages/s (run 1; PDF → DOCX)
pdf folder pdf.extract-images pdf supported partial images, transparency JPEG and JPEG 2000 images are written as their original data; every other image is decoded and written as PNG (CMYK is converted to sRGB). CCITT fax and JBIG2 images, images in Separation / DeviceN / Lab colour spaces and inline images are not extracted; they are counted in a warning. Images are written as stored in the file, not as placed on the page (no cropping, rotation or stencil masks). ImageExtraction/jpegBytesAndLosslessPixels, handWrittenColourSpaces, unsupportedFiltersAndInlineImagesAreCounted —
pdf folder pdf.split pdf supported lossless chapters, dimensions, images, links, metadata, vectorGeometry The accessibility tag structure, file attachments, XMP metadata, document JavaScript and digital signatures are not carried over. Form fields stay on their pages with their values, but the form definition (AcroForm dictionary) is not carried over. Internal links that point to pages left out of the result stop working. PageOperations/splitModes —
pdf jpeg — pdf supported lossy dimensions Pages become flat images: text is no longer selectable and links, form fields and the outline are gone. Several pages produce a folder with one image per page (or one multi-page TIFF when that option is on). A page above 256 megapixels or 32000 pixels a side is rendered at a lower resolution, with a warning. JPEG has no transparency and is lossy. Rendering/pageSelectionWidthAndGreyJPEG 69.9 pages/s (run 1; PDF → JPEG, 150 dpi)
pdf markdown — pdf supported partial — Only the text is kept, in the order PDFKit reads it; fonts, layout, tables, images and links are dropped. Multi-column pages and tables can come out in an unexpected order. Markdown output only adds a “Page N” heading per page; headings, lists and emphasis are not detected, and characters such as * or # in the text are not escaped. TextExport/markdownAddsOnlyPageHeadings, scannedDocumentIsReportedAndReadWithOCR —
pdf pdf pdf.compress pdf supported lossy chapters, dimensions, links, metadata, vectorGeometry Only images are recompressed: those above the chosen resolution are downsampled, and opaque images are stored as JPEG at the chosen quality. Text and vector drawing are kept as they are. Images with transparency are downsampled but stay lossless, so they shrink less. If the result is not smaller than the source, the original file is returned unchanged. Pages are re-emitted by Quartz: the accessibility tag structure (tagged PDF) is not carried over. File attachments, JavaScript and digital signatures of the source are not carried over. Compression/imageHeavyDocumentShrinks, neverReturnsABiggerFile, structureSurvivesCompression —
pdf pdf pdf.delete-pages pdf supported lossless chapters, dimensions, images, links, metadata, vectorGeometry The accessibility tag structure, file attachments, XMP metadata, document JavaScript and digital signatures are not carried over. Form fields stay on their pages with their values, but the form definition (AcroForm dictionary) is not carried over. Internal links that point to pages left out of the result stop working. PageOperations/extractDeleteReorder —
pdf pdf pdf.extract-pages pdf supported lossless chapters, dimensions, images, links, metadata, vectorGeometry The accessibility tag structure, file attachments, XMP metadata, document JavaScript and digital signatures are not carried over. Form fields stay on their pages with their values, but the form definition (AcroForm dictionary) is not carried over. Internal links that point to pages left out of the result stop working. PageOperations/extractDeleteReorder —
pdf pdf pdf.merge pdf supported lossless chapters, dimensions, images, links, metadata, vectorGeometry Each file's outline is nested under an entry named after the file. Document metadata comes from the first file. Form fields that share a name across files end up sharing their value. The accessibility tag structure, file attachments, XMP metadata, document JavaScript and digital signatures are not carried over. Form fields stay on their pages with their values, but the form definition (AcroForm dictionary) is not carried over. Internal links that point to pages left out of the result stop working. PageOperations/mergeKeepsOrderOutlineAndAnnotations —
pdf pdf pdf.ocr pdf supported structural chapters, dimensions, images, links, metadata, vectorGeometry Recognition runs on this Mac with Vision and can make mistakes, especially on handwriting, small or skewed text. The added text is invisible and positioned word by word; it is meant for search, selection and copy, not for editing. Pages are re-emitted by Quartz: the accessibility tag structure (tagged PDF) is not carried over. File attachments, JavaScript and digital signatures of the source are not carried over. SearchablePDF/invisibleLayerMatchesTheScan (150 and 300 dpi), outlineLinksAnnotationsAndFormsSurvive, documentsWithTextAreReturnedUnchanged, everyPageOnRequest —
pdf pdf pdf.reorder-pages pdf supported lossless chapters, dimensions, images, links, metadata, vectorGeometry The accessibility tag structure, file attachments, XMP metadata, document JavaScript and digital signatures are not carried over. Form fields stay on their pages with their values, but the form definition (AcroForm dictionary) is not carried over. Internal links that point to pages left out of the result stop working. PageOperations/extractDeleteReorder —
pdf pdf pdf.rotate pdf supported lossless chapters, dimensions, images, links, metadata, vectorGeometry Rotation is stored as the page's rotation attribute (multiples of 90°); page content is not redrawn. The file is rewritten, so a digital signature does not stay valid. PageOperations/rotationIsWrittenAsPageRotation —
pdf pdf pdf.set-metadata pdf supported lossless chapters, dimensions, images, links, vectorGeometry Only the Info dictionary (title, author, subject, keywords, creator, producer, dates) is edited or removed; an XMP packet is reported, not rewritten. The file is rewritten, so a digital signature does not stay valid. PageOperations/metadataCanBeSetAndStripped —
pdf png — pdf supported lossy dimensions Pages become flat images: text is no longer selectable and links, form fields and the outline are gone. Several pages produce a folder with one image per page (or one multi-page TIFF when that option is on). A page above 256 megapixels or 32000 pixels a side is rendered at a lower resolution, with a warning. Rendering/singlePageToPNG, severalPagesGoToAFolder, transparentBackground, annotationsCanBeLeftOut, cropBoxVersusMediaBox, stripRenderingIsSeamless 19.7 pages/s (run 1; PDF → PNG, 150 dpi (1275 × 1650))
pdf rtf — pdf experimental partial textStyles This is a text reconstruction, not a faithful editable twin: page layout, columns, tables, images and vector graphics are not rebuilt. Paragraphs are guessed from line geometry and headings from font size (kept as larger text, not as Word heading styles); both can be wrong. Fonts that are not installed on this Mac are replaced by Helvetica, Times New Roman or Courier New with the same size, bold and italic. Scanned PDFs have no text to rebuild: make them searchable with OCR first. RichText/rebuildsParagraphsAndStyles(rtf) —
pdf tiff — pdf supported lossy dimensions Pages become flat images: text is no longer selectable and links, form fields and the outline are gone. Several pages produce a folder with one image per page (or one multi-page TIFF when that option is on). A page above 256 megapixels or 32000 pixels a side is rendered at a lower resolution, with a warning. Rendering/multiPageTIFFAndPerPageTIFF 24.4 pages/s (run 1; PDF → TIFF (LZW), 150 dpi)
pdf txt — pdf supported partial — Only the text is kept, in the order PDFKit reads it; fonts, layout, tables, images and links are dropped. Multi-column pages and tables can come out in an unexpected order. Pages without a text layer are read with on-device text recognition (unless turned off), which can make mistakes. TextExport/plainTextWithSeparators, hyphenationRepair, positionalOrderSortsTopToBottom, scannedDocumentIsReportedAndReadWithOCR 646 pages/s (run 1; PDF → TXT)

# spreadsheet sources (42 paths)

Source Destination Operation Engine Status Fidelity Preserves Limitations Tests Performance
ods arrow — data supported partial tables Only cell values are converted. Formulas (their last calculated values are kept), styles, number formats, column widths, charts, images, pivot tables, comments, data validation and conditional formatting are not carried over. Merged cells keep their value in the first cell only. One sheet per output file: the first one by default, a chosen one, or all of them as separate files. Columns are written as int64, float64, bool, utf8, binary, date32, time64[us], timestamp[us] and decimal128; nested values as JSON text. Buffers are not compressed. MatrixTests.path (ods → arrow) —
ods csv — data supported partial tables Only cell values are converted. Formulas (their last calculated values are kept), styles, number formats, column widths, charts, images, pivot tables, comments, data validation and conditional formatting are not carried over. Merged cells keep their value in the first cell only. One sheet per output file: the first one by default, a chosen one, or all of them as separate files. Text formats carry no types: dates are written in ISO 8601, binary values as Base64, nested values as JSON text. MatrixTests.path (ods → csv) —
ods duckdb — data supported partial multipleSheets, tables Only cell values are converted. Formulas (their last calculated values are kept), styles, number formats, column widths, charts, images, pivot tables, comments, data validation and conditional formatting are not carried over. Merged cells keep their value in the first cell only. MatrixTests.path (ods → duckdb) —
ods json — data supported partial tables Only cell values are converted. Formulas (their last calculated values are kept), styles, number formats, column widths, charts, images, pivot tables, comments, data validation and conditional formatting are not carried over. Merged cells keep their value in the first cell only. One sheet per output file: the first one by default, a chosen one, or all of them as separate files. JSON has no date, binary, NaN or infinity: dates become ISO 8601 text, binary values Base64 text, NaN and infinities null. MatrixTests.path (ods → json) —
ods jsonl — data supported partial tables Only cell values are converted. Formulas (their last calculated values are kept), styles, number formats, column widths, charts, images, pivot tables, comments, data validation and conditional formatting are not carried over. Merged cells keep their value in the first cell only. One sheet per output file: the first one by default, a chosen one, or all of them as separate files. JSON has no date, binary, NaN or infinity: dates become ISO 8601 text, binary values Base64 text, NaN and infinities null. MatrixTests.path (ods → jsonl) —
ods parquet — data supported partial tables Only cell values are converted. Formulas (their last calculated values are kept), styles, number formats, column widths, charts, images, pivot tables, comments, data validation and conditional formatting are not carried over. Merged cells keep their value in the first cell only. One sheet per output file: the first one by default, a chosen one, or all of them as separate files. MatrixTests.path (ods → parquet) —
ods plist — data supported partial tables Only cell values are converted. Formulas (their last calculated values are kept), styles, number formats, column widths, charts, images, pivot tables, comments, data validation and conditional formatting are not carried over. Merged cells keep their value in the first cell only. One sheet per output file: the first one by default, a chosen one, or all of them as separate files. Property lists have no null (empty values are omitted or stop the conversion) and are built in memory (20 million cells at most). MatrixTests.path (ods → plist) —
ods sqlite — data supported partial multipleSheets, tables Only cell values are converted. Formulas (their last calculated values are kept), styles, number formats, column widths, charts, images, pivot tables, comments, data validation and conditional formatting are not carried over. Merged cells keep their value in the first cell only. SQLite numbers are 64-bit: exact decimals and 128-bit integers are stored as TEXT; dates and times as ISO 8601 text in DATE / DATETIME / TIME columns. MatrixTests.path (ods → sqlite) —
ods toml — data supported partial tables Only cell values are converted. Formulas (their last calculated values are kept), styles, number formats, column widths, charts, images, pivot tables, comments, data validation and conditional formatting are not carried over. Merged cells keep their value in the first cell only. One sheet per output file: the first one by default, a chosen one, or all of them as separate files. TOML has no null: empty values are omitted (or stop the conversion, as chosen). MatrixTests.path (ods → toml) —
ods tsv — data supported partial tables Only cell values are converted. Formulas (their last calculated values are kept), styles, number formats, column widths, charts, images, pivot tables, comments, data validation and conditional formatting are not carried over. Merged cells keep their value in the first cell only. One sheet per output file: the first one by default, a chosen one, or all of them as separate files. Text formats carry no types: dates are written in ISO 8601, binary values as Base64, nested values as JSON text. MatrixTests.path (ods → tsv) —
ods xls — data supported partial multipleSheets, tables Only cell values are converted. Formulas (their last calculated values are kept), styles, number formats, column widths, charts, images, pivot tables, comments, data validation and conditional formatting are not carried over. Merged cells keep their value in the first cell only. Spreadsheet numbers are 64-bit floats: integers beyond 2^53 and long decimals are written as text; NaN and infinities become empty cells; time-zoned instants are written as UTC wall-clock times. Date-times are stored to the millisecond. At most 65,536 rows and 256 columns per sheet (legacy format limit); produced through the bundled LibreOffice. MatrixTests.path (ods → xls) —
ods xlsx — data supported partial multipleSheets, tables Only cell values are converted. Formulas (their last calculated values are kept), styles, number formats, column widths, charts, images, pivot tables, comments, data validation and conditional formatting are not carried over. Merged cells keep their value in the first cell only. Spreadsheet numbers are 64-bit floats: integers beyond 2^53 and long decimals are written as text; NaN and infinities become empty cells; time-zoned instants are written as UTC wall-clock times. Date-times are stored to the millisecond. At most 1,048,576 rows and 16,384 columns per sheet (larger tables stop with an error, or are split across sheets on request). MatrixTests.path (ods → xlsx) —
ods xml — data supported partial tables Only cell values are converted. Formulas (their last calculated values are kept), styles, number formats, column widths, charts, images, pivot tables, comments, data validation and conditional formatting are not carried over. Merged cells keep their value in the first cell only. One sheet per output file: the first one by default, a chosen one, or all of them as separate files. XML carries text only; empty values are omitted, so empty text and missing values cannot be told apart when read back. MatrixTests.path (ods → xml) —
ods yaml — data supported partial tables Only cell values are converted. Formulas (their last calculated values are kept), styles, number formats, column widths, charts, images, pivot tables, comments, data validation and conditional formatting are not carried over. Merged cells keep their value in the first cell only. One sheet per output file: the first one by default, a chosen one, or all of them as separate files. Binary values are written as Base64 text. MatrixTests.path (ods → yaml) —
xls arrow — data supported partial tables Only cell values are converted. Formulas (their last calculated values are kept), styles, number formats, column widths, charts, images, pivot tables, comments, data validation and conditional formatting are not carried over. Merged cells keep their value in the first cell only. One sheet per output file: the first one by default, a chosen one, or all of them as separate files. Columns are written as int64, float64, bool, utf8, binary, date32, time64[us], timestamp[us] and decimal128; nested values as JSON text. Buffers are not compressed. MatrixTests.path (xls → arrow) —
xls csv — data supported partial tables Only cell values are converted. Formulas (their last calculated values are kept), styles, number formats, column widths, charts, images, pivot tables, comments, data validation and conditional formatting are not carried over. Merged cells keep their value in the first cell only. One sheet per output file: the first one by default, a chosen one, or all of them as separate files. Text formats carry no types: dates are written in ISO 8601, binary values as Base64, nested values as JSON text. MatrixTests.path (xls → csv) —
xls duckdb — data supported partial multipleSheets, tables Only cell values are converted. Formulas (their last calculated values are kept), styles, number formats, column widths, charts, images, pivot tables, comments, data validation and conditional formatting are not carried over. Merged cells keep their value in the first cell only. MatrixTests.path (xls → duckdb) —
xls json — data supported partial tables Only cell values are converted. Formulas (their last calculated values are kept), styles, number formats, column widths, charts, images, pivot tables, comments, data validation and conditional formatting are not carried over. Merged cells keep their value in the first cell only. One sheet per output file: the first one by default, a chosen one, or all of them as separate files. JSON has no date, binary, NaN or infinity: dates become ISO 8601 text, binary values Base64 text, NaN and infinities null. MatrixTests.path (xls → json) —
xls jsonl — data supported partial tables Only cell values are converted. Formulas (their last calculated values are kept), styles, number formats, column widths, charts, images, pivot tables, comments, data validation and conditional formatting are not carried over. Merged cells keep their value in the first cell only. One sheet per output file: the first one by default, a chosen one, or all of them as separate files. JSON has no date, binary, NaN or infinity: dates become ISO 8601 text, binary values Base64 text, NaN and infinities null. MatrixTests.path (xls → jsonl) —
xls ods — data supported partial multipleSheets, tables Only cell values are converted. Formulas (their last calculated values are kept), styles, number formats, column widths, charts, images, pivot tables, comments, data validation and conditional formatting are not carried over. Merged cells keep their value in the first cell only. Spreadsheet numbers are 64-bit floats: integers beyond 2^53 and long decimals are written as text; NaN and infinities become empty cells; time-zoned instants are written as UTC wall-clock times. At most 1,048,576 rows and 16,384 columns per sheet (larger tables stop with an error, or are split across sheets on request). MatrixTests.path (xls → ods) —
xls parquet — data supported partial tables Only cell values are converted. Formulas (their last calculated values are kept), styles, number formats, column widths, charts, images, pivot tables, comments, data validation and conditional formatting are not carried over. Merged cells keep their value in the first cell only. One sheet per output file: the first one by default, a chosen one, or all of them as separate files. MatrixTests.path (xls → parquet) —
xls plist — data supported partial tables Only cell values are converted. Formulas (their last calculated values are kept), styles, number formats, column widths, charts, images, pivot tables, comments, data validation and conditional formatting are not carried over. Merged cells keep their value in the first cell only. One sheet per output file: the first one by default, a chosen one, or all of them as separate files. Property lists have no null (empty values are omitted or stop the conversion) and are built in memory (20 million cells at most). MatrixTests.path (xls → plist) —
xls sqlite — data supported partial multipleSheets, tables Only cell values are converted. Formulas (their last calculated values are kept), styles, number formats, column widths, charts, images, pivot tables, comments, data validation and conditional formatting are not carried over. Merged cells keep their value in the first cell only. SQLite numbers are 64-bit: exact decimals and 128-bit integers are stored as TEXT; dates and times as ISO 8601 text in DATE / DATETIME / TIME columns. MatrixTests.path (xls → sqlite) —
xls toml — data supported partial tables Only cell values are converted. Formulas (their last calculated values are kept), styles, number formats, column widths, charts, images, pivot tables, comments, data validation and conditional formatting are not carried over. Merged cells keep their value in the first cell only. One sheet per output file: the first one by default, a chosen one, or all of them as separate files. TOML has no null: empty values are omitted (or stop the conversion, as chosen). MatrixTests.path (xls → toml) —
xls tsv — data supported partial tables Only cell values are converted. Formulas (their last calculated values are kept), styles, number formats, column widths, charts, images, pivot tables, comments, data validation and conditional formatting are not carried over. Merged cells keep their value in the first cell only. One sheet per output file: the first one by default, a chosen one, or all of them as separate files. Text formats carry no types: dates are written in ISO 8601, binary values as Base64, nested values as JSON text. MatrixTests.path (xls → tsv) —
xls xlsx — data supported partial multipleSheets, tables Only cell values are converted. Formulas (their last calculated values are kept), styles, number formats, column widths, charts, images, pivot tables, comments, data validation and conditional formatting are not carried over. Merged cells keep their value in the first cell only. Spreadsheet numbers are 64-bit floats: integers beyond 2^53 and long decimals are written as text; NaN and infinities become empty cells; time-zoned instants are written as UTC wall-clock times. Date-times are stored to the millisecond. At most 1,048,576 rows and 16,384 columns per sheet (larger tables stop with an error, or are split across sheets on request). MatrixTests.path (xls → xlsx) —
xls xml — data supported partial tables Only cell values are converted. Formulas (their last calculated values are kept), styles, number formats, column widths, charts, images, pivot tables, comments, data validation and conditional formatting are not carried over. Merged cells keep their value in the first cell only. One sheet per output file: the first one by default, a chosen one, or all of them as separate files. XML carries text only; empty values are omitted, so empty text and missing values cannot be told apart when read back. MatrixTests.path (xls → xml) —
xls yaml — data supported partial tables Only cell values are converted. Formulas (their last calculated values are kept), styles, number formats, column widths, charts, images, pivot tables, comments, data validation and conditional formatting are not carried over. Merged cells keep their value in the first cell only. One sheet per output file: the first one by default, a chosen one, or all of them as separate files. Binary values are written as Base64 text. MatrixTests.path (xls → yaml) —
xlsx arrow — data supported partial tables Only cell values are converted. Formulas (their last calculated values are kept), styles, number formats, column widths, charts, images, pivot tables, comments, data validation and conditional formatting are not carried over. Merged cells keep their value in the first cell only. One sheet per output file: the first one by default, a chosen one, or all of them as separate files. Columns are written as int64, float64, bool, utf8, binary, date32, time64[us], timestamp[us] and decimal128; nested values as JSON text. Buffers are not compressed. MatrixTests.path (xlsx → arrow) —
xlsx csv — data supported partial tables Only cell values are converted. Formulas (their last calculated values are kept), styles, number formats, column widths, charts, images, pivot tables, comments, data validation and conditional formatting are not carried over. Merged cells keep their value in the first cell only. One sheet per output file: the first one by default, a chosen one, or all of them as separate files. Text formats carry no types: dates are written in ISO 8601, binary values as Base64, nested values as JSON text. MatrixTests.path (xlsx → csv) 4.55 s (time; XLSX (1 M rows) → CSV)
xlsx duckdb — data supported partial multipleSheets, tables Only cell values are converted. Formulas (their last calculated values are kept), styles, number formats, column widths, charts, images, pivot tables, comments, data validation and conditional formatting are not carried over. Merged cells keep their value in the first cell only. MatrixTests.path (xlsx → duckdb) —
xlsx json — data supported partial tables Only cell values are converted. Formulas (their last calculated values are kept), styles, number formats, column widths, charts, images, pivot tables, comments, data validation and conditional formatting are not carried over. Merged cells keep their value in the first cell only. One sheet per output file: the first one by default, a chosen one, or all of them as separate files. JSON has no date, binary, NaN or infinity: dates become ISO 8601 text, binary values Base64 text, NaN and infinities null. MatrixTests.path (xlsx → json) —
xlsx jsonl — data supported partial tables Only cell values are converted. Formulas (their last calculated values are kept), styles, number formats, column widths, charts, images, pivot tables, comments, data validation and conditional formatting are not carried over. Merged cells keep their value in the first cell only. One sheet per output file: the first one by default, a chosen one, or all of them as separate files. JSON has no date, binary, NaN or infinity: dates become ISO 8601 text, binary values Base64 text, NaN and infinities null. MatrixTests.path (xlsx → jsonl) —
xlsx ods — data supported partial multipleSheets, tables Only cell values are converted. Formulas (their last calculated values are kept), styles, number formats, column widths, charts, images, pivot tables, comments, data validation and conditional formatting are not carried over. Merged cells keep their value in the first cell only. Spreadsheet numbers are 64-bit floats: integers beyond 2^53 and long decimals are written as text; NaN and infinities become empty cells; time-zoned instants are written as UTC wall-clock times. At most 1,048,576 rows and 16,384 columns per sheet (larger tables stop with an error, or are split across sheets on request). MatrixTests.path (xlsx → ods) —
xlsx parquet — data supported partial tables Only cell values are converted. Formulas (their last calculated values are kept), styles, number formats, column widths, charts, images, pivot tables, comments, data validation and conditional formatting are not carried over. Merged cells keep their value in the first cell only. One sheet per output file: the first one by default, a chosen one, or all of them as separate files. MatrixTests.path (xlsx → parquet) —
xlsx plist — data supported partial tables Only cell values are converted. Formulas (their last calculated values are kept), styles, number formats, column widths, charts, images, pivot tables, comments, data validation and conditional formatting are not carried over. Merged cells keep their value in the first cell only. One sheet per output file: the first one by default, a chosen one, or all of them as separate files. Property lists have no null (empty values are omitted or stop the conversion) and are built in memory (20 million cells at most). MatrixTests.path (xlsx → plist) —
xlsx sqlite — data supported partial multipleSheets, tables Only cell values are converted. Formulas (their last calculated values are kept), styles, number formats, column widths, charts, images, pivot tables, comments, data validation and conditional formatting are not carried over. Merged cells keep their value in the first cell only. SQLite numbers are 64-bit: exact decimals and 128-bit integers are stored as TEXT; dates and times as ISO 8601 text in DATE / DATETIME / TIME columns. MatrixTests.path (xlsx → sqlite) —
xlsx toml — data supported partial tables Only cell values are converted. Formulas (their last calculated values are kept), styles, number formats, column widths, charts, images, pivot tables, comments, data validation and conditional formatting are not carried over. Merged cells keep their value in the first cell only. One sheet per output file: the first one by default, a chosen one, or all of them as separate files. TOML has no null: empty values are omitted (or stop the conversion, as chosen). MatrixTests.path (xlsx → toml) —
xlsx tsv — data supported partial tables Only cell values are converted. Formulas (their last calculated values are kept), styles, number formats, column widths, charts, images, pivot tables, comments, data validation and conditional formatting are not carried over. Merged cells keep their value in the first cell only. One sheet per output file: the first one by default, a chosen one, or all of them as separate files. Text formats carry no types: dates are written in ISO 8601, binary values as Base64, nested values as JSON text. MatrixTests.path (xlsx → tsv) —
xlsx xls — data supported partial multipleSheets, tables Only cell values are converted. Formulas (their last calculated values are kept), styles, number formats, column widths, charts, images, pivot tables, comments, data validation and conditional formatting are not carried over. Merged cells keep their value in the first cell only. Spreadsheet numbers are 64-bit floats: integers beyond 2^53 and long decimals are written as text; NaN and infinities become empty cells; time-zoned instants are written as UTC wall-clock times. Date-times are stored to the millisecond. At most 65,536 rows and 256 columns per sheet (legacy format limit); produced through the bundled LibreOffice. MatrixTests.path (xlsx → xls) —
xlsx xml — data supported partial tables Only cell values are converted. Formulas (their last calculated values are kept), styles, number formats, column widths, charts, images, pivot tables, comments, data validation and conditional formatting are not carried over. Merged cells keep their value in the first cell only. One sheet per output file: the first one by default, a chosen one, or all of them as separate files. XML carries text only; empty values are omitted, so empty text and missing values cannot be told apart when read back. MatrixTests.path (xlsx → xml) —
xlsx yaml — data supported partial tables Only cell values are converted. Formulas (their last calculated values are kept), styles, number formats, column widths, charts, images, pivot tables, comments, data validation and conditional formatting are not carried over. Merged cells keep their value in the first cell only. One sheet per output file: the first one by default, a chosen one, or all of them as separate files. Binary values are written as Base64 text. MatrixTests.path (xlsx → yaml) —

# data sources (182 paths)

Source Destination Operation Engine Status Fidelity Preserves Limitations Tests Performance
arrow csv — data supported structural tables ZSTD-compressed Arrow buffers, union, run-end-encoded and view types are not supported. Nested columns are read as JSON text; unsigned 64-bit integers become exact decimals. Text formats carry no types: dates are written in ISO 8601, binary values as Base64, nested values as JSON text. MatrixTests.path (arrow → csv) —
arrow duckdb — data supported structural tables ZSTD-compressed Arrow buffers, union, run-end-encoded and view types are not supported. Nested columns are read as JSON text; unsigned 64-bit integers become exact decimals. MatrixTests.path (arrow → duckdb) —
arrow json — data supported structural tables ZSTD-compressed Arrow buffers, union, run-end-encoded and view types are not supported. Nested columns are read as JSON text; unsigned 64-bit integers become exact decimals. JSON has no date, binary, NaN or infinity: dates become ISO 8601 text, binary values Base64 text, NaN and infinities null. MatrixTests.path (arrow → json) —
arrow jsonl — data supported structural tables ZSTD-compressed Arrow buffers, union, run-end-encoded and view types are not supported. Nested columns are read as JSON text; unsigned 64-bit integers become exact decimals. JSON has no date, binary, NaN or infinity: dates become ISO 8601 text, binary values Base64 text, NaN and infinities null. MatrixTests.path (arrow → jsonl) —
arrow ods — data supported structural tables ZSTD-compressed Arrow buffers, union, run-end-encoded and view types are not supported. Nested columns are read as JSON text; unsigned 64-bit integers become exact decimals. Spreadsheet numbers are 64-bit floats: integers beyond 2^53 and long decimals are written as text; NaN and infinities become empty cells; time-zoned instants are written as UTC wall-clock times. At most 1,048,576 rows and 16,384 columns per sheet (larger tables stop with an error, or are split across sheets on request). MatrixTests.path (arrow → ods) —
arrow parquet — data supported structural tables ZSTD-compressed Arrow buffers, union, run-end-encoded and view types are not supported. Nested columns are read as JSON text; unsigned 64-bit integers become exact decimals. MatrixTests.path (arrow → parquet) —
arrow plist — data supported structural tables ZSTD-compressed Arrow buffers, union, run-end-encoded and view types are not supported. Nested columns are read as JSON text; unsigned 64-bit integers become exact decimals. Property lists have no null (empty values are omitted or stop the conversion) and are built in memory (20 million cells at most). MatrixTests.path (arrow → plist) —
arrow sqlite — data supported structural tables ZSTD-compressed Arrow buffers, union, run-end-encoded and view types are not supported. Nested columns are read as JSON text; unsigned 64-bit integers become exact decimals. SQLite numbers are 64-bit: exact decimals and 128-bit integers are stored as TEXT; dates and times as ISO 8601 text in DATE / DATETIME / TIME columns. MatrixTests.path (arrow → sqlite) —
arrow toml — data supported structural tables ZSTD-compressed Arrow buffers, union, run-end-encoded and view types are not supported. Nested columns are read as JSON text; unsigned 64-bit integers become exact decimals. TOML has no null: empty values are omitted (or stop the conversion, as chosen). MatrixTests.path (arrow → toml) —
arrow tsv — data supported structural tables ZSTD-compressed Arrow buffers, union, run-end-encoded and view types are not supported. Nested columns are read as JSON text; unsigned 64-bit integers become exact decimals. Text formats carry no types: dates are written in ISO 8601, binary values as Base64, nested values as JSON text. MatrixTests.path (arrow → tsv) —
arrow xls — data supported structural tables ZSTD-compressed Arrow buffers, union, run-end-encoded and view types are not supported. Nested columns are read as JSON text; unsigned 64-bit integers become exact decimals. Spreadsheet numbers are 64-bit floats: integers beyond 2^53 and long decimals are written as text; NaN and infinities become empty cells; time-zoned instants are written as UTC wall-clock times. Date-times are stored to the millisecond. At most 65,536 rows and 256 columns per sheet (legacy format limit); produced through the bundled LibreOffice. MatrixTests.path (arrow → xls) —
arrow xlsx — data supported structural tables ZSTD-compressed Arrow buffers, union, run-end-encoded and view types are not supported. Nested columns are read as JSON text; unsigned 64-bit integers become exact decimals. Spreadsheet numbers are 64-bit floats: integers beyond 2^53 and long decimals are written as text; NaN and infinities become empty cells; time-zoned instants are written as UTC wall-clock times. Date-times are stored to the millisecond. At most 1,048,576 rows and 16,384 columns per sheet (larger tables stop with an error, or are split across sheets on request). MatrixTests.path (arrow → xlsx) —
arrow xml — data supported structural tables ZSTD-compressed Arrow buffers, union, run-end-encoded and view types are not supported. Nested columns are read as JSON text; unsigned 64-bit integers become exact decimals. XML carries text only; empty values are omitted, so empty text and missing values cannot be told apart when read back. MatrixTests.path (arrow → xml) —
arrow yaml — data supported structural tables ZSTD-compressed Arrow buffers, union, run-end-encoded and view types are not supported. Nested columns are read as JSON text; unsigned 64-bit integers become exact decimals. Binary values are written as Base64 text. MatrixTests.path (arrow → yaml) —
csv arrow — data supported structural tables Column types are detected from the text (verified on every row); numbers are re-written in canonical form (1.50 → 1.5). Identifiers with leading zeros stay text. Turn detection off to keep every cell as written. Columns are written as int64, float64, bool, utf8, binary, date32, time64[us], timestamp[us] and decimal128; nested values as JSON text. Buffers are not compressed. MatrixTests.path (csv → arrow) 13.43 s (time; CSV (2.36 GB) → Arrow IPC file)
csv csv — data supported structural tables Column types are detected from the text (verified on every row); numbers are re-written in canonical form (1.50 → 1.5). Identifiers with leading zeros stay text. Turn detection off to keep every cell as written. Text formats carry no types: dates are written in ISO 8601, binary values as Base64, nested values as JSON text. Same-format conversion: the file is re-written with the chosen options. MatrixTests.path (csv → csv) —
csv duckdb — data supported structural tables Column types are detected from the text (verified on every row); numbers are re-written in canonical form (1.50 → 1.5). Identifiers with leading zeros stay text. Turn detection off to keep every cell as written. MatrixTests.path (csv → duckdb) —
csv geojson — geo supported structural attributes, vectorGeometry CSV does not record a reference system: coordinates are assumed to be WGS 84 only when the columns are named like longitude / latitude; otherwise state it in the settings. GeoJSON is always WGS 84 longitude / latitude: sources in WGS 84 or NAD83 geographic, Web Mercator, UTM or MTM coordinates are reprojected; any other reference system is refused. Measure (M) values are dropped. PathTests.convertsCities(_😃 —
csv gpkg — geo supported structural attributes, coordinateSystem, vectorGeometry CSV does not record a reference system: coordinates are assumed to be WGS 84 only when the columns are named like longitude / latitude; otherwise state it in the settings. PathTests.convertsCities(_😃 —
csv gpx — geo supported partial vectorGeometry CSV does not record a reference system: coordinates are assumed to be WGS 84 only when the columns are named like longitude / latitude; otherwise state it in the settings. GPX is always WGS 84 and holds only points and lines: polygons are refused unless written as closed tracks. Only the fixed GPX fields (name, desc, cmt, src, sym, type, time, link…) are kept; other attributes are dropped and counted. PathTests.convertsCities(_😃 —
csv json — data supported structural tables Column types are detected from the text (verified on every row); numbers are re-written in canonical form (1.50 → 1.5). Identifiers with leading zeros stay text. Turn detection off to keep every cell as written. JSON has no date, binary, NaN or infinity: dates become ISO 8601 text, binary values Base64 text, NaN and infinities null. MatrixTests.path (csv → json) —
csv jsonl — data supported structural tables Column types are detected from the text (verified on every row); numbers are re-written in canonical form (1.50 → 1.5). Identifiers with leading zeros stay text. Turn detection off to keep every cell as written. JSON has no date, binary, NaN or infinity: dates become ISO 8601 text, binary values Base64 text, NaN and infinities null. MatrixTests.path (csv → jsonl) —
csv kml — geo supported structural attributes, vectorGeometry CSV does not record a reference system: coordinates are assumed to be WGS 84 only when the columns are named like longitude / latitude; otherwise state it in the settings. KML is always WGS 84 longitude / latitude: sources in WGS 84 or NAD83 geographic, Web Mercator, UTM or MTM coordinates are reprojected; any other reference system is refused. Measure (M) values are dropped. Dates are written as text. PathTests.convertsCities(_😃 —
csv ods — data supported structural tables Column types are detected from the text (verified on every row); numbers are re-written in canonical form (1.50 → 1.5). Identifiers with leading zeros stay text. Turn detection off to keep every cell as written. Spreadsheet numbers are 64-bit floats: integers beyond 2^53 and long decimals are written as text; NaN and infinities become empty cells; time-zoned instants are written as UTC wall-clock times. At most 1,048,576 rows and 16,384 columns per sheet (larger tables stop with an error, or are split across sheets on request). MatrixTests.path (csv → ods) 5.43 s (time; CSV → ODS (1 M rows))
csv parquet — data supported structural tables Column types are detected from the text (verified on every row); numbers are re-written in canonical form (1.50 → 1.5). Identifiers with leading zeros stay text. Turn detection off to keep every cell as written. MatrixTests.path (csv → parquet) 6.90 s (time; CSV (2.36 GB) → Parquet, zstd)
csv plist — data supported structural tables Column types are detected from the text (verified on every row); numbers are re-written in canonical form (1.50 → 1.5). Identifiers with leading zeros stay text. Turn detection off to keep every cell as written. Property lists have no null (empty values are omitted or stop the conversion) and are built in memory (20 million cells at most). MatrixTests.path (csv → plist) —
csv shp — geo supported structural attributes, coordinateSystem, vectorGeometry CSV does not record a reference system: coordinates are assumed to be WGS 84 only when the columns are named like longitude / latitude; otherwise state it in the settings. The result is a folder holding the .shp, .shx, .dbf, .cpg and .prj files. dBASE limits apply: attribute names of 10 characters, text of 254 bytes, no date-and-time type, no nested values. One geometry type per shapefile: mixed layers are split. PathTests.convertsCities(_😃 —
csv sqlite — data supported structural tables Column types are detected from the text (verified on every row); numbers are re-written in canonical form (1.50 → 1.5). Identifiers with leading zeros stay text. Turn detection off to keep every cell as written. SQLite numbers are 64-bit: exact decimals and 128-bit integers are stored as TEXT; dates and times as ISO 8601 text in DATE / DATETIME / TIME columns. MatrixTests.path (csv → sqlite) 33.07 s (time; CSV (2.36 GB) → SQLite)
csv toml — data supported structural tables Column types are detected from the text (verified on every row); numbers are re-written in canonical form (1.50 → 1.5). Identifiers with leading zeros stay text. Turn detection off to keep every cell as written. TOML has no null: empty values are omitted (or stop the conversion, as chosen). MatrixTests.path (csv → toml) —
csv tsv — data supported structural tables Column types are detected from the text (verified on every row); numbers are re-written in canonical form (1.50 → 1.5). Identifiers with leading zeros stay text. Turn detection off to keep every cell as written. Text formats carry no types: dates are written in ISO 8601, binary values as Base64, nested values as JSON text. MatrixTests.path (csv → tsv) —
csv xls — data supported structural tables Column types are detected from the text (verified on every row); numbers are re-written in canonical form (1.50 → 1.5). Identifiers with leading zeros stay text. Turn detection off to keep every cell as written. Spreadsheet numbers are 64-bit floats: integers beyond 2^53 and long decimals are written as text; NaN and infinities become empty cells; time-zoned instants are written as UTC wall-clock times. Date-times are stored to the millisecond. At most 65,536 rows and 256 columns per sheet (legacy format limit); produced through the bundled LibreOffice. MatrixTests.path (csv → xls) —
csv xlsx — data supported structural tables Column types are detected from the text (verified on every row); numbers are re-written in canonical form (1.50 → 1.5). Identifiers with leading zeros stay text. Turn detection off to keep every cell as written. Spreadsheet numbers are 64-bit floats: integers beyond 2^53 and long decimals are written as text; NaN and infinities become empty cells; time-zoned instants are written as UTC wall-clock times. Date-times are stored to the millisecond. At most 1,048,576 rows and 16,384 columns per sheet (larger tables stop with an error, or are split across sheets on request). MatrixTests.path (csv → xlsx) 6.16 s (time; CSV → XLSX (1 M rows, shared strings))
csv xml — data supported structural tables Column types are detected from the text (verified on every row); numbers are re-written in canonical form (1.50 → 1.5). Identifiers with leading zeros stay text. Turn detection off to keep every cell as written. XML carries text only; empty values are omitted, so empty text and missing values cannot be told apart when read back. MatrixTests.path (csv → xml) —
csv yaml — data supported structural tables Column types are detected from the text (verified on every row); numbers are re-written in canonical form (1.50 → 1.5). Identifiers with leading zeros stay text. Turn detection off to keep every cell as written. Binary values are written as Base64 text. MatrixTests.path (csv → yaml) —
duckdb arrow — data supported partial tables One table per output file: the first one by default, a chosen one, or all of them as separate files. Only tables are read. Views, indexes, constraints and triggers are not carried over; stored SQL is never executed. Columns are written as int64, float64, bool, utf8, binary, date32, time64[us], timestamp[us] and decimal128; nested values as JSON text. Buffers are not compressed. MatrixTests.path (duckdb → arrow) —
duckdb csv — data supported partial tables One table per output file: the first one by default, a chosen one, or all of them as separate files. Only tables are read. Views, indexes, constraints and triggers are not carried over; stored SQL is never executed. Text formats carry no types: dates are written in ISO 8601, binary values as Base64, nested values as JSON text. MatrixTests.path (duckdb → csv) —
duckdb json — data supported partial tables One table per output file: the first one by default, a chosen one, or all of them as separate files. Only tables are read. Views, indexes, constraints and triggers are not carried over; stored SQL is never executed. JSON has no date, binary, NaN or infinity: dates become ISO 8601 text, binary values Base64 text, NaN and infinities null. MatrixTests.path (duckdb → json) —
duckdb jsonl — data supported partial tables One table per output file: the first one by default, a chosen one, or all of them as separate files. Only tables are read. Views, indexes, constraints and triggers are not carried over; stored SQL is never executed. JSON has no date, binary, NaN or infinity: dates become ISO 8601 text, binary values Base64 text, NaN and infinities null. MatrixTests.path (duckdb → jsonl) —
duckdb ods — data supported structural multipleSheets, tables Only tables are read. Views, indexes, constraints and triggers are not carried over; stored SQL is never executed. Spreadsheet numbers are 64-bit floats: integers beyond 2^53 and long decimals are written as text; NaN and infinities become empty cells; time-zoned instants are written as UTC wall-clock times. At most 1,048,576 rows and 16,384 columns per sheet (larger tables stop with an error, or are split across sheets on request). MatrixTests.path (duckdb → ods) —
duckdb parquet — data supported partial tables One table per output file: the first one by default, a chosen one, or all of them as separate files. Only tables are read. Views, indexes, constraints and triggers are not carried over; stored SQL is never executed. MatrixTests.path (duckdb → parquet) —
duckdb plist — data supported partial tables One table per output file: the first one by default, a chosen one, or all of them as separate files. Only tables are read. Views, indexes, constraints and triggers are not carried over; stored SQL is never executed. Property lists have no null (empty values are omitted or stop the conversion) and are built in memory (20 million cells at most). MatrixTests.path (duckdb → plist) —
duckdb sqlite — data supported structural multipleSheets, tables Only tables are read. Views, indexes, constraints and triggers are not carried over; stored SQL is never executed. SQLite numbers are 64-bit: exact decimals and 128-bit integers are stored as TEXT; dates and times as ISO 8601 text in DATE / DATETIME / TIME columns. MatrixTests.path (duckdb → sqlite) —
duckdb toml — data supported partial tables One table per output file: the first one by default, a chosen one, or all of them as separate files. Only tables are read. Views, indexes, constraints and triggers are not carried over; stored SQL is never executed. TOML has no null: empty values are omitted (or stop the conversion, as chosen). MatrixTests.path (duckdb → toml) —
duckdb tsv — data supported partial tables One table per output file: the first one by default, a chosen one, or all of them as separate files. Only tables are read. Views, indexes, constraints and triggers are not carried over; stored SQL is never executed. Text formats carry no types: dates are written in ISO 8601, binary values as Base64, nested values as JSON text. MatrixTests.path (duckdb → tsv) —
duckdb xls — data supported structural multipleSheets, tables Only tables are read. Views, indexes, constraints and triggers are not carried over; stored SQL is never executed. Spreadsheet numbers are 64-bit floats: integers beyond 2^53 and long decimals are written as text; NaN and infinities become empty cells; time-zoned instants are written as UTC wall-clock times. Date-times are stored to the millisecond. At most 65,536 rows and 256 columns per sheet (legacy format limit); produced through the bundled LibreOffice. MatrixTests.path (duckdb → xls) —
duckdb xlsx — data supported structural multipleSheets, tables Only tables are read. Views, indexes, constraints and triggers are not carried over; stored SQL is never executed. Spreadsheet numbers are 64-bit floats: integers beyond 2^53 and long decimals are written as text; NaN and infinities become empty cells; time-zoned instants are written as UTC wall-clock times. Date-times are stored to the millisecond. At most 1,048,576 rows and 16,384 columns per sheet (larger tables stop with an error, or are split across sheets on request). MatrixTests.path (duckdb → xlsx) —
duckdb xml — data supported partial tables One table per output file: the first one by default, a chosen one, or all of them as separate files. Only tables are read. Views, indexes, constraints and triggers are not carried over; stored SQL is never executed. XML carries text only; empty values are omitted, so empty text and missing values cannot be told apart when read back. MatrixTests.path (duckdb → xml) —
duckdb yaml — data supported partial tables One table per output file: the first one by default, a chosen one, or all of them as separate files. Only tables are read. Views, indexes, constraints and triggers are not carried over; stored SQL is never executed. Binary values are written as Base64 text. MatrixTests.path (duckdb → yaml) —
json arrow — data supported structural tables Integers beyond 64 bits are read as 64-bit floats by the JSON reader. Text that looks like an ISO date becomes a date. The table is built from the document's list of records (detected, or given by “Record list”); other content of the document is not converted. Columns are written as int64, float64, bool, utf8, binary, date32, time64[us], timestamp[us] and decimal128; nested values as JSON text. Buffers are not compressed. MatrixTests.path (json → arrow) —
json csv — data supported structural tables Integers beyond 64 bits are read as 64-bit floats by the JSON reader. Text that looks like an ISO date becomes a date. The table is built from the document's list of records (detected, or given by “Record list”); other content of the document is not converted. Text formats carry no types: dates are written in ISO 8601, binary values as Base64, nested values as JSON text. MatrixTests.path (json → csv) —
json duckdb — data supported structural tables Integers beyond 64 bits are read as 64-bit floats by the JSON reader. Text that looks like an ISO date becomes a date. The table is built from the document's list of records (detected, or given by “Record list”); other content of the document is not converted. MatrixTests.path (json → duckdb) —
json json — data supported lossless — — TreeTests.path (json → json) —
json jsonl — data supported structural tables Integers beyond 64 bits are read as 64-bit floats by the JSON reader. Text that looks like an ISO date becomes a date. The table is built from the document's list of records (detected, or given by “Record list”); other content of the document is not converted. JSON has no date, binary, NaN or infinity: dates become ISO 8601 text, binary values Base64 text, NaN and infinities null. MatrixTests.path (json → jsonl) —
json ods — data supported structural tables Integers beyond 64 bits are read as 64-bit floats by the JSON reader. Text that looks like an ISO date becomes a date. The table is built from the document's list of records (detected, or given by “Record list”); other content of the document is not converted. Spreadsheet numbers are 64-bit floats: integers beyond 2^53 and long decimals are written as text; NaN and infinities become empty cells; time-zoned instants are written as UTC wall-clock times. At most 1,048,576 rows and 16,384 columns per sheet (larger tables stop with an error, or are split across sheets on request). MatrixTests.path (json → ods) —
json parquet — data supported structural tables Integers beyond 64 bits are read as 64-bit floats by the JSON reader. Text that looks like an ISO date becomes a date. The table is built from the document's list of records (detected, or given by “Record list”); other content of the document is not converted. MatrixTests.path (json → parquet) —
json plist — data supported structural — Property lists have no null and no key order (keys are sorted when read back); integers beyond 64 bits and exact decimals are written as text. TreeTests.path (json → plist) —
json sqlite — data supported structural tables Integers beyond 64 bits are read as 64-bit floats by the JSON reader. Text that looks like an ISO date becomes a date. The table is built from the document's list of records (detected, or given by “Record list”); other content of the document is not converted. SQLite numbers are 64-bit: exact decimals and 128-bit integers are stored as TEXT; dates and times as ISO 8601 text in DATE / DATETIME / TIME columns. MatrixTests.path (json → sqlite) —
json toml — data supported structural — TOML has no null (nulls are omitted or stop the conversion, as chosen) and needs a table at the top level; plain values are moved before sub-tables. TreeTests.path (json → toml) —
json tsv — data supported structural tables Integers beyond 64 bits are read as 64-bit floats by the JSON reader. Text that looks like an ISO date becomes a date. The table is built from the document's list of records (detected, or given by “Record list”); other content of the document is not converted. Text formats carry no types: dates are written in ISO 8601, binary values as Base64, nested values as JSON text. MatrixTests.path (json → tsv) —
json xls — data supported structural tables Integers beyond 64 bits are read as 64-bit floats by the JSON reader. Text that looks like an ISO date becomes a date. The table is built from the document's list of records (detected, or given by “Record list”); other content of the document is not converted. Spreadsheet numbers are 64-bit floats: integers beyond 2^53 and long decimals are written as text; NaN and infinities become empty cells; time-zoned instants are written as UTC wall-clock times. Date-times are stored to the millisecond. At most 65,536 rows and 256 columns per sheet (legacy format limit); produced through the bundled LibreOffice. MatrixTests.path (json → xls) —
json xlsx — data supported structural tables Integers beyond 64 bits are read as 64-bit floats by the JSON reader. Text that looks like an ISO date becomes a date. The table is built from the document's list of records (detected, or given by “Record list”); other content of the document is not converted. Spreadsheet numbers are 64-bit floats: integers beyond 2^53 and long decimals are written as text; NaN and infinities become empty cells; time-zoned instants are written as UTC wall-clock times. Date-times are stored to the millisecond. At most 1,048,576 rows and 16,384 columns per sheet (larger tables stop with an error, or are split across sheets on request). MatrixTests.path (json → xlsx) —
json xml — data supported structural — XML carries text only: numbers and booleans become text, empty strings and nulls both become empty elements. Keys starting with “@” become attributes, “#text” becomes element text, lists become repeated elements. TreeTests.path (json → xml) —
json yaml — data supported lossless — — TreeTests.path (json → yaml) —
jsonl arrow — data supported structural tables Integers beyond 64 bits are read as 64-bit floats by the JSON reader. Text that looks like an ISO date becomes a date. Columns are written as int64, float64, bool, utf8, binary, date32, time64[us], timestamp[us] and decimal128; nested values as JSON text. Buffers are not compressed. MatrixTests.path (jsonl → arrow) —
jsonl csv — data supported structural tables Integers beyond 64 bits are read as 64-bit floats by the JSON reader. Text that looks like an ISO date becomes a date. Text formats carry no types: dates are written in ISO 8601, binary values as Base64, nested values as JSON text. MatrixTests.path (jsonl → csv) —
jsonl duckdb — data supported structural tables Integers beyond 64 bits are read as 64-bit floats by the JSON reader. Text that looks like an ISO date becomes a date. MatrixTests.path (jsonl → duckdb) —
jsonl json — data supported structural tables Integers beyond 64 bits are read as 64-bit floats by the JSON reader. Text that looks like an ISO date becomes a date. JSON has no date, binary, NaN or infinity: dates become ISO 8601 text, binary values Base64 text, NaN and infinities null. MatrixTests.path (jsonl → json) —
jsonl jsonl — data supported structural tables Integers beyond 64 bits are read as 64-bit floats by the JSON reader. Text that looks like an ISO date becomes a date. JSON has no date, binary, NaN or infinity: dates become ISO 8601 text, binary values Base64 text, NaN and infinities null. Same-format conversion: the file is re-written with the chosen options. MatrixTests.path (jsonl → jsonl) —
jsonl ods — data supported structural tables Integers beyond 64 bits are read as 64-bit floats by the JSON reader. Text that looks like an ISO date becomes a date. Spreadsheet numbers are 64-bit floats: integers beyond 2^53 and long decimals are written as text; NaN and infinities become empty cells; time-zoned instants are written as UTC wall-clock times. At most 1,048,576 rows and 16,384 columns per sheet (larger tables stop with an error, or are split across sheets on request). MatrixTests.path (jsonl → ods) —
jsonl parquet — data supported structural tables Integers beyond 64 bits are read as 64-bit floats by the JSON reader. Text that looks like an ISO date becomes a date. MatrixTests.path (jsonl → parquet) 12.99 s (time; JSON Lines (1.45 GB) → Parquet, zstd)
jsonl plist — data supported structural tables Integers beyond 64 bits are read as 64-bit floats by the JSON reader. Text that looks like an ISO date becomes a date. Property lists have no null (empty values are omitted or stop the conversion) and are built in memory (20 million cells at most). MatrixTests.path (jsonl → plist) —
jsonl sqlite — data supported structural tables Integers beyond 64 bits are read as 64-bit floats by the JSON reader. Text that looks like an ISO date becomes a date. SQLite numbers are 64-bit: exact decimals and 128-bit integers are stored as TEXT; dates and times as ISO 8601 text in DATE / DATETIME / TIME columns. MatrixTests.path (jsonl → sqlite) —
jsonl toml — data supported structural tables Integers beyond 64 bits are read as 64-bit floats by the JSON reader. Text that looks like an ISO date becomes a date. TOML has no null: empty values are omitted (or stop the conversion, as chosen). MatrixTests.path (jsonl → toml) —
jsonl tsv — data supported structural tables Integers beyond 64 bits are read as 64-bit floats by the JSON reader. Text that looks like an ISO date becomes a date. Text formats carry no types: dates are written in ISO 8601, binary values as Base64, nested values as JSON text. MatrixTests.path (jsonl → tsv) —
jsonl xls — data supported structural tables Integers beyond 64 bits are read as 64-bit floats by the JSON reader. Text that looks like an ISO date becomes a date. Spreadsheet numbers are 64-bit floats: integers beyond 2^53 and long decimals are written as text; NaN and infinities become empty cells; time-zoned instants are written as UTC wall-clock times. Date-times are stored to the millisecond. At most 65,536 rows and 256 columns per sheet (legacy format limit); produced through the bundled LibreOffice. MatrixTests.path (jsonl → xls) —
jsonl xlsx — data supported structural tables Integers beyond 64 bits are read as 64-bit floats by the JSON reader. Text that looks like an ISO date becomes a date. Spreadsheet numbers are 64-bit floats: integers beyond 2^53 and long decimals are written as text; NaN and infinities become empty cells; time-zoned instants are written as UTC wall-clock times. Date-times are stored to the millisecond. At most 1,048,576 rows and 16,384 columns per sheet (larger tables stop with an error, or are split across sheets on request). MatrixTests.path (jsonl → xlsx) —
jsonl xml — data supported structural tables Integers beyond 64 bits are read as 64-bit floats by the JSON reader. Text that looks like an ISO date becomes a date. XML carries text only; empty values are omitted, so empty text and missing values cannot be told apart when read back. MatrixTests.path (jsonl → xml) —
jsonl yaml — data supported structural tables Integers beyond 64 bits are read as 64-bit floats by the JSON reader. Text that looks like an ISO date becomes a date. Binary values are written as Base64 text. MatrixTests.path (jsonl → yaml) —
parquet arrow — data supported structural tables Columns are written as int64, float64, bool, utf8, binary, date32, time64[us], timestamp[us] and decimal128; nested values as JSON text. Buffers are not compressed. MatrixTests.path (parquet → arrow) —
parquet csv — data supported structural tables Text formats carry no types: dates are written in ISO 8601, binary values as Base64, nested values as JSON text. MatrixTests.path (parquet → csv) 3.10 s (time; Parquet (zstd) → CSV)
parquet duckdb — data supported lossless tables — MatrixTests.path (parquet → duckdb) —
parquet json — data supported structural tables JSON has no date, binary, NaN or infinity: dates become ISO 8601 text, binary values Base64 text, NaN and infinities null. MatrixTests.path (parquet → json) —
parquet jsonl — data supported structural tables JSON has no date, binary, NaN or infinity: dates become ISO 8601 text, binary values Base64 text, NaN and infinities null. MatrixTests.path (parquet → jsonl) —
parquet ods — data supported structural tables Spreadsheet numbers are 64-bit floats: integers beyond 2^53 and long decimals are written as text; NaN and infinities become empty cells; time-zoned instants are written as UTC wall-clock times. At most 1,048,576 rows and 16,384 columns per sheet (larger tables stop with an error, or are split across sheets on request). MatrixTests.path (parquet → ods) —
parquet parquet — data supported lossless tables Same-format conversion: the file is re-written with the chosen options. MatrixTests.path (parquet → parquet) —
parquet plist — data supported structural tables Property lists have no null (empty values are omitted or stop the conversion) and are built in memory (20 million cells at most). MatrixTests.path (parquet → plist) —
parquet sqlite — data supported structural tables SQLite numbers are 64-bit: exact decimals and 128-bit integers are stored as TEXT; dates and times as ISO 8601 text in DATE / DATETIME / TIME columns. MatrixTests.path (parquet → sqlite) —
parquet toml — data supported structural tables TOML has no null: empty values are omitted (or stop the conversion, as chosen). MatrixTests.path (parquet → toml) —
parquet tsv — data supported structural tables Text formats carry no types: dates are written in ISO 8601, binary values as Base64, nested values as JSON text. MatrixTests.path (parquet → tsv) —
parquet xls — data supported structural tables Spreadsheet numbers are 64-bit floats: integers beyond 2^53 and long decimals are written as text; NaN and infinities become empty cells; time-zoned instants are written as UTC wall-clock times. Date-times are stored to the millisecond. At most 65,536 rows and 256 columns per sheet (legacy format limit); produced through the bundled LibreOffice. MatrixTests.path (parquet → xls) —
parquet xlsx — data supported structural tables Spreadsheet numbers are 64-bit floats: integers beyond 2^53 and long decimals are written as text; NaN and infinities become empty cells; time-zoned instants are written as UTC wall-clock times. Date-times are stored to the millisecond. At most 1,048,576 rows and 16,384 columns per sheet (larger tables stop with an error, or are split across sheets on request). MatrixTests.path (parquet → xlsx) —
parquet xml — data supported structural tables XML carries text only; empty values are omitted, so empty text and missing values cannot be told apart when read back. MatrixTests.path (parquet → xml) —
parquet yaml — data supported structural tables Binary values are written as Base64 text. MatrixTests.path (parquet → yaml) —
plist arrow — data supported structural tables The table is built from the document's list of records (detected, or given by “Record list”); other content of the document is not converted. Columns are written as int64, float64, bool, utf8, binary, date32, time64[us], timestamp[us] and decimal128; nested values as JSON text. Buffers are not compressed. MatrixTests.path (plist → arrow) —
plist csv — data supported structural tables The table is built from the document's list of records (detected, or given by “Record list”); other content of the document is not converted. Text formats carry no types: dates are written in ISO 8601, binary values as Base64, nested values as JSON text. MatrixTests.path (plist → csv) —
plist duckdb — data supported structural tables The table is built from the document's list of records (detected, or given by “Record list”); other content of the document is not converted. MatrixTests.path (plist → duckdb) —
plist json — data supported structural — Dates and binary values become text (ISO 8601 / Base64); JSON has no such types. TreeTests.path (plist → json) —
plist jsonl — data supported structural tables The table is built from the document's list of records (detected, or given by “Record list”); other content of the document is not converted. JSON has no date, binary, NaN or infinity: dates become ISO 8601 text, binary values Base64 text, NaN and infinities null. MatrixTests.path (plist → jsonl) —
plist ods — data supported structural tables The table is built from the document's list of records (detected, or given by “Record list”); other content of the document is not converted. Spreadsheet numbers are 64-bit floats: integers beyond 2^53 and long decimals are written as text; NaN and infinities become empty cells; time-zoned instants are written as UTC wall-clock times. At most 1,048,576 rows and 16,384 columns per sheet (larger tables stop with an error, or are split across sheets on request). MatrixTests.path (plist → ods) —
plist parquet — data supported structural tables The table is built from the document's list of records (detected, or given by “Record list”); other content of the document is not converted. MatrixTests.path (plist → parquet) —
plist plist — data supported lossless — Property lists have no null and no key order (keys are sorted when read back); integers beyond 64 bits and exact decimals are written as text. TreeTests.path (plist → plist) —
plist sqlite — data supported structural tables The table is built from the document's list of records (detected, or given by “Record list”); other content of the document is not converted. SQLite numbers are 64-bit: exact decimals and 128-bit integers are stored as TEXT; dates and times as ISO 8601 text in DATE / DATETIME / TIME columns. MatrixTests.path (plist → sqlite) —
plist toml — data supported structural — TOML has no null (nulls are omitted or stop the conversion, as chosen) and needs a table at the top level; plain values are moved before sub-tables. TreeTests.path (plist → toml) —
plist tsv — data supported structural tables The table is built from the document's list of records (detected, or given by “Record list”); other content of the document is not converted. Text formats carry no types: dates are written in ISO 8601, binary values as Base64, nested values as JSON text. MatrixTests.path (plist → tsv) —
plist xls — data supported structural tables The table is built from the document's list of records (detected, or given by “Record list”); other content of the document is not converted. Spreadsheet numbers are 64-bit floats: integers beyond 2^53 and long decimals are written as text; NaN and infinities become empty cells; time-zoned instants are written as UTC wall-clock times. Date-times are stored to the millisecond. At most 65,536 rows and 256 columns per sheet (legacy format limit); produced through the bundled LibreOffice. MatrixTests.path (plist → xls) —
plist xlsx — data supported structural tables The table is built from the document's list of records (detected, or given by “Record list”); other content of the document is not converted. Spreadsheet numbers are 64-bit floats: integers beyond 2^53 and long decimals are written as text; NaN and infinities become empty cells; time-zoned instants are written as UTC wall-clock times. Date-times are stored to the millisecond. At most 1,048,576 rows and 16,384 columns per sheet (larger tables stop with an error, or are split across sheets on request). MatrixTests.path (plist → xlsx) —
plist xml — data supported structural — XML carries text only: numbers and booleans become text, empty strings and nulls both become empty elements. Keys starting with “@” become attributes, “#text” becomes element text, lists become repeated elements. TreeTests.path (plist → xml) —
plist yaml — data supported structural — Comments, anchors and formatting of the source are not kept. TreeTests.path (plist → yaml) —
sqlite arrow — data supported partial tables One table per output file: the first one by default, a chosen one, or all of them as separate files. Only tables are read. Views, indexes, constraints and triggers are not carried over; stored SQL is never executed. Columns are written as int64, float64, bool, utf8, binary, date32, time64[us], timestamp[us] and decimal128; nested values as JSON text. Buffers are not compressed. MatrixTests.path (sqlite → arrow) —
sqlite csv — data supported partial tables One table per output file: the first one by default, a chosen one, or all of them as separate files. Only tables are read. Views, indexes, constraints and triggers are not carried over; stored SQL is never executed. Text formats carry no types: dates are written in ISO 8601, binary values as Base64, nested values as JSON text. MatrixTests.path (sqlite → csv) —
sqlite duckdb — data supported structural multipleSheets, tables Only tables are read. Views, indexes, constraints and triggers are not carried over; stored SQL is never executed. MatrixTests.path (sqlite → duckdb) —
sqlite json — data supported partial tables One table per output file: the first one by default, a chosen one, or all of them as separate files. Only tables are read. Views, indexes, constraints and triggers are not carried over; stored SQL is never executed. JSON has no date, binary, NaN or infinity: dates become ISO 8601 text, binary values Base64 text, NaN and infinities null. MatrixTests.path (sqlite → json) —
sqlite jsonl — data supported partial tables One table per output file: the first one by default, a chosen one, or all of them as separate files. Only tables are read. Views, indexes, constraints and triggers are not carried over; stored SQL is never executed. JSON has no date, binary, NaN or infinity: dates become ISO 8601 text, binary values Base64 text, NaN and infinities null. MatrixTests.path (sqlite → jsonl) —
sqlite ods — data supported structural multipleSheets, tables Only tables are read. Views, indexes, constraints and triggers are not carried over; stored SQL is never executed. Spreadsheet numbers are 64-bit floats: integers beyond 2^53 and long decimals are written as text; NaN and infinities become empty cells; time-zoned instants are written as UTC wall-clock times. At most 1,048,576 rows and 16,384 columns per sheet (larger tables stop with an error, or are split across sheets on request). MatrixTests.path (sqlite → ods) —
sqlite parquet — data supported partial tables One table per output file: the first one by default, a chosen one, or all of them as separate files. Only tables are read. Views, indexes, constraints and triggers are not carried over; stored SQL is never executed. MatrixTests.path (sqlite → parquet) —
sqlite plist — data supported partial tables One table per output file: the first one by default, a chosen one, or all of them as separate files. Only tables are read. Views, indexes, constraints and triggers are not carried over; stored SQL is never executed. Property lists have no null (empty values are omitted or stop the conversion) and are built in memory (20 million cells at most). MatrixTests.path (sqlite → plist) —
sqlite toml — data supported partial tables One table per output file: the first one by default, a chosen one, or all of them as separate files. Only tables are read. Views, indexes, constraints and triggers are not carried over; stored SQL is never executed. TOML has no null: empty values are omitted (or stop the conversion, as chosen). MatrixTests.path (sqlite → toml) —
sqlite tsv — data supported partial tables One table per output file: the first one by default, a chosen one, or all of them as separate files. Only tables are read. Views, indexes, constraints and triggers are not carried over; stored SQL is never executed. Text formats carry no types: dates are written in ISO 8601, binary values as Base64, nested values as JSON text. MatrixTests.path (sqlite → tsv) —
sqlite xls — data supported structural multipleSheets, tables Only tables are read. Views, indexes, constraints and triggers are not carried over; stored SQL is never executed. Spreadsheet numbers are 64-bit floats: integers beyond 2^53 and long decimals are written as text; NaN and infinities become empty cells; time-zoned instants are written as UTC wall-clock times. Date-times are stored to the millisecond. At most 65,536 rows and 256 columns per sheet (legacy format limit); produced through the bundled LibreOffice. MatrixTests.path (sqlite → xls) —
sqlite xlsx — data supported structural multipleSheets, tables Only tables are read. Views, indexes, constraints and triggers are not carried over; stored SQL is never executed. Spreadsheet numbers are 64-bit floats: integers beyond 2^53 and long decimals are written as text; NaN and infinities become empty cells; time-zoned instants are written as UTC wall-clock times. Date-times are stored to the millisecond. At most 1,048,576 rows and 16,384 columns per sheet (larger tables stop with an error, or are split across sheets on request). MatrixTests.path (sqlite → xlsx) —
sqlite xml — data supported partial tables One table per output file: the first one by default, a chosen one, or all of them as separate files. Only tables are read. Views, indexes, constraints and triggers are not carried over; stored SQL is never executed. XML carries text only; empty values are omitted, so empty text and missing values cannot be told apart when read back. MatrixTests.path (sqlite → xml) —
sqlite yaml — data supported partial tables One table per output file: the first one by default, a chosen one, or all of them as separate files. Only tables are read. Views, indexes, constraints and triggers are not carried over; stored SQL is never executed. Binary values are written as Base64 text. MatrixTests.path (sqlite → yaml) —
toml arrow — data supported structural tables The table is built from the document's list of records (detected, or given by “Record list”); other content of the document is not converted. Columns are written as int64, float64, bool, utf8, binary, date32, time64[us], timestamp[us] and decimal128; nested values as JSON text. Buffers are not compressed. MatrixTests.path (toml → arrow) —
toml csv — data supported structural tables The table is built from the document's list of records (detected, or given by “Record list”); other content of the document is not converted. Text formats carry no types: dates are written in ISO 8601, binary values as Base64, nested values as JSON text. MatrixTests.path (toml → csv) —
toml duckdb — data supported structural tables The table is built from the document's list of records (detected, or given by “Record list”); other content of the document is not converted. MatrixTests.path (toml → duckdb) —
toml json — data supported structural — Dates and binary values become text (ISO 8601 / Base64); JSON has no such types. TOML comments are not carried over. TreeTests.path (toml → json) —
toml jsonl — data supported structural tables The table is built from the document's list of records (detected, or given by “Record list”); other content of the document is not converted. JSON has no date, binary, NaN or infinity: dates become ISO 8601 text, binary values Base64 text, NaN and infinities null. MatrixTests.path (toml → jsonl) —
toml ods — data supported structural tables The table is built from the document's list of records (detected, or given by “Record list”); other content of the document is not converted. Spreadsheet numbers are 64-bit floats: integers beyond 2^53 and long decimals are written as text; NaN and infinities become empty cells; time-zoned instants are written as UTC wall-clock times. At most 1,048,576 rows and 16,384 columns per sheet (larger tables stop with an error, or are split across sheets on request). MatrixTests.path (toml → ods) —
toml parquet — data supported structural tables The table is built from the document's list of records (detected, or given by “Record list”); other content of the document is not converted. MatrixTests.path (toml → parquet) —
toml plist — data supported structural — Property lists have no null and no key order (keys are sorted when read back); integers beyond 64 bits and exact decimals are written as text. TOML comments are not carried over. TreeTests.path (toml → plist) —
toml sqlite — data supported structural tables The table is built from the document's list of records (detected, or given by “Record list”); other content of the document is not converted. SQLite numbers are 64-bit: exact decimals and 128-bit integers are stored as TEXT; dates and times as ISO 8601 text in DATE / DATETIME / TIME columns. MatrixTests.path (toml → sqlite) —
toml toml — data supported structural — TOML has no null (nulls are omitted or stop the conversion, as chosen) and needs a table at the top level; plain values are moved before sub-tables. TOML comments are not carried over. The document is re-serialised: comments and original formatting are not kept. TreeTests.path (toml → toml) —
toml tsv — data supported structural tables The table is built from the document's list of records (detected, or given by “Record list”); other content of the document is not converted. Text formats carry no types: dates are written in ISO 8601, binary values as Base64, nested values as JSON text. MatrixTests.path (toml → tsv) —
toml xls — data supported structural tables The table is built from the document's list of records (detected, or given by “Record list”); other content of the document is not converted. Spreadsheet numbers are 64-bit floats: integers beyond 2^53 and long decimals are written as text; NaN and infinities become empty cells; time-zoned instants are written as UTC wall-clock times. Date-times are stored to the millisecond. At most 65,536 rows and 256 columns per sheet (legacy format limit); produced through the bundled LibreOffice. MatrixTests.path (toml → xls) —
toml xlsx — data supported structural tables The table is built from the document's list of records (detected, or given by “Record list”); other content of the document is not converted. Spreadsheet numbers are 64-bit floats: integers beyond 2^53 and long decimals are written as text; NaN and infinities become empty cells; time-zoned instants are written as UTC wall-clock times. Date-times are stored to the millisecond. At most 1,048,576 rows and 16,384 columns per sheet (larger tables stop with an error, or are split across sheets on request). MatrixTests.path (toml → xlsx) —
toml xml — data supported structural — XML carries text only: numbers and booleans become text, empty strings and nulls both become empty elements. Keys starting with “@” become attributes, “#text” becomes element text, lists become repeated elements. TOML comments are not carried over. TreeTests.path (toml → xml) —
toml yaml — data supported structural — Comments, anchors and formatting of the source are not kept. TOML comments are not carried over. TreeTests.path (toml → yaml) —
tsv arrow — data supported structural tables Column types are detected from the text (verified on every row); numbers are re-written in canonical form (1.50 → 1.5). Identifiers with leading zeros stay text. Turn detection off to keep every cell as written. Columns are written as int64, float64, bool, utf8, binary, date32, time64[us], timestamp[us] and decimal128; nested values as JSON text. Buffers are not compressed. MatrixTests.path (tsv → arrow) —
tsv csv — data supported structural tables Column types are detected from the text (verified on every row); numbers are re-written in canonical form (1.50 → 1.5). Identifiers with leading zeros stay text. Turn detection off to keep every cell as written. Text formats carry no types: dates are written in ISO 8601, binary values as Base64, nested values as JSON text. MatrixTests.path (tsv → csv) —
tsv duckdb — data supported structural tables Column types are detected from the text (verified on every row); numbers are re-written in canonical form (1.50 → 1.5). Identifiers with leading zeros stay text. Turn detection off to keep every cell as written. MatrixTests.path (tsv → duckdb) —
tsv json — data supported structural tables Column types are detected from the text (verified on every row); numbers are re-written in canonical form (1.50 → 1.5). Identifiers with leading zeros stay text. Turn detection off to keep every cell as written. JSON has no date, binary, NaN or infinity: dates become ISO 8601 text, binary values Base64 text, NaN and infinities null. MatrixTests.path (tsv → json) —
tsv jsonl — data supported structural tables Column types are detected from the text (verified on every row); numbers are re-written in canonical form (1.50 → 1.5). Identifiers with leading zeros stay text. Turn detection off to keep every cell as written. JSON has no date, binary, NaN or infinity: dates become ISO 8601 text, binary values Base64 text, NaN and infinities null. MatrixTests.path (tsv → jsonl) —
tsv ods — data supported structural tables Column types are detected from the text (verified on every row); numbers are re-written in canonical form (1.50 → 1.5). Identifiers with leading zeros stay text. Turn detection off to keep every cell as written. Spreadsheet numbers are 64-bit floats: integers beyond 2^53 and long decimals are written as text; NaN and infinities become empty cells; time-zoned instants are written as UTC wall-clock times. At most 1,048,576 rows and 16,384 columns per sheet (larger tables stop with an error, or are split across sheets on request). MatrixTests.path (tsv → ods) —
tsv parquet — data supported structural tables Column types are detected from the text (verified on every row); numbers are re-written in canonical form (1.50 → 1.5). Identifiers with leading zeros stay text. Turn detection off to keep every cell as written. MatrixTests.path (tsv → parquet) —
tsv plist — data supported structural tables Column types are detected from the text (verified on every row); numbers are re-written in canonical form (1.50 → 1.5). Identifiers with leading zeros stay text. Turn detection off to keep every cell as written. Property lists have no null (empty values are omitted or stop the conversion) and are built in memory (20 million cells at most). MatrixTests.path (tsv → plist) —
tsv sqlite — data supported structural tables Column types are detected from the text (verified on every row); numbers are re-written in canonical form (1.50 → 1.5). Identifiers with leading zeros stay text. Turn detection off to keep every cell as written. SQLite numbers are 64-bit: exact decimals and 128-bit integers are stored as TEXT; dates and times as ISO 8601 text in DATE / DATETIME / TIME columns. MatrixTests.path (tsv → sqlite) —
tsv toml — data supported structural tables Column types are detected from the text (verified on every row); numbers are re-written in canonical form (1.50 → 1.5). Identifiers with leading zeros stay text. Turn detection off to keep every cell as written. TOML has no null: empty values are omitted (or stop the conversion, as chosen). MatrixTests.path (tsv → toml) —
tsv tsv — data supported structural tables Column types are detected from the text (verified on every row); numbers are re-written in canonical form (1.50 → 1.5). Identifiers with leading zeros stay text. Turn detection off to keep every cell as written. Text formats carry no types: dates are written in ISO 8601, binary values as Base64, nested values as JSON text. Same-format conversion: the file is re-written with the chosen options. MatrixTests.path (tsv → tsv) —
tsv xls — data supported structural tables Column types are detected from the text (verified on every row); numbers are re-written in canonical form (1.50 → 1.5). Identifiers with leading zeros stay text. Turn detection off to keep every cell as written. Spreadsheet numbers are 64-bit floats: integers beyond 2^53 and long decimals are written as text; NaN and infinities become empty cells; time-zoned instants are written as UTC wall-clock times. Date-times are stored to the millisecond. At most 65,536 rows and 256 columns per sheet (legacy format limit); produced through the bundled LibreOffice. MatrixTests.path (tsv → xls) —
tsv xlsx — data supported structural tables Column types are detected from the text (verified on every row); numbers are re-written in canonical form (1.50 → 1.5). Identifiers with leading zeros stay text. Turn detection off to keep every cell as written. Spreadsheet numbers are 64-bit floats: integers beyond 2^53 and long decimals are written as text; NaN and infinities become empty cells; time-zoned instants are written as UTC wall-clock times. Date-times are stored to the millisecond. At most 1,048,576 rows and 16,384 columns per sheet (larger tables stop with an error, or are split across sheets on request). MatrixTests.path (tsv → xlsx) —
tsv xml — data supported structural tables Column types are detected from the text (verified on every row); numbers are re-written in canonical form (1.50 → 1.5). Identifiers with leading zeros stay text. Turn detection off to keep every cell as written. XML carries text only; empty values are omitted, so empty text and missing values cannot be told apart when read back. MatrixTests.path (tsv → xml) —
tsv yaml — data supported structural tables Column types are detected from the text (verified on every row); numbers are re-written in canonical form (1.50 → 1.5). Identifiers with leading zeros stay text. Turn detection off to keep every cell as written. Binary values are written as Base64 text. MatrixTests.path (tsv → yaml) —
xml arrow — data supported structural tables Column types are detected from the text (verified on every row); numbers are re-written in canonical form (1.50 → 1.5). Identifiers with leading zeros stay text. Turn detection off to keep every cell as written. The table is built from the document's list of records (detected, or given by “Record list”); other content of the document is not converted. Columns are written as int64, float64, bool, utf8, binary, date32, time64[us], timestamp[us] and decimal128; nested values as JSON text. Buffers are not compressed. MatrixTests.path (xml → arrow) —
xml csv — data supported structural tables Column types are detected from the text (verified on every row); numbers are re-written in canonical form (1.50 → 1.5). Identifiers with leading zeros stay text. Turn detection off to keep every cell as written. The table is built from the document's list of records (detected, or given by “Record list”); other content of the document is not converted. Text formats carry no types: dates are written in ISO 8601, binary values as Base64, nested values as JSON text. MatrixTests.path (xml → csv) —
xml duckdb — data supported structural tables Column types are detected from the text (verified on every row); numbers are re-written in canonical form (1.50 → 1.5). Identifiers with leading zeros stay text. Turn detection off to keep every cell as written. The table is built from the document's list of records (detected, or given by “Record list”); other content of the document is not converted. MatrixTests.path (xml → duckdb) —
xml json — data supported structural — Dates and binary values become text (ISO 8601 / Base64); JSON has no such types. XML comments and processing instructions are dropped; attribute order is not kept; every value is read as text. Documents declaring entities (DTD) are refused. TreeTests.path (xml → json) —
xml jsonl — data supported structural tables Column types are detected from the text (verified on every row); numbers are re-written in canonical form (1.50 → 1.5). Identifiers with leading zeros stay text. Turn detection off to keep every cell as written. The table is built from the document's list of records (detected, or given by “Record list”); other content of the document is not converted. JSON has no date, binary, NaN or infinity: dates become ISO 8601 text, binary values Base64 text, NaN and infinities null. MatrixTests.path (xml → jsonl) —
xml ods — data supported structural tables Column types are detected from the text (verified on every row); numbers are re-written in canonical form (1.50 → 1.5). Identifiers with leading zeros stay text. Turn detection off to keep every cell as written. The table is built from the document's list of records (detected, or given by “Record list”); other content of the document is not converted. Spreadsheet numbers are 64-bit floats: integers beyond 2^53 and long decimals are written as text; NaN and infinities become empty cells; time-zoned instants are written as UTC wall-clock times. At most 1,048,576 rows and 16,384 columns per sheet (larger tables stop with an error, or are split across sheets on request). MatrixTests.path (xml → ods) —
xml parquet — data supported structural tables Column types are detected from the text (verified on every row); numbers are re-written in canonical form (1.50 → 1.5). Identifiers with leading zeros stay text. Turn detection off to keep every cell as written. The table is built from the document's list of records (detected, or given by “Record list”); other content of the document is not converted. MatrixTests.path (xml → parquet) —
xml plist — data supported structural — Property lists have no null and no key order (keys are sorted when read back); integers beyond 64 bits and exact decimals are written as text. XML comments and processing instructions are dropped; attribute order is not kept; every value is read as text. Documents declaring entities (DTD) are refused. TreeTests.path (xml → plist) —
xml sqlite — data supported structural tables Column types are detected from the text (verified on every row); numbers are re-written in canonical form (1.50 → 1.5). Identifiers with leading zeros stay text. Turn detection off to keep every cell as written. The table is built from the document's list of records (detected, or given by “Record list”); other content of the document is not converted. SQLite numbers are 64-bit: exact decimals and 128-bit integers are stored as TEXT; dates and times as ISO 8601 text in DATE / DATETIME / TIME columns. MatrixTests.path (xml → sqlite) —
xml toml — data supported structural — TOML has no null (nulls are omitted or stop the conversion, as chosen) and needs a table at the top level; plain values are moved before sub-tables. XML comments and processing instructions are dropped; attribute order is not kept; every value is read as text. Documents declaring entities (DTD) are refused. TreeTests.path (xml → toml) —
xml tsv — data supported structural tables Column types are detected from the text (verified on every row); numbers are re-written in canonical form (1.50 → 1.5). Identifiers with leading zeros stay text. Turn detection off to keep every cell as written. The table is built from the document's list of records (detected, or given by “Record list”); other content of the document is not converted. Text formats carry no types: dates are written in ISO 8601, binary values as Base64, nested values as JSON text. MatrixTests.path (xml → tsv) —
xml xls — data supported structural tables Column types are detected from the text (verified on every row); numbers are re-written in canonical form (1.50 → 1.5). Identifiers with leading zeros stay text. Turn detection off to keep every cell as written. The table is built from the document's list of records (detected, or given by “Record list”); other content of the document is not converted. Spreadsheet numbers are 64-bit floats: integers beyond 2^53 and long decimals are written as text; NaN and infinities become empty cells; time-zoned instants are written as UTC wall-clock times. Date-times are stored to the millisecond. At most 65,536 rows and 256 columns per sheet (legacy format limit); produced through the bundled LibreOffice. MatrixTests.path (xml → xls) —
xml xlsx — data supported structural tables Column types are detected from the text (verified on every row); numbers are re-written in canonical form (1.50 → 1.5). Identifiers with leading zeros stay text. Turn detection off to keep every cell as written. The table is built from the document's list of records (detected, or given by “Record list”); other content of the document is not converted. Spreadsheet numbers are 64-bit floats: integers beyond 2^53 and long decimals are written as text; NaN and infinities become empty cells; time-zoned instants are written as UTC wall-clock times. Date-times are stored to the millisecond. At most 1,048,576 rows and 16,384 columns per sheet (larger tables stop with an error, or are split across sheets on request). MatrixTests.path (xml → xlsx) —
xml xml — data supported structural — XML carries text only: numbers and booleans become text, empty strings and nulls both become empty elements. Keys starting with “@” become attributes, “#text” becomes element text, lists become repeated elements. XML comments and processing instructions are dropped; attribute order is not kept; every value is read as text. Documents declaring entities (DTD) are refused. The document is re-serialised: comments and original formatting are not kept. TreeTests.path (xml → xml) —
xml yaml — data supported structural — Comments, anchors and formatting of the source are not kept. XML comments and processing instructions are dropped; attribute order is not kept; every value is read as text. Documents declaring entities (DTD) are refused. TreeTests.path (xml → yaml) —
yaml arrow — data supported structural tables The table is built from the document's list of records (detected, or given by “Record list”); other content of the document is not converted. Columns are written as int64, float64, bool, utf8, binary, date32, time64[us], timestamp[us] and decimal128; nested values as JSON text. Buffers are not compressed. MatrixTests.path (yaml → arrow) —
yaml csv — data supported structural tables The table is built from the document's list of records (detected, or given by “Record list”); other content of the document is not converted. Text formats carry no types: dates are written in ISO 8601, binary values as Base64, nested values as JSON text. MatrixTests.path (yaml → csv) —
yaml duckdb — data supported structural tables The table is built from the document's list of records (detected, or given by “Record list”); other content of the document is not converted. MatrixTests.path (yaml → duckdb) —
yaml json — data supported structural — Dates and binary values become text (ISO 8601 / Base64); JSON has no such types. YAML comments, anchors and tags are not carried over. TreeTests.path (yaml → json) —
yaml jsonl — data supported structural tables The table is built from the document's list of records (detected, or given by “Record list”); other content of the document is not converted. JSON has no date, binary, NaN or infinity: dates become ISO 8601 text, binary values Base64 text, NaN and infinities null. MatrixTests.path (yaml → jsonl) —
yaml ods — data supported structural tables The table is built from the document's list of records (detected, or given by “Record list”); other content of the document is not converted. Spreadsheet numbers are 64-bit floats: integers beyond 2^53 and long decimals are written as text; NaN and infinities become empty cells; time-zoned instants are written as UTC wall-clock times. At most 1,048,576 rows and 16,384 columns per sheet (larger tables stop with an error, or are split across sheets on request). MatrixTests.path (yaml → ods) —
yaml parquet — data supported structural tables The table is built from the document's list of records (detected, or given by “Record list”); other content of the document is not converted. MatrixTests.path (yaml → parquet) —
yaml plist — data supported structural — Property lists have no null and no key order (keys are sorted when read back); integers beyond 64 bits and exact decimals are written as text. YAML comments, anchors and tags are not carried over. TreeTests.path (yaml → plist) —
yaml sqlite — data supported structural tables The table is built from the document's list of records (detected, or given by “Record list”); other content of the document is not converted. SQLite numbers are 64-bit: exact decimals and 128-bit integers are stored as TEXT; dates and times as ISO 8601 text in DATE / DATETIME / TIME columns. MatrixTests.path (yaml → sqlite) —
yaml toml — data supported structural — TOML has no null (nulls are omitted or stop the conversion, as chosen) and needs a table at the top level; plain values are moved before sub-tables. YAML comments, anchors and tags are not carried over. TreeTests.path (yaml → toml) —
yaml tsv — data supported structural tables The table is built from the document's list of records (detected, or given by “Record list”); other content of the document is not converted. Text formats carry no types: dates are written in ISO 8601, binary values as Base64, nested values as JSON text. MatrixTests.path (yaml → tsv) —
yaml xls — data supported structural tables The table is built from the document's list of records (detected, or given by “Record list”); other content of the document is not converted. Spreadsheet numbers are 64-bit floats: integers beyond 2^53 and long decimals are written as text; NaN and infinities become empty cells; time-zoned instants are written as UTC wall-clock times. Date-times are stored to the millisecond. At most 65,536 rows and 256 columns per sheet (legacy format limit); produced through the bundled LibreOffice. MatrixTests.path (yaml → xls) —
yaml xlsx — data supported structural tables The table is built from the document's list of records (detected, or given by “Record list”); other content of the document is not converted. Spreadsheet numbers are 64-bit floats: integers beyond 2^53 and long decimals are written as text; NaN and infinities become empty cells; time-zoned instants are written as UTC wall-clock times. Date-times are stored to the millisecond. At most 1,048,576 rows and 16,384 columns per sheet (larger tables stop with an error, or are split across sheets on request). MatrixTests.path (yaml → xlsx) —
yaml xml — data supported structural — XML carries text only: numbers and booleans become text, empty strings and nulls both become empty elements. Keys starting with “@” become attributes, “#text” becomes element text, lists become repeated elements. YAML comments, anchors and tags are not carried over. TreeTests.path (yaml → xml) —
yaml yaml — data supported structural — Comments, anchors and formatting of the source are not kept. The document is re-serialised: comments and original formatting are not kept. TreeTests.path (yaml → yaml) —

# presentation sources (24 paths)

Source Destination Operation Engine Status Fidelity Preserves Limitations Tests Performance
odp folder media.extract presentations supported lossless images Embedded image, audio and video files are copied byte for byte; linked (external) media are not fetched. PresentationTests.declaredPath —
odp jpeg — presentations supported lossy dimensions Slides are rendered by LibreOffice, not PowerPoint or Keynote: SmartArt, charts, text fitting and theme effects can look different, and missing fonts are substituted. One image file per slide (rendered through an intermediate PDF). Animations, transitions, video, audio, speaker notes and links are not part of an image. PresentationTests.declaredPath —
odp markdown — presentations supported partial tables Text only, slide by slide in presentation order, with speaker notes: no layout, images, charts or formatting. Text inside pictures, charts and SmartArt is not read. PresentationTests.declaredPath —
odp pdf — presentations supported structural dimensions, images, links, metadata, textStyles Slides are rendered by LibreOffice, not PowerPoint or Keynote: SmartArt, charts, text fitting and theme effects can look different, and missing fonts are substituted. Animations, transitions, video and audio do not exist in a PDF: each slide is exported in its final state. Speaker notes and hidden slides are not included unless enabled. PresentationTests.declaredPath —
odp png — presentations supported structural dimensions Slides are rendered by LibreOffice, not PowerPoint or Keynote: SmartArt, charts, text fitting and theme effects can look different, and missing fonts are substituted. One image file per slide (rendered through an intermediate PDF). Animations, transitions, video, audio, speaker notes and links are not part of an image. PresentationTests.declaredPath —
odp ppt — presentations supported structural dimensions, images, metadata, tables, textStyles Converted by LibreOffice: compatibility with PowerPoint is good but not complete. Animations and transitions are mapped where an equivalent exists, SmartArt becomes plain shapes, embedded video may not play, macros are dropped. Fonts are referenced, not embedded. PresentationTests.declaredPath —
odp pptx — presentations supported structural dimensions, images, metadata, tables, textStyles Converted by LibreOffice: compatibility with PowerPoint is good but not complete. Animations and transitions are mapped where an equivalent exists, SmartArt becomes plain shapes, embedded video may not play, macros are dropped. Fonts are referenced, not embedded. PresentationTests.declaredPath —
odp txt — presentations supported partial — Text only, slide by slide in presentation order, with speaker notes: no layout, images, charts or formatting. Text inside pictures, charts and SmartArt is not read. PresentationTests.declaredPath —
ppt folder media.extract presentations experimental structural images PPT has no package to read: the file is converted to PPTX by LibreOffice first, so images are the ones LibreOffice re-saved (they can be re-encoded). PresentationTests.declaredPath —
ppt jpeg — presentations supported lossy dimensions Slides are rendered by LibreOffice, not PowerPoint or Keynote: SmartArt, charts, text fitting and theme effects can look different, and missing fonts are substituted. One image file per slide (rendered through an intermediate PDF). Animations, transitions, video, audio, speaker notes and links are not part of an image. PresentationTests.declaredPath —
ppt markdown — presentations supported partial tables Text only, slide by slide in presentation order, with speaker notes: no layout, images, charts or formatting. Text inside pictures, charts and SmartArt is not read. PPT is first converted to PPTX by LibreOffice, then read. PresentationTests.declaredPath —
ppt odp — presentations supported structural dimensions, images, metadata, tables, textStyles Converted by LibreOffice: compatibility with PowerPoint is good but not complete. Animations and transitions are mapped where an equivalent exists, SmartArt becomes plain shapes, embedded video may not play, macros are dropped. Fonts are referenced, not embedded. PresentationTests.declaredPath —
ppt pdf — presentations supported structural dimensions, images, links, metadata, textStyles Slides are rendered by LibreOffice, not PowerPoint or Keynote: SmartArt, charts, text fitting and theme effects can look different, and missing fonts are substituted. Animations, transitions, video and audio do not exist in a PDF: each slide is exported in its final state. Speaker notes and hidden slides are not included unless enabled. PresentationTests.declaredPath —
ppt png — presentations supported structural dimensions Slides are rendered by LibreOffice, not PowerPoint or Keynote: SmartArt, charts, text fitting and theme effects can look different, and missing fonts are substituted. One image file per slide (rendered through an intermediate PDF). Animations, transitions, video, audio, speaker notes and links are not part of an image. PresentationTests.declaredPath —
ppt pptx — presentations supported structural dimensions, images, metadata, tables, textStyles Converted by LibreOffice: compatibility with PowerPoint is good but not complete. Animations and transitions are mapped where an equivalent exists, SmartArt becomes plain shapes, embedded video may not play, macros are dropped. Fonts are referenced, not embedded. PresentationTests.declaredPath —
ppt txt — presentations supported partial — Text only, slide by slide in presentation order, with speaker notes: no layout, images, charts or formatting. Text inside pictures, charts and SmartArt is not read. PPT is first converted to PPTX by LibreOffice, then read. PresentationTests.declaredPath —
pptx folder media.extract presentations supported lossless images Embedded image, audio and video files are copied byte for byte; linked (external) media are not fetched. PresentationTests.declaredPath —
pptx jpeg — presentations supported lossy dimensions Slides are rendered by LibreOffice, not PowerPoint or Keynote: SmartArt, charts, text fitting and theme effects can look different, and missing fonts are substituted. One image file per slide (rendered through an intermediate PDF). Animations, transitions, video, audio, speaker notes and links are not part of an image. PresentationTests.declaredPath —
pptx markdown — presentations supported partial tables Text only, slide by slide in presentation order, with speaker notes: no layout, images, charts or formatting. Text inside pictures, charts and SmartArt is not read. PresentationTests.declaredPath 0.01 s (time; PPTX→Markdown text (package XML only))
pptx odp — presentations supported structural dimensions, images, metadata, tables, textStyles Converted by LibreOffice: compatibility with PowerPoint is good but not complete. Animations and transitions are mapped where an equivalent exists, SmartArt becomes plain shapes, embedded video may not play, macros are dropped. Fonts are referenced, not embedded. PresentationTests.declaredPath —
pptx pdf — presentations supported structural dimensions, images, links, metadata, textStyles Slides are rendered by LibreOffice, not PowerPoint or Keynote: SmartArt, charts, text fitting and theme effects can look different, and missing fonts are substituted. Animations, transitions, video and audio do not exist in a PDF: each slide is exported in its final state. Speaker notes and hidden slides are not included unless enabled. PresentationTests.declaredPath 1.63 s (time; LibreOffice PPTX→PDF, 3 slides)
pptx png — presentations supported structural dimensions Slides are rendered by LibreOffice, not PowerPoint or Keynote: SmartArt, charts, text fitting and theme effects can look different, and missing fonts are substituted. One image file per slide (rendered through an intermediate PDF). Animations, transitions, video, audio, speaker notes and links are not part of an image. PresentationTests.declaredPath 1.51 s (1654 × 930 px each) (time; PPTX→PNG, 3 slides at 150 dpi)
pptx ppt — presentations supported structural dimensions, images, metadata, tables, textStyles Converted by LibreOffice: compatibility with PowerPoint is good but not complete. Animations and transitions are mapped where an equivalent exists, SmartArt becomes plain shapes, embedded video may not play, macros are dropped. Fonts are referenced, not embedded. PresentationTests.declaredPath —
pptx txt — presentations supported partial — Text only, slide by slide in presentation order, with speaker notes: no layout, images, charts or formatting. Text inside pictures, charts and SmartArt is not read. PresentationTests.declaredPath —

# archive sources (108 paths)

Source Destination Operation Engine Status Fidelity Preserves Limitations Tests Performance
7z 7z archive.recompress archive supported lossless attributes, metadata Entries are copied one by one from the old archive to the new one, without extracting them to disk. Owner and group are not stored; set-ID and sticky bits are not stored; hard links become separate files. Extended attributes are not stored. 7z archives cannot be encrypted by this app. Compression runs on one processor core and is slow at Normal and Maximum (a few MB per second); ZIP and tar.zst are many times faster. Entries with unsafe names are cleaned or left out, devices and FIFOs are left out, as when extracting. Password-protected 7z archives cannot be opened. Hard links cannot be stored in 7-Zip and are left out. EngineArchiveTests/MatrixTests (suite) —
7z folder archive.extract archive supported lossless attributes, metadata Entries whose name climbs out of the folder (“..”) are left out; absolute paths, drive letters and invisible characters are cleaned; symbolic links pointing outside the folder, devices and FIFOs are not created. Owner and group are not restored; set-user-ID, set-group-ID and sticky bits are removed. An archive inside the archive stays an archive: nothing nested is extracted. Password-protected 7z archives cannot be opened. MatrixTests.everyDeclaredPathConvertsAndValidates (fixtures made by other tools, see "Fixtures"), RoundTripTests, SecurityTests —
7z json archive.verify archive supported partial — CRC-32 of every file that stores one, checked by the decoder. The result is a report file; the archive itself is not changed. Password-protected 7z archives cannot be opened. EngineArchiveTests/MatrixTests (suite) —
7z tar — archive supported lossless attributes, metadata Entries are copied one by one from the old archive to the new one, without extracting them to disk. Owner and group are not stored; set-ID and sticky bits are not stored; hard links become separate files. Extended attributes are stored only when you ask for them. Entries with unsafe names are cleaned or left out, devices and FIFOs are left out, as when extracting. Password-protected 7z archives cannot be opened. EngineArchiveTests/MatrixTests (suite) —
7z tbz2 — archive supported lossless attributes, metadata Entries are copied one by one from the old archive to the new one, without extracting them to disk. Owner and group are not stored; set-ID and sticky bits are not stored; hard links become separate files. Extended attributes are stored only when you ask for them. Entries with unsafe names are cleaned or left out, devices and FIFOs are left out, as when extracting. Password-protected 7z archives cannot be opened. EngineArchiveTests/MatrixTests (suite) —
7z tgz — archive supported lossless attributes, metadata Entries are copied one by one from the old archive to the new one, without extracting them to disk. Owner and group are not stored; set-ID and sticky bits are not stored; hard links become separate files. Extended attributes are stored only when you ask for them. Entries with unsafe names are cleaned or left out, devices and FIFOs are left out, as when extracting. Password-protected 7z archives cannot be opened. EngineArchiveTests/MatrixTests (suite) —
7z txt archive.verify archive supported partial — CRC-32 of every file that stores one, checked by the decoder. The result is a report file; the archive itself is not changed. Password-protected 7z archives cannot be opened. EngineArchiveTests/MatrixTests (suite) —
7z txz — archive supported lossless attributes, metadata Entries are copied one by one from the old archive to the new one, without extracting them to disk. Owner and group are not stored; set-ID and sticky bits are not stored; hard links become separate files. Extended attributes are stored only when you ask for them. Entries with unsafe names are cleaned or left out, devices and FIFOs are left out, as when extracting. Password-protected 7z archives cannot be opened. EngineArchiveTests/MatrixTests (suite) —
7z tzst — archive supported lossless attributes, metadata Entries are copied one by one from the old archive to the new one, without extracting them to disk. Owner and group are not stored; set-ID and sticky bits are not stored; hard links become separate files. Extended attributes are stored only when you ask for them. Entries with unsafe names are cleaned or left out, devices and FIFOs are left out, as when extracting. Password-protected 7z archives cannot be opened. EngineArchiveTests/MatrixTests (suite) —
7z zip — archive supported lossless attributes, metadata Entries are copied one by one from the old archive to the new one, without extracting them to disk. Owner and group are not stored; set-ID and sticky bits are not stored; hard links become separate files. Extended attributes and resource forks are not stored. Encryption (optional) protects file contents only: file names, sizes and the targets of symbolic links stay readable. Entries with unsafe names are cleaned or left out, devices and FIFOs are left out, as when extracting. Password-protected 7z archives cannot be opened. Hard links cannot be stored in ZIP and are left out. EngineArchiveTests/MatrixTests (suite) —
bz2 bz2 archive.recompress archive supported lossless — The data is decompressed and compressed again in memory, block by block; a gzip file name or comment is not carried over. EngineArchiveTests/MatrixTests (suite) —
bz2 folder archive.extract archive supported lossless — The stream holds one file; it is named after the archive, without its last extension. MatrixTests.everyDeclaredPathConvertsAndValidates (fixtures made by other tools, see "Fixtures"), RoundTripTests, SecurityTests —
bz2 gz — archive supported lossless — The data is decompressed and compressed again in memory, block by block; a gzip file name or comment is not carried over. EngineArchiveTests/MatrixTests (suite) —
bz2 json archive.verify archive supported partial — The compressed stream's own integrity check. The result is a report file; the archive itself is not changed. EngineArchiveTests/MatrixTests (suite) —
bz2 txt archive.verify archive supported partial — The compressed stream's own integrity check. The result is a report file; the archive itself is not changed. EngineArchiveTests/MatrixTests (suite) —
bz2 xz — archive supported lossless — The data is decompressed and compressed again in memory, block by block; a gzip file name or comment is not carried over. EngineArchiveTests/MatrixTests (suite) —
bz2 zst — archive supported lossless — The data is decompressed and compressed again in memory, block by block; a gzip file name or comment is not carried over. EngineArchiveTests/MatrixTests (suite) —
gz bz2 — archive supported lossless — The data is decompressed and compressed again in memory, block by block; a gzip file name or comment is not carried over. EngineArchiveTests/MatrixTests (suite) —
gz folder archive.extract archive supported lossless — The stream holds one file; it is named after the archive, without its last extension. MatrixTests.everyDeclaredPathConvertsAndValidates (fixtures made by other tools, see "Fixtures"), RoundTripTests, SecurityTests —
gz gz archive.recompress archive supported lossless — The data is decompressed and compressed again in memory, block by block; a gzip file name or comment is not carried over. EngineArchiveTests/MatrixTests (suite) —
gz json archive.verify archive supported partial — The gzip CRC-32 and length of the stream, checked by this app with zlib. The result is a report file; the archive itself is not changed. EngineArchiveTests/MatrixTests (suite) —
gz txt archive.verify archive supported partial — The gzip CRC-32 and length of the stream, checked by this app with zlib. The result is a report file; the archive itself is not changed. EngineArchiveTests/MatrixTests (suite) —
gz xz — archive supported lossless — The data is decompressed and compressed again in memory, block by block; a gzip file name or comment is not carried over. EngineArchiveTests/MatrixTests (suite) —
gz zst — archive supported lossless — The data is decompressed and compressed again in memory, block by block; a gzip file name or comment is not carried over. EngineArchiveTests/MatrixTests (suite) —
rar 7z — archive experimental lossless attributes, metadata RAR reading was verified on this build only with archives whose files are stored without compression (RAR 4 and RAR 5); compressed, solid and multi-volume RAR archives are decoded by the same library but were not tested. Password-protected RAR archives cannot be opened. RAR archives cannot be created: no free encoder exists. Entries are copied one by one from the old archive to the new one, without extracting them to disk. Owner and group are not stored; set-ID and sticky bits are not stored; hard links become separate files. Extended attributes are not stored. 7z archives cannot be encrypted by this app. Compression runs on one processor core and is slow at Normal and Maximum (a few MB per second); ZIP and tar.zst are many times faster. Entries with unsafe names are cleaned or left out, devices and FIFOs are left out, as when extracting. Hard links cannot be stored in 7-Zip and are left out. EngineArchiveTests/MatrixTests (suite) —
rar folder archive.extract archive experimental lossless attributes, metadata RAR reading was verified on this build only with archives whose files are stored without compression (RAR 4 and RAR 5); compressed, solid and multi-volume RAR archives are decoded by the same library but were not tested. Password-protected RAR archives cannot be opened. RAR archives cannot be created: no free encoder exists. Entries whose name climbs out of the folder (“..”) are left out; absolute paths, drive letters and invisible characters are cleaned; symbolic links pointing outside the folder, devices and FIFOs are not created. Owner and group are not restored; set-user-ID, set-group-ID and sticky bits are removed. An archive inside the archive stays an archive: nothing nested is extracted. MatrixTests.everyDeclaredPathConvertsAndValidates (fixtures made by other tools, see "Fixtures"), RoundTripTests, SecurityTests —
rar json archive.verify archive experimental partial — RAR reading was verified on this build only with archives whose files are stored without compression (RAR 4 and RAR 5); compressed, solid and multi-volume RAR archives are decoded by the same library but were not tested. Password-protected RAR archives cannot be opened. RAR archives cannot be created: no free encoder exists. CRC-32 of every file: checked by the decoder for compressed files, and by this app against the file headers (single-volume, unencrypted archives). The result is a report file; the archive itself is not changed. EngineArchiveTests/MatrixTests (suite) —
rar tar — archive experimental lossless attributes, metadata RAR reading was verified on this build only with archives whose files are stored without compression (RAR 4 and RAR 5); compressed, solid and multi-volume RAR archives are decoded by the same library but were not tested. Password-protected RAR archives cannot be opened. RAR archives cannot be created: no free encoder exists. Entries are copied one by one from the old archive to the new one, without extracting them to disk. Owner and group are not stored; set-ID and sticky bits are not stored; hard links become separate files. Extended attributes are stored only when you ask for them. Entries with unsafe names are cleaned or left out, devices and FIFOs are left out, as when extracting. EngineArchiveTests/MatrixTests (suite) —
rar tbz2 — archive experimental lossless attributes, metadata RAR reading was verified on this build only with archives whose files are stored without compression (RAR 4 and RAR 5); compressed, solid and multi-volume RAR archives are decoded by the same library but were not tested. Password-protected RAR archives cannot be opened. RAR archives cannot be created: no free encoder exists. Entries are copied one by one from the old archive to the new one, without extracting them to disk. Owner and group are not stored; set-ID and sticky bits are not stored; hard links become separate files. Extended attributes are stored only when you ask for them. Entries with unsafe names are cleaned or left out, devices and FIFOs are left out, as when extracting. EngineArchiveTests/MatrixTests (suite) —
rar tgz — archive experimental lossless attributes, metadata RAR reading was verified on this build only with archives whose files are stored without compression (RAR 4 and RAR 5); compressed, solid and multi-volume RAR archives are decoded by the same library but were not tested. Password-protected RAR archives cannot be opened. RAR archives cannot be created: no free encoder exists. Entries are copied one by one from the old archive to the new one, without extracting them to disk. Owner and group are not stored; set-ID and sticky bits are not stored; hard links become separate files. Extended attributes are stored only when you ask for them. Entries with unsafe names are cleaned or left out, devices and FIFOs are left out, as when extracting. EngineArchiveTests/MatrixTests (suite) —
rar txt archive.verify archive experimental partial — RAR reading was verified on this build only with archives whose files are stored without compression (RAR 4 and RAR 5); compressed, solid and multi-volume RAR archives are decoded by the same library but were not tested. Password-protected RAR archives cannot be opened. RAR archives cannot be created: no free encoder exists. CRC-32 of every file: checked by the decoder for compressed files, and by this app against the file headers (single-volume, unencrypted archives). The result is a report file; the archive itself is not changed. EngineArchiveTests/MatrixTests (suite) —
rar txz — archive experimental lossless attributes, metadata RAR reading was verified on this build only with archives whose files are stored without compression (RAR 4 and RAR 5); compressed, solid and multi-volume RAR archives are decoded by the same library but were not tested. Password-protected RAR archives cannot be opened. RAR archives cannot be created: no free encoder exists. Entries are copied one by one from the old archive to the new one, without extracting them to disk. Owner and group are not stored; set-ID and sticky bits are not stored; hard links become separate files. Extended attributes are stored only when you ask for them. Entries with unsafe names are cleaned or left out, devices and FIFOs are left out, as when extracting. EngineArchiveTests/MatrixTests (suite) —
rar tzst — archive experimental lossless attributes, metadata RAR reading was verified on this build only with archives whose files are stored without compression (RAR 4 and RAR 5); compressed, solid and multi-volume RAR archives are decoded by the same library but were not tested. Password-protected RAR archives cannot be opened. RAR archives cannot be created: no free encoder exists. Entries are copied one by one from the old archive to the new one, without extracting them to disk. Owner and group are not stored; set-ID and sticky bits are not stored; hard links become separate files. Extended attributes are stored only when you ask for them. Entries with unsafe names are cleaned or left out, devices and FIFOs are left out, as when extracting. EngineArchiveTests/MatrixTests (suite) —
rar zip — archive experimental lossless attributes, metadata RAR reading was verified on this build only with archives whose files are stored without compression (RAR 4 and RAR 5); compressed, solid and multi-volume RAR archives are decoded by the same library but were not tested. Password-protected RAR archives cannot be opened. RAR archives cannot be created: no free encoder exists. Entries are copied one by one from the old archive to the new one, without extracting them to disk. Owner and group are not stored; set-ID and sticky bits are not stored; hard links become separate files. Extended attributes and resource forks are not stored. Encryption (optional) protects file contents only: file names, sizes and the targets of symbolic links stay readable. Entries with unsafe names are cleaned or left out, devices and FIFOs are left out, as when extracting. Hard links cannot be stored in ZIP and are left out. EngineArchiveTests/MatrixTests (suite) —
tar 7z — archive supported lossless attributes, metadata Entries are copied one by one from the old archive to the new one, without extracting them to disk. Owner and group are not stored; set-ID and sticky bits are not stored; hard links become separate files. Extended attributes are not stored. 7z archives cannot be encrypted by this app. Compression runs on one processor core and is slow at Normal and Maximum (a few MB per second); ZIP and tar.zst are many times faster. Entries with unsafe names are cleaned or left out, devices and FIFOs are left out, as when extracting. Hard links cannot be stored in 7-Zip and are left out. EngineArchiveTests/MatrixTests (suite) —
tar folder archive.extract archive supported lossless attributes, metadata Entries whose name climbs out of the folder (“..”) are left out; absolute paths, drive letters and invisible characters are cleaned; symbolic links pointing outside the folder, devices and FIFOs are not created. Owner and group are not restored; set-user-ID, set-group-ID and sticky bits are removed. An archive inside the archive stays an archive: nothing nested is extracted. MatrixTests.everyDeclaredPathConvertsAndValidates (fixtures made by other tools, see "Fixtures"), RoundTripTests, SecurityTests —
tar json archive.verify archive supported partial — TAR stores no checksum of file data: only the header checksums and the structure are checked. The result is a report file; the archive itself is not changed. EngineArchiveTests/MatrixTests (suite) —
tar tar archive.recompress archive supported lossless attributes, metadata Entries are copied one by one from the old archive to the new one, without extracting them to disk. Owner and group are not stored; set-ID and sticky bits are not stored; hard links become separate files. Extended attributes are stored only when you ask for them. Entries with unsafe names are cleaned or left out, devices and FIFOs are left out, as when extracting. EngineArchiveTests/MatrixTests (suite) —
tar tbz2 — archive supported lossless attributes, metadata Entries are copied one by one from the old archive to the new one, without extracting them to disk. Owner and group are not stored; set-ID and sticky bits are not stored; hard links become separate files. Extended attributes are stored only when you ask for them. Entries with unsafe names are cleaned or left out, devices and FIFOs are left out, as when extracting. EngineArchiveTests/MatrixTests (suite) —
tar tgz — archive supported lossless attributes, metadata Entries are copied one by one from the old archive to the new one, without extracting them to disk. Owner and group are not stored; set-ID and sticky bits are not stored; hard links become separate files. Extended attributes are stored only when you ask for them. Entries with unsafe names are cleaned or left out, devices and FIFOs are left out, as when extracting. EngineArchiveTests/MatrixTests (suite) —
tar txt archive.verify archive supported partial — TAR stores no checksum of file data: only the header checksums and the structure are checked. The result is a report file; the archive itself is not changed. EngineArchiveTests/MatrixTests (suite) —
tar txz — archive supported lossless attributes, metadata Entries are copied one by one from the old archive to the new one, without extracting them to disk. Owner and group are not stored; set-ID and sticky bits are not stored; hard links become separate files. Extended attributes are stored only when you ask for them. Entries with unsafe names are cleaned or left out, devices and FIFOs are left out, as when extracting. EngineArchiveTests/MatrixTests (suite) —
tar tzst — archive supported lossless attributes, metadata Entries are copied one by one from the old archive to the new one, without extracting them to disk. Owner and group are not stored; set-ID and sticky bits are not stored; hard links become separate files. Extended attributes are stored only when you ask for them. Entries with unsafe names are cleaned or left out, devices and FIFOs are left out, as when extracting. EngineArchiveTests/MatrixTests (suite) —
tar zip — archive supported lossless attributes, metadata Entries are copied one by one from the old archive to the new one, without extracting them to disk. Owner and group are not stored; set-ID and sticky bits are not stored; hard links become separate files. Extended attributes and resource forks are not stored. Encryption (optional) protects file contents only: file names, sizes and the targets of symbolic links stay readable. Entries with unsafe names are cleaned or left out, devices and FIFOs are left out, as when extracting. Hard links cannot be stored in ZIP and are left out. EngineArchiveTests/MatrixTests (suite) —
tbz2 7z — archive supported lossless attributes, metadata Entries are copied one by one from the old archive to the new one, without extracting them to disk. Owner and group are not stored; set-ID and sticky bits are not stored; hard links become separate files. Extended attributes are not stored. 7z archives cannot be encrypted by this app. Compression runs on one processor core and is slow at Normal and Maximum (a few MB per second); ZIP and tar.zst are many times faster. Entries with unsafe names are cleaned or left out, devices and FIFOs are left out, as when extracting. Hard links cannot be stored in 7-Zip and are left out. EngineArchiveTests/MatrixTests (suite) —
tbz2 folder archive.extract archive supported lossless attributes, metadata Entries whose name climbs out of the folder (“..”) are left out; absolute paths, drive letters and invisible characters are cleaned; symbolic links pointing outside the folder, devices and FIFOs are not created. Owner and group are not restored; set-user-ID, set-group-ID and sticky bits are removed. An archive inside the archive stays an archive: nothing nested is extracted. MatrixTests.everyDeclaredPathConvertsAndValidates (fixtures made by other tools, see "Fixtures"), RoundTripTests, SecurityTests —
tbz2 json archive.verify archive supported partial — The compressed stream's own integrity check covers the whole archive; TAR stores no per-file checksum. The result is a report file; the archive itself is not changed. EngineArchiveTests/MatrixTests (suite) —
tbz2 tar — archive supported lossless attributes, metadata Entries are copied one by one from the old archive to the new one, without extracting them to disk. Owner and group are not stored; set-ID and sticky bits are not stored; hard links become separate files. Extended attributes are stored only when you ask for them. Entries with unsafe names are cleaned or left out, devices and FIFOs are left out, as when extracting. EngineArchiveTests/MatrixTests (suite) —
tbz2 tbz2 archive.recompress archive supported lossless attributes, metadata Entries are copied one by one from the old archive to the new one, without extracting them to disk. Owner and group are not stored; set-ID and sticky bits are not stored; hard links become separate files. Extended attributes are stored only when you ask for them. Entries with unsafe names are cleaned or left out, devices and FIFOs are left out, as when extracting. EngineArchiveTests/MatrixTests (suite) —
tbz2 tgz — archive supported lossless attributes, metadata Entries are copied one by one from the old archive to the new one, without extracting them to disk. Owner and group are not stored; set-ID and sticky bits are not stored; hard links become separate files. Extended attributes are stored only when you ask for them. Entries with unsafe names are cleaned or left out, devices and FIFOs are left out, as when extracting. EngineArchiveTests/MatrixTests (suite) —
tbz2 txt archive.verify archive supported partial — The compressed stream's own integrity check covers the whole archive; TAR stores no per-file checksum. The result is a report file; the archive itself is not changed. EngineArchiveTests/MatrixTests (suite) —
tbz2 txz — archive supported lossless attributes, metadata Entries are copied one by one from the old archive to the new one, without extracting them to disk. Owner and group are not stored; set-ID and sticky bits are not stored; hard links become separate files. Extended attributes are stored only when you ask for them. Entries with unsafe names are cleaned or left out, devices and FIFOs are left out, as when extracting. EngineArchiveTests/MatrixTests (suite) —
tbz2 tzst — archive supported lossless attributes, metadata Entries are copied one by one from the old archive to the new one, without extracting them to disk. Owner and group are not stored; set-ID and sticky bits are not stored; hard links become separate files. Extended attributes are stored only when you ask for them. Entries with unsafe names are cleaned or left out, devices and FIFOs are left out, as when extracting. EngineArchiveTests/MatrixTests (suite) —
tbz2 zip — archive supported lossless attributes, metadata Entries are copied one by one from the old archive to the new one, without extracting them to disk. Owner and group are not stored; set-ID and sticky bits are not stored; hard links become separate files. Extended attributes and resource forks are not stored. Encryption (optional) protects file contents only: file names, sizes and the targets of symbolic links stay readable. Entries with unsafe names are cleaned or left out, devices and FIFOs are left out, as when extracting. Hard links cannot be stored in ZIP and are left out. EngineArchiveTests/MatrixTests (suite) —
tgz 7z — archive supported lossless attributes, metadata Entries are copied one by one from the old archive to the new one, without extracting them to disk. Owner and group are not stored; set-ID and sticky bits are not stored; hard links become separate files. Extended attributes are not stored. 7z archives cannot be encrypted by this app. Compression runs on one processor core and is slow at Normal and Maximum (a few MB per second); ZIP and tar.zst are many times faster. Entries with unsafe names are cleaned or left out, devices and FIFOs are left out, as when extracting. Hard links cannot be stored in 7-Zip and are left out. EngineArchiveTests/MatrixTests (suite) —
tgz folder archive.extract archive supported lossless attributes, metadata Entries whose name climbs out of the folder (“..”) are left out; absolute paths, drive letters and invisible characters are cleaned; symbolic links pointing outside the folder, devices and FIFOs are not created. Owner and group are not restored; set-user-ID, set-group-ID and sticky bits are removed. An archive inside the archive stays an archive: nothing nested is extracted. MatrixTests.everyDeclaredPathConvertsAndValidates (fixtures made by other tools, see "Fixtures"), RoundTripTests, SecurityTests —
tgz json archive.verify archive supported partial — The gzip CRC-32 and length cover the whole archive (checked by this app with zlib); TAR stores no per-file checksum. The result is a report file; the archive itself is not changed. EngineArchiveTests/MatrixTests (suite) —
tgz tar — archive supported lossless attributes, metadata Entries are copied one by one from the old archive to the new one, without extracting them to disk. Owner and group are not stored; set-ID and sticky bits are not stored; hard links become separate files. Extended attributes are stored only when you ask for them. Entries with unsafe names are cleaned or left out, devices and FIFOs are left out, as when extracting. EngineArchiveTests/MatrixTests (suite) —
tgz tbz2 — archive supported lossless attributes, metadata Entries are copied one by one from the old archive to the new one, without extracting them to disk. Owner and group are not stored; set-ID and sticky bits are not stored; hard links become separate files. Extended attributes are stored only when you ask for them. Entries with unsafe names are cleaned or left out, devices and FIFOs are left out, as when extracting. EngineArchiveTests/MatrixTests (suite) —
tgz tgz archive.recompress archive supported lossless attributes, metadata Entries are copied one by one from the old archive to the new one, without extracting them to disk. Owner and group are not stored; set-ID and sticky bits are not stored; hard links become separate files. Extended attributes are stored only when you ask for them. Entries with unsafe names are cleaned or left out, devices and FIFOs are left out, as when extracting. EngineArchiveTests/MatrixTests (suite) —
tgz txt archive.verify archive supported partial — The gzip CRC-32 and length cover the whole archive (checked by this app with zlib); TAR stores no per-file checksum. The result is a report file; the archive itself is not changed. EngineArchiveTests/MatrixTests (suite) —
tgz txz — archive supported lossless attributes, metadata Entries are copied one by one from the old archive to the new one, without extracting them to disk. Owner and group are not stored; set-ID and sticky bits are not stored; hard links become separate files. Extended attributes are stored only when you ask for them. Entries with unsafe names are cleaned or left out, devices and FIFOs are left out, as when extracting. EngineArchiveTests/MatrixTests (suite) —
tgz tzst — archive supported lossless attributes, metadata Entries are copied one by one from the old archive to the new one, without extracting them to disk. Owner and group are not stored; set-ID and sticky bits are not stored; hard links become separate files. Extended attributes are stored only when you ask for them. Entries with unsafe names are cleaned or left out, devices and FIFOs are left out, as when extracting. EngineArchiveTests/MatrixTests (suite) —
tgz zip — archive supported lossless attributes, metadata Entries are copied one by one from the old archive to the new one, without extracting them to disk. Owner and group are not stored; set-ID and sticky bits are not stored; hard links become separate files. Extended attributes and resource forks are not stored. Encryption (optional) protects file contents only: file names, sizes and the targets of symbolic links stay readable. Entries with unsafe names are cleaned or left out, devices and FIFOs are left out, as when extracting. Hard links cannot be stored in ZIP and are left out. EngineArchiveTests/MatrixTests (suite) —
txz 7z — archive supported lossless attributes, metadata Entries are copied one by one from the old archive to the new one, without extracting them to disk. Owner and group are not stored; set-ID and sticky bits are not stored; hard links become separate files. Extended attributes are not stored. 7z archives cannot be encrypted by this app. Compression runs on one processor core and is slow at Normal and Maximum (a few MB per second); ZIP and tar.zst are many times faster. Entries with unsafe names are cleaned or left out, devices and FIFOs are left out, as when extracting. Hard links cannot be stored in 7-Zip and are left out. EngineArchiveTests/MatrixTests (suite) —
txz folder archive.extract archive supported lossless attributes, metadata Entries whose name climbs out of the folder (“..”) are left out; absolute paths, drive letters and invisible characters are cleaned; symbolic links pointing outside the folder, devices and FIFOs are not created. Owner and group are not restored; set-user-ID, set-group-ID and sticky bits are removed. An archive inside the archive stays an archive: nothing nested is extracted. MatrixTests.everyDeclaredPathConvertsAndValidates (fixtures made by other tools, see "Fixtures"), RoundTripTests, SecurityTests —
txz json archive.verify archive supported partial — The compressed stream's own integrity check covers the whole archive; TAR stores no per-file checksum. The result is a report file; the archive itself is not changed. EngineArchiveTests/MatrixTests (suite) —
txz tar — archive supported lossless attributes, metadata Entries are copied one by one from the old archive to the new one, without extracting them to disk. Owner and group are not stored; set-ID and sticky bits are not stored; hard links become separate files. Extended attributes are stored only when you ask for them. Entries with unsafe names are cleaned or left out, devices and FIFOs are left out, as when extracting. EngineArchiveTests/MatrixTests (suite) —
txz tbz2 — archive supported lossless attributes, metadata Entries are copied one by one from the old archive to the new one, without extracting them to disk. Owner and group are not stored; set-ID and sticky bits are not stored; hard links become separate files. Extended attributes are stored only when you ask for them. Entries with unsafe names are cleaned or left out, devices and FIFOs are left out, as when extracting. EngineArchiveTests/MatrixTests (suite) —
txz tgz — archive supported lossless attributes, metadata Entries are copied one by one from the old archive to the new one, without extracting them to disk. Owner and group are not stored; set-ID and sticky bits are not stored; hard links become separate files. Extended attributes are stored only when you ask for them. Entries with unsafe names are cleaned or left out, devices and FIFOs are left out, as when extracting. EngineArchiveTests/MatrixTests (suite) —
txz txt archive.verify archive supported partial — The compressed stream's own integrity check covers the whole archive; TAR stores no per-file checksum. The result is a report file; the archive itself is not changed. EngineArchiveTests/MatrixTests (suite) —
txz txz archive.recompress archive supported lossless attributes, metadata Entries are copied one by one from the old archive to the new one, without extracting them to disk. Owner and group are not stored; set-ID and sticky bits are not stored; hard links become separate files. Extended attributes are stored only when you ask for them. Entries with unsafe names are cleaned or left out, devices and FIFOs are left out, as when extracting. EngineArchiveTests/MatrixTests (suite) —
txz tzst — archive supported lossless attributes, metadata Entries are copied one by one from the old archive to the new one, without extracting them to disk. Owner and group are not stored; set-ID and sticky bits are not stored; hard links become separate files. Extended attributes are stored only when you ask for them. Entries with unsafe names are cleaned or left out, devices and FIFOs are left out, as when extracting. EngineArchiveTests/MatrixTests (suite) —
txz zip — archive supported lossless attributes, metadata Entries are copied one by one from the old archive to the new one, without extracting them to disk. Owner and group are not stored; set-ID and sticky bits are not stored; hard links become separate files. Extended attributes and resource forks are not stored. Encryption (optional) protects file contents only: file names, sizes and the targets of symbolic links stay readable. Entries with unsafe names are cleaned or left out, devices and FIFOs are left out, as when extracting. Hard links cannot be stored in ZIP and are left out. EngineArchiveTests/MatrixTests (suite) —
tzst 7z — archive supported lossless attributes, metadata Entries are copied one by one from the old archive to the new one, without extracting them to disk. Owner and group are not stored; set-ID and sticky bits are not stored; hard links become separate files. Extended attributes are not stored. 7z archives cannot be encrypted by this app. Compression runs on one processor core and is slow at Normal and Maximum (a few MB per second); ZIP and tar.zst are many times faster. Entries with unsafe names are cleaned or left out, devices and FIFOs are left out, as when extracting. Hard links cannot be stored in 7-Zip and are left out. EngineArchiveTests/MatrixTests (suite) —
tzst folder archive.extract archive supported lossless attributes, metadata Entries whose name climbs out of the folder (“..”) are left out; absolute paths, drive letters and invisible characters are cleaned; symbolic links pointing outside the folder, devices and FIFOs are not created. Owner and group are not restored; set-user-ID, set-group-ID and sticky bits are removed. An archive inside the archive stays an archive: nothing nested is extracted. MatrixTests.everyDeclaredPathConvertsAndValidates (fixtures made by other tools, see "Fixtures"), RoundTripTests, SecurityTests —
tzst json archive.verify archive supported partial — The compressed stream's own integrity check covers the whole archive; TAR stores no per-file checksum. The result is a report file; the archive itself is not changed. EngineArchiveTests/MatrixTests (suite) —
tzst tar — archive supported lossless attributes, metadata Entries are copied one by one from the old archive to the new one, without extracting them to disk. Owner and group are not stored; set-ID and sticky bits are not stored; hard links become separate files. Extended attributes are stored only when you ask for them. Entries with unsafe names are cleaned or left out, devices and FIFOs are left out, as when extracting. EngineArchiveTests/MatrixTests (suite) —
tzst tbz2 — archive supported lossless attributes, metadata Entries are copied one by one from the old archive to the new one, without extracting them to disk. Owner and group are not stored; set-ID and sticky bits are not stored; hard links become separate files. Extended attributes are stored only when you ask for them. Entries with unsafe names are cleaned or left out, devices and FIFOs are left out, as when extracting. EngineArchiveTests/MatrixTests (suite) —
tzst tgz — archive supported lossless attributes, metadata Entries are copied one by one from the old archive to the new one, without extracting them to disk. Owner and group are not stored; set-ID and sticky bits are not stored; hard links become separate files. Extended attributes are stored only when you ask for them. Entries with unsafe names are cleaned or left out, devices and FIFOs are left out, as when extracting. EngineArchiveTests/MatrixTests (suite) —
tzst txt archive.verify archive supported partial — The compressed stream's own integrity check covers the whole archive; TAR stores no per-file checksum. The result is a report file; the archive itself is not changed. EngineArchiveTests/MatrixTests (suite) —
tzst txz — archive supported lossless attributes, metadata Entries are copied one by one from the old archive to the new one, without extracting them to disk. Owner and group are not stored; set-ID and sticky bits are not stored; hard links become separate files. Extended attributes are stored only when you ask for them. Entries with unsafe names are cleaned or left out, devices and FIFOs are left out, as when extracting. EngineArchiveTests/MatrixTests (suite) —
tzst tzst archive.recompress archive supported lossless attributes, metadata Entries are copied one by one from the old archive to the new one, without extracting them to disk. Owner and group are not stored; set-ID and sticky bits are not stored; hard links become separate files. Extended attributes are stored only when you ask for them. Entries with unsafe names are cleaned or left out, devices and FIFOs are left out, as when extracting. EngineArchiveTests/MatrixTests (suite) —
tzst zip — archive supported lossless attributes, metadata Entries are copied one by one from the old archive to the new one, without extracting them to disk. Owner and group are not stored; set-ID and sticky bits are not stored; hard links become separate files. Extended attributes and resource forks are not stored. Encryption (optional) protects file contents only: file names, sizes and the targets of symbolic links stay readable. Entries with unsafe names are cleaned or left out, devices and FIFOs are left out, as when extracting. Hard links cannot be stored in ZIP and are left out. EngineArchiveTests/MatrixTests (suite) —
xz bz2 — archive supported lossless — The data is decompressed and compressed again in memory, block by block; a gzip file name or comment is not carried over. EngineArchiveTests/MatrixTests (suite) —
xz folder archive.extract archive supported lossless — The stream holds one file; it is named after the archive, without its last extension. MatrixTests.everyDeclaredPathConvertsAndValidates (fixtures made by other tools, see "Fixtures"), RoundTripTests, SecurityTests —
xz gz — archive supported lossless — The data is decompressed and compressed again in memory, block by block; a gzip file name or comment is not carried over. EngineArchiveTests/MatrixTests (suite) —
xz json archive.verify archive supported partial — The compressed stream's own integrity check. The result is a report file; the archive itself is not changed. EngineArchiveTests/MatrixTests (suite) —
xz txt archive.verify archive supported partial — The compressed stream's own integrity check. The result is a report file; the archive itself is not changed. EngineArchiveTests/MatrixTests (suite) —
xz xz archive.recompress archive supported lossless — The data is decompressed and compressed again in memory, block by block; a gzip file name or comment is not carried over. EngineArchiveTests/MatrixTests (suite) —
xz zst — archive supported lossless — The data is decompressed and compressed again in memory, block by block; a gzip file name or comment is not carried over. EngineArchiveTests/MatrixTests (suite) —
zip 7z — archive supported lossless attributes, metadata Entries are copied one by one from the old archive to the new one, without extracting them to disk. Owner and group are not stored; set-ID and sticky bits are not stored; hard links become separate files. Extended attributes are not stored. 7z archives cannot be encrypted by this app. Compression runs on one processor core and is slow at Normal and Maximum (a few MB per second); ZIP and tar.zst are many times faster. Entries with unsafe names are cleaned or left out, devices and FIFOs are left out, as when extracting. Password-protected ZIP files (ZipCrypto and AES) need their password. A symbolic link whose target is itself encrypted cannot be decoded and is left out. Hard links cannot be stored in 7-Zip and are left out. EngineArchiveTests/MatrixTests (suite) —
zip folder archive.extract archive supported lossless attributes, metadata Entries whose name climbs out of the folder (“..”) are left out; absolute paths, drive letters and invisible characters are cleaned; symbolic links pointing outside the folder, devices and FIFOs are not created. Owner and group are not restored; set-user-ID, set-group-ID and sticky bits are removed. An archive inside the archive stays an archive: nothing nested is extracted. Password-protected ZIP files (ZipCrypto and AES) need their password. A symbolic link whose target is itself encrypted cannot be decoded and is left out. MatrixTests.everyDeclaredPathConvertsAndValidates (fixtures made by other tools, see "Fixtures"), RoundTripTests, SecurityTests —
zip json archive.verify archive supported partial — CRC-32 of every file, checked by the decoder and against the central directory. The result is a report file; the archive itself is not changed. Password-protected ZIP files (ZipCrypto and AES) need their password. A symbolic link whose target is itself encrypted cannot be decoded and is left out. EngineArchiveTests/MatrixTests (suite) —
zip tar — archive supported lossless attributes, metadata Entries are copied one by one from the old archive to the new one, without extracting them to disk. Owner and group are not stored; set-ID and sticky bits are not stored; hard links become separate files. Extended attributes are stored only when you ask for them. Entries with unsafe names are cleaned or left out, devices and FIFOs are left out, as when extracting. Password-protected ZIP files (ZipCrypto and AES) need their password. A symbolic link whose target is itself encrypted cannot be decoded and is left out. EngineArchiveTests/MatrixTests (suite) —
zip tbz2 — archive supported lossless attributes, metadata Entries are copied one by one from the old archive to the new one, without extracting them to disk. Owner and group are not stored; set-ID and sticky bits are not stored; hard links become separate files. Extended attributes are stored only when you ask for them. Entries with unsafe names are cleaned or left out, devices and FIFOs are left out, as when extracting. Password-protected ZIP files (ZipCrypto and AES) need their password. A symbolic link whose target is itself encrypted cannot be decoded and is left out. EngineArchiveTests/MatrixTests (suite) —
zip tgz — archive supported lossless attributes, metadata Entries are copied one by one from the old archive to the new one, without extracting them to disk. Owner and group are not stored; set-ID and sticky bits are not stored; hard links become separate files. Extended attributes are stored only when you ask for them. Entries with unsafe names are cleaned or left out, devices and FIFOs are left out, as when extracting. Password-protected ZIP files (ZipCrypto and AES) need their password. A symbolic link whose target is itself encrypted cannot be decoded and is left out. EngineArchiveTests/MatrixTests (suite) —
zip txt archive.verify archive supported partial — CRC-32 of every file, checked by the decoder and against the central directory. The result is a report file; the archive itself is not changed. Password-protected ZIP files (ZipCrypto and AES) need their password. A symbolic link whose target is itself encrypted cannot be decoded and is left out. EngineArchiveTests/MatrixTests (suite) —
zip txz — archive supported lossless attributes, metadata Entries are copied one by one from the old archive to the new one, without extracting them to disk. Owner and group are not stored; set-ID and sticky bits are not stored; hard links become separate files. Extended attributes are stored only when you ask for them. Entries with unsafe names are cleaned or left out, devices and FIFOs are left out, as when extracting. Password-protected ZIP files (ZipCrypto and AES) need their password. A symbolic link whose target is itself encrypted cannot be decoded and is left out. EngineArchiveTests/MatrixTests (suite) —
zip tzst — archive supported lossless attributes, metadata Entries are copied one by one from the old archive to the new one, without extracting them to disk. Owner and group are not stored; set-ID and sticky bits are not stored; hard links become separate files. Extended attributes are stored only when you ask for them. Entries with unsafe names are cleaned or left out, devices and FIFOs are left out, as when extracting. Password-protected ZIP files (ZipCrypto and AES) need their password. A symbolic link whose target is itself encrypted cannot be decoded and is left out. EngineArchiveTests/MatrixTests (suite) —
zip zip archive.recompress archive supported lossless attributes, metadata Entries are copied one by one from the old archive to the new one, without extracting them to disk. Owner and group are not stored; set-ID and sticky bits are not stored; hard links become separate files. Extended attributes and resource forks are not stored. Encryption (optional) protects file contents only: file names, sizes and the targets of symbolic links stay readable. Entries with unsafe names are cleaned or left out, devices and FIFOs are left out, as when extracting. Password-protected ZIP files (ZipCrypto and AES) need their password. A symbolic link whose target is itself encrypted cannot be decoded and is left out. Hard links cannot be stored in ZIP and are left out. EngineArchiveTests/MatrixTests (suite) —
zst bz2 — archive supported lossless — The data is decompressed and compressed again in memory, block by block; a gzip file name or comment is not carried over. EngineArchiveTests/MatrixTests (suite) —
zst folder archive.extract archive supported lossless — The stream holds one file; it is named after the archive, without its last extension. MatrixTests.everyDeclaredPathConvertsAndValidates (fixtures made by other tools, see "Fixtures"), RoundTripTests, SecurityTests —
zst gz — archive supported lossless — The data is decompressed and compressed again in memory, block by block; a gzip file name or comment is not carried over. EngineArchiveTests/MatrixTests (suite) —
zst json archive.verify archive supported partial — The compressed stream's own integrity check. The result is a report file; the archive itself is not changed. EngineArchiveTests/MatrixTests (suite) —
zst txt archive.verify archive supported partial — The compressed stream's own integrity check. The result is a report file; the archive itself is not changed. EngineArchiveTests/MatrixTests (suite) —
zst xz — archive supported lossless — The data is decompressed and compressed again in memory, block by block; a gzip file name or comment is not carried over. EngineArchiveTests/MatrixTests (suite) —
zst zst archive.recompress archive supported lossless — The data is decompressed and compressed again in memory, block by block; a gzip file name or comment is not carried over. EngineArchiveTests/MatrixTests (suite) —

# model3D sources (132 paths)

Source Destination Operation Engine Status Fidelity Preserves Limitations Tests Performance
3mf 3mf model.reprocess model3d supported partial coordinateSystem, dimensions, materials Decoded by ORIX reader, written by ORIX writer, verified by re-reading with Assimp. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. The node hierarchy is flattened to one placed object per mesh. Only the base colour of each material is kept. Vertex colours are not written. Texture coordinates are not written. 3MF textures, colour groups and multi-file (production) packages are not read. declaredPath(id:) with model3d:3mf>3mf#model.reprocess —
3mf dae — model3d supported partial attributes, coordinateSystem, dimensions, materials Decoded by ORIX reader, written by Assimp, verified by re-reading with ORIX COLLADA checker. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. PBR materials are approximated with COLLADA Phong values. Only the first set of texture coordinates is written. Written by Assimp's COLLADA exporter; metallic and roughness are not part of COLLADA 1.4. 3MF textures, colour groups and multi-file (production) packages are not read. declaredPath(id:) with model3d:3mf>dae —
3mf glb — model3d supported structural attributes, coordinateSystem, dimensions, materials Decoded by ORIX reader, written by ORIX writer, verified by re-reading with ORIX glTF checker + Assimp. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. glTF is in metres: a known source unit is converted, an unknown one is written unchanged. 3MF textures, colour groups and multi-file (production) packages are not read. declaredPath(id:) with model3d:3mf>glb —
3mf gltf — model3d supported structural attributes, coordinateSystem, dimensions, materials Decoded by ORIX reader, written by ORIX writer, verified by re-reading with ORIX glTF checker + Assimp. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. glTF is in metres: a known source unit is converted, an unknown one is written unchanged. 3MF textures, colour groups and multi-file (production) packages are not read. declaredPath(id:) with model3d:3mf>gltf —
3mf obj — model3d supported partial attributes, dimensions, materials Decoded by ORIX reader, written by ORIX writer, verified by re-reading with Assimp. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. The node hierarchy is flattened: every mesh is written in world coordinates. PBR materials are approximated with MTL Phong values (plus Pr/Pm lines). Vertex colours are not written. Only the first set of texture coordinates is written. Wavefront OBJ records neither a unit nor an up axis. 3MF textures, colour groups and multi-file (production) packages are not read. declaredPath(id:) with model3d:3mf>obj —
3mf ply — model3d supported partial attributes, dimensions Decoded by ORIX reader, written by ORIX writer, verified by re-reading with Assimp. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. The node hierarchy is flattened: every mesh is written in world coordinates. PLY has no materials: they are dropped. Only the first set of texture coordinates is written. PLY records neither a unit nor an up axis. 3MF textures, colour groups and multi-file (production) packages are not read. declaredPath(id:) with model3d:3mf>ply —
3mf stl — model3d supported partial dimensions Decoded by ORIX reader, written by ORIX writer, verified by re-reading with Assimp. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. The node hierarchy is flattened: every mesh is written in world coordinates. STL has no materials: they are dropped. Vertex colours are not written. Texture coordinates are not written. STL records neither a unit nor an up axis. Facet normals are recomputed from the triangle winding. 3MF textures, colour groups and multi-file (production) packages are not read. declaredPath(id:) with model3d:3mf>stl —
3mf usd — model3d supported structural attributes, coordinateSystem, dimensions, materials Decoded by ORIX reader, written by ORIX writer, verified by re-reading with Model I/O. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. 3MF textures, colour groups and multi-file (production) packages are not read. declaredPath(id:) with model3d:3mf>usd —
3mf usda — model3d supported structural attributes, coordinateSystem, dimensions, materials Decoded by ORIX reader, written by ORIX writer, verified by re-reading with Model I/O. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. 3MF textures, colour groups and multi-file (production) packages are not read. declaredPath(id:) with model3d:3mf>usda —
3mf usdc — model3d supported structural attributes, dimensions, materials Decoded by ORIX reader, written by ORIX writer + Model I/O, verified by re-reading with SceneKit. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. Only the first set of texture coordinates is written. Model I/O cannot record metersPerUnit in a crate file: coordinates are converted to centimetres (USD's default) when the source unit is known. The crate layer is produced by Model I/O from the text layer ORIX writes: Model I/O re-authors the materials (their names and number can change). 3MF textures, colour groups and multi-file (production) packages are not read. declaredPath(id:) with model3d:3mf>usdc —
3mf usdz — model3d supported structural attributes, coordinateSystem, dimensions, materials Decoded by ORIX reader, written by ORIX writer, verified by re-reading with Model I/O. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. 3MF textures, colour groups and multi-file (production) packages are not read. declaredPath(id:) with model3d:3mf>usdz —
dae 3mf — model3d supported partial coordinateSystem, dimensions, materials Decoded by Assimp, written by ORIX writer, verified by re-reading with Assimp. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. The node hierarchy is flattened to one placed object per mesh. Only the base colour of each material is kept. 3MF cannot hold textures: they are dropped. Vertex colours are not written. Texture coordinates are not written. declaredPath(id:) with model3d:dae>3mf —
dae dae model.reprocess model3d supported partial attributes, coordinateSystem, dimensions, materials, textures Decoded by Assimp, written by Assimp, verified by re-reading with ORIX COLLADA checker. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. PBR materials are approximated with COLLADA Phong values. Textures are written as files next to the model; the output is then a folder. Only the first set of texture coordinates is written. Written by Assimp's COLLADA exporter; metallic and roughness are not part of COLLADA 1.4. declaredPath(id:) with model3d:dae>dae#model.reprocess —
dae glb — model3d supported structural attributes, coordinateSystem, dimensions, materials, textures Decoded by Assimp, written by ORIX writer, verified by re-reading with ORIX glTF checker + Assimp. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. Only PNG and JPEG textures are carried. glTF is in metres: a known source unit is converted, an unknown one is written unchanged. declaredPath(id:) with model3d:dae>glb —
dae gltf — model3d supported structural attributes, coordinateSystem, dimensions, materials, textures Decoded by Assimp, written by ORIX writer, verified by re-reading with ORIX glTF checker + Assimp. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. Textures are written as files next to the model; the output is then a folder. Only PNG and JPEG textures are carried. glTF is in metres: a known source unit is converted, an unknown one is written unchanged. declaredPath(id:) with model3d:dae>gltf —
dae obj — model3d supported partial attributes, dimensions, materials, textures Decoded by Assimp, written by ORIX writer, verified by re-reading with Assimp. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. The node hierarchy is flattened: every mesh is written in world coordinates. PBR materials are approximated with MTL Phong values (plus Pr/Pm lines). Textures are written as files next to the model; the output is then a folder. Vertex colours are not written. Only the first set of texture coordinates is written. Wavefront OBJ records neither a unit nor an up axis. declaredPath(id:) with model3d:dae>obj —
dae ply — model3d supported partial attributes, dimensions Decoded by Assimp, written by ORIX writer, verified by re-reading with Assimp. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. The node hierarchy is flattened: every mesh is written in world coordinates. PLY has no materials: they are dropped. PLY cannot hold textures: they are dropped. Only the first set of texture coordinates is written. PLY records neither a unit nor an up axis. declaredPath(id:) with model3d:dae>ply —
dae stl — model3d supported partial dimensions Decoded by Assimp, written by ORIX writer, verified by re-reading with Assimp. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. The node hierarchy is flattened: every mesh is written in world coordinates. STL has no materials: they are dropped. STL cannot hold textures: they are dropped. Vertex colours are not written. Texture coordinates are not written. STL records neither a unit nor an up axis. Facet normals are recomputed from the triangle winding. declaredPath(id:) with model3d:dae>stl —
dae usd — model3d supported structural attributes, coordinateSystem, dimensions, materials, textures Decoded by Assimp, written by ORIX writer, verified by re-reading with Model I/O. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. Textures are written as files next to the model; the output is then a folder. Only PNG and JPEG textures are carried. declaredPath(id:) with model3d:dae>usd —
dae usda — model3d supported structural attributes, coordinateSystem, dimensions, materials, textures Decoded by Assimp, written by ORIX writer, verified by re-reading with Model I/O. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. Textures are written as files next to the model; the output is then a folder. Only PNG and JPEG textures are carried. declaredPath(id:) with model3d:dae>usda —
dae usdc — model3d supported structural attributes, dimensions, materials, textures Decoded by Assimp, written by ORIX writer + Model I/O, verified by re-reading with SceneKit. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. Textures are written as files next to the model; the output is then a folder. Only PNG and JPEG textures are carried. Only the first set of texture coordinates is written. Model I/O cannot record metersPerUnit in a crate file: coordinates are converted to centimetres (USD's default) when the source unit is known. The crate layer is produced by Model I/O from the text layer ORIX writes: Model I/O re-authors the materials (their names and number can change). declaredPath(id:) with model3d:dae>usdc —
dae usdz — model3d supported structural attributes, coordinateSystem, dimensions, materials, textures Decoded by Assimp, written by ORIX writer, verified by re-reading with Model I/O. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. Only PNG and JPEG textures are carried. declaredPath(id:) with model3d:dae>usdz —
fbx 3mf — model3d experimental partial coordinateSystem, dimensions, materials Decoded by Assimp, written by ORIX writer, verified by re-reading with Assimp. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. The node hierarchy is flattened to one placed object per mesh. Only the base colour of each material is kept. 3MF cannot hold textures: they are dropped. Vertex colours are not written. Texture coordinates are not written. FBX import is experimental: it is only tested on files written by Assimp's own FBX exporter (no third-party FBX file was available). declaredPath(id:) with model3d:fbx>3mf —
fbx dae — model3d experimental partial attributes, coordinateSystem, dimensions, materials, textures Decoded by Assimp, written by Assimp, verified by re-reading with ORIX COLLADA checker. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. PBR materials are approximated with COLLADA Phong values. Textures are written as files next to the model; the output is then a folder. Only the first set of texture coordinates is written. Written by Assimp's COLLADA exporter; metallic and roughness are not part of COLLADA 1.4. FBX import is experimental: it is only tested on files written by Assimp's own FBX exporter (no third-party FBX file was available). declaredPath(id:) with model3d:fbx>dae —
fbx glb — model3d experimental structural attributes, coordinateSystem, dimensions, materials, textures Decoded by Assimp, written by ORIX writer, verified by re-reading with ORIX glTF checker + Assimp. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. Only PNG and JPEG textures are carried. glTF is in metres: a known source unit is converted, an unknown one is written unchanged. FBX import is experimental: it is only tested on files written by Assimp's own FBX exporter (no third-party FBX file was available). declaredPath(id:) with model3d:fbx>glb —
fbx gltf — model3d experimental structural attributes, coordinateSystem, dimensions, materials, textures Decoded by Assimp, written by ORIX writer, verified by re-reading with ORIX glTF checker + Assimp. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. Textures are written as files next to the model; the output is then a folder. Only PNG and JPEG textures are carried. glTF is in metres: a known source unit is converted, an unknown one is written unchanged. FBX import is experimental: it is only tested on files written by Assimp's own FBX exporter (no third-party FBX file was available). declaredPath(id:) with model3d:fbx>gltf —
fbx obj — model3d experimental partial attributes, dimensions, materials, textures Decoded by Assimp, written by ORIX writer, verified by re-reading with Assimp. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. The node hierarchy is flattened: every mesh is written in world coordinates. PBR materials are approximated with MTL Phong values (plus Pr/Pm lines). Textures are written as files next to the model; the output is then a folder. Vertex colours are not written. Only the first set of texture coordinates is written. Wavefront OBJ records neither a unit nor an up axis. FBX import is experimental: it is only tested on files written by Assimp's own FBX exporter (no third-party FBX file was available). declaredPath(id:) with model3d:fbx>obj —
fbx ply — model3d experimental partial attributes, dimensions Decoded by Assimp, written by ORIX writer, verified by re-reading with Assimp. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. The node hierarchy is flattened: every mesh is written in world coordinates. PLY has no materials: they are dropped. PLY cannot hold textures: they are dropped. Only the first set of texture coordinates is written. PLY records neither a unit nor an up axis. FBX import is experimental: it is only tested on files written by Assimp's own FBX exporter (no third-party FBX file was available). declaredPath(id:) with model3d:fbx>ply —
fbx stl — model3d experimental partial dimensions Decoded by Assimp, written by ORIX writer, verified by re-reading with Assimp. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. The node hierarchy is flattened: every mesh is written in world coordinates. STL has no materials: they are dropped. STL cannot hold textures: they are dropped. Vertex colours are not written. Texture coordinates are not written. STL records neither a unit nor an up axis. Facet normals are recomputed from the triangle winding. FBX import is experimental: it is only tested on files written by Assimp's own FBX exporter (no third-party FBX file was available). declaredPath(id:) with model3d:fbx>stl —
fbx usd — model3d experimental structural attributes, coordinateSystem, dimensions, materials, textures Decoded by Assimp, written by ORIX writer, verified by re-reading with Model I/O. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. Textures are written as files next to the model; the output is then a folder. Only PNG and JPEG textures are carried. FBX import is experimental: it is only tested on files written by Assimp's own FBX exporter (no third-party FBX file was available). declaredPath(id:) with model3d:fbx>usd —
fbx usda — model3d experimental structural attributes, coordinateSystem, dimensions, materials, textures Decoded by Assimp, written by ORIX writer, verified by re-reading with Model I/O. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. Textures are written as files next to the model; the output is then a folder. Only PNG and JPEG textures are carried. FBX import is experimental: it is only tested on files written by Assimp's own FBX exporter (no third-party FBX file was available). declaredPath(id:) with model3d:fbx>usda —
fbx usdc — model3d experimental structural attributes, dimensions, materials, textures Decoded by Assimp, written by ORIX writer + Model I/O, verified by re-reading with SceneKit. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. Textures are written as files next to the model; the output is then a folder. Only PNG and JPEG textures are carried. Only the first set of texture coordinates is written. Model I/O cannot record metersPerUnit in a crate file: coordinates are converted to centimetres (USD's default) when the source unit is known. The crate layer is produced by Model I/O from the text layer ORIX writes: Model I/O re-authors the materials (their names and number can change). FBX import is experimental: it is only tested on files written by Assimp's own FBX exporter (no third-party FBX file was available). declaredPath(id:) with model3d:fbx>usdc —
fbx usdz — model3d experimental structural attributes, coordinateSystem, dimensions, materials, textures Decoded by Assimp, written by ORIX writer, verified by re-reading with Model I/O. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. Only PNG and JPEG textures are carried. FBX import is experimental: it is only tested on files written by Assimp's own FBX exporter (no third-party FBX file was available). declaredPath(id:) with model3d:fbx>usdz —
glb 3mf — model3d supported partial coordinateSystem, dimensions, materials Decoded by Assimp, written by ORIX writer, verified by re-reading with Assimp. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. The node hierarchy is flattened to one placed object per mesh. Only the base colour of each material is kept. 3MF cannot hold textures: they are dropped. Vertex colours are not written. Texture coordinates are not written. declaredPath(id:) with model3d:glb>3mf —
glb dae — model3d supported partial attributes, coordinateSystem, dimensions, materials, textures Decoded by Assimp, written by Assimp, verified by re-reading with ORIX COLLADA checker. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. PBR materials are approximated with COLLADA Phong values. Textures are written as files next to the model; the output is then a folder. Only the first set of texture coordinates is written. Written by Assimp's COLLADA exporter; metallic and roughness are not part of COLLADA 1.4. declaredPath(id:) with model3d:glb>dae —
glb glb model.reprocess model3d supported structural attributes, coordinateSystem, dimensions, materials, textures Decoded by Assimp, written by ORIX writer, verified by re-reading with ORIX glTF checker + Assimp. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. Only PNG and JPEG textures are carried. glTF is in metres: a known source unit is converted, an unknown one is written unchanged. declaredPath(id:) with model3d:glb>glb#model.reprocess —
glb gltf — model3d supported structural attributes, coordinateSystem, dimensions, materials, textures Decoded by Assimp, written by ORIX writer, verified by re-reading with ORIX glTF checker + Assimp. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. Textures are written as files next to the model; the output is then a folder. Only PNG and JPEG textures are carried. glTF is in metres: a known source unit is converted, an unknown one is written unchanged. declaredPath(id:) with model3d:glb>gltf —
glb obj — model3d supported partial attributes, dimensions, materials, textures Decoded by Assimp, written by ORIX writer, verified by re-reading with Assimp. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. The node hierarchy is flattened: every mesh is written in world coordinates. PBR materials are approximated with MTL Phong values (plus Pr/Pm lines). Textures are written as files next to the model; the output is then a folder. Vertex colours are not written. Only the first set of texture coordinates is written. Wavefront OBJ records neither a unit nor an up axis. declaredPath(id:) with model3d:glb>obj —
glb ply — model3d supported partial attributes, dimensions Decoded by Assimp, written by ORIX writer, verified by re-reading with Assimp. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. The node hierarchy is flattened: every mesh is written in world coordinates. PLY has no materials: they are dropped. PLY cannot hold textures: they are dropped. Only the first set of texture coordinates is written. PLY records neither a unit nor an up axis. declaredPath(id:) with model3d:glb>ply —
glb stl — model3d supported partial dimensions Decoded by Assimp, written by ORIX writer, verified by re-reading with Assimp. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. The node hierarchy is flattened: every mesh is written in world coordinates. STL has no materials: they are dropped. STL cannot hold textures: they are dropped. Vertex colours are not written. Texture coordinates are not written. STL records neither a unit nor an up axis. Facet normals are recomputed from the triangle winding. declaredPath(id:) with model3d:glb>stl —
glb usd — model3d supported structural attributes, coordinateSystem, dimensions, materials, textures Decoded by Assimp, written by ORIX writer, verified by re-reading with Model I/O. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. Textures are written as files next to the model; the output is then a folder. Only PNG and JPEG textures are carried. declaredPath(id:) with model3d:glb>usd —
glb usda — model3d supported structural attributes, coordinateSystem, dimensions, materials, textures Decoded by Assimp, written by ORIX writer, verified by re-reading with Model I/O. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. Textures are written as files next to the model; the output is then a folder. Only PNG and JPEG textures are carried. declaredPath(id:) with model3d:glb>usda —
glb usdc — model3d supported structural attributes, dimensions, materials, textures Decoded by Assimp, written by ORIX writer + Model I/O, verified by re-reading with SceneKit. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. Textures are written as files next to the model; the output is then a folder. Only PNG and JPEG textures are carried. Only the first set of texture coordinates is written. Model I/O cannot record metersPerUnit in a crate file: coordinates are converted to centimetres (USD's default) when the source unit is known. The crate layer is produced by Model I/O from the text layer ORIX writes: Model I/O re-authors the materials (their names and number can change). declaredPath(id:) with model3d:glb>usdc —
glb usdz — model3d supported structural attributes, coordinateSystem, dimensions, materials, textures Decoded by Assimp, written by ORIX writer, verified by re-reading with Model I/O. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. Only PNG and JPEG textures are carried. declaredPath(id:) with model3d:glb>usdz 998 000 (triangles; GLB → USDZ)
gltf 3mf — model3d supported partial coordinateSystem, dimensions, materials Decoded by Assimp, written by ORIX writer, verified by re-reading with Assimp. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. The node hierarchy is flattened to one placed object per mesh. Only the base colour of each material is kept. 3MF cannot hold textures: they are dropped. Vertex colours are not written. Texture coordinates are not written. declaredPath(id:) with model3d:gltf>3mf —
gltf dae — model3d supported partial attributes, coordinateSystem, dimensions, materials, textures Decoded by Assimp, written by Assimp, verified by re-reading with ORIX COLLADA checker. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. PBR materials are approximated with COLLADA Phong values. Textures are written as files next to the model; the output is then a folder. Only the first set of texture coordinates is written. Written by Assimp's COLLADA exporter; metallic and roughness are not part of COLLADA 1.4. declaredPath(id:) with model3d:gltf>dae —
gltf glb — model3d supported structural attributes, coordinateSystem, dimensions, materials, textures Decoded by Assimp, written by ORIX writer, verified by re-reading with ORIX glTF checker + Assimp. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. Only PNG and JPEG textures are carried. glTF is in metres: a known source unit is converted, an unknown one is written unchanged. declaredPath(id:) with model3d:gltf>glb —
gltf gltf model.reprocess model3d supported structural attributes, coordinateSystem, dimensions, materials, textures Decoded by Assimp, written by ORIX writer, verified by re-reading with ORIX glTF checker + Assimp. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. Textures are written as files next to the model; the output is then a folder. Only PNG and JPEG textures are carried. glTF is in metres: a known source unit is converted, an unknown one is written unchanged. declaredPath(id:) with model3d:gltf>gltf#model.reprocess —
gltf obj — model3d supported partial attributes, dimensions, materials, textures Decoded by Assimp, written by ORIX writer, verified by re-reading with Assimp. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. The node hierarchy is flattened: every mesh is written in world coordinates. PBR materials are approximated with MTL Phong values (plus Pr/Pm lines). Textures are written as files next to the model; the output is then a folder. Vertex colours are not written. Only the first set of texture coordinates is written. Wavefront OBJ records neither a unit nor an up axis. declaredPath(id:) with model3d:gltf>obj —
gltf ply — model3d supported partial attributes, dimensions Decoded by Assimp, written by ORIX writer, verified by re-reading with Assimp. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. The node hierarchy is flattened: every mesh is written in world coordinates. PLY has no materials: they are dropped. PLY cannot hold textures: they are dropped. Only the first set of texture coordinates is written. PLY records neither a unit nor an up axis. declaredPath(id:) with model3d:gltf>ply —
gltf stl — model3d supported partial dimensions Decoded by Assimp, written by ORIX writer, verified by re-reading with Assimp. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. The node hierarchy is flattened: every mesh is written in world coordinates. STL has no materials: they are dropped. STL cannot hold textures: they are dropped. Vertex colours are not written. Texture coordinates are not written. STL records neither a unit nor an up axis. Facet normals are recomputed from the triangle winding. declaredPath(id:) with model3d:gltf>stl —
gltf usd — model3d supported structural attributes, coordinateSystem, dimensions, materials, textures Decoded by Assimp, written by ORIX writer, verified by re-reading with Model I/O. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. Textures are written as files next to the model; the output is then a folder. Only PNG and JPEG textures are carried. declaredPath(id:) with model3d:gltf>usd —
gltf usda — model3d supported structural attributes, coordinateSystem, dimensions, materials, textures Decoded by Assimp, written by ORIX writer, verified by re-reading with Model I/O. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. Textures are written as files next to the model; the output is then a folder. Only PNG and JPEG textures are carried. declaredPath(id:) with model3d:gltf>usda —
gltf usdc — model3d supported structural attributes, dimensions, materials, textures Decoded by Assimp, written by ORIX writer + Model I/O, verified by re-reading with SceneKit. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. Textures are written as files next to the model; the output is then a folder. Only PNG and JPEG textures are carried. Only the first set of texture coordinates is written. Model I/O cannot record metersPerUnit in a crate file: coordinates are converted to centimetres (USD's default) when the source unit is known. The crate layer is produced by Model I/O from the text layer ORIX writes: Model I/O re-authors the materials (their names and number can change). declaredPath(id:) with model3d:gltf>usdc —
gltf usdz — model3d supported structural attributes, coordinateSystem, dimensions, materials, textures Decoded by Assimp, written by ORIX writer, verified by re-reading with Model I/O. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. Only PNG and JPEG textures are carried. declaredPath(id:) with model3d:gltf>usdz —
obj 3mf — model3d supported partial coordinateSystem, dimensions, materials Decoded by Assimp, written by ORIX writer, verified by re-reading with Assimp. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. Only the base colour of each material is kept. 3MF cannot hold textures: they are dropped. Vertex colours are not written. Texture coordinates are not written. Wavefront OBJ states no unit: set the source unit, otherwise numbers are written unchanged. declaredPath(id:) with model3d:obj>3mf —
obj dae — model3d supported partial attributes, coordinateSystem, dimensions, materials, textures Decoded by Assimp, written by Assimp, verified by re-reading with ORIX COLLADA checker. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. PBR materials are approximated with COLLADA Phong values. Textures are written as files next to the model; the output is then a folder. Only the first set of texture coordinates is written. Wavefront OBJ states no unit: set the source unit, otherwise numbers are written unchanged. Written by Assimp's COLLADA exporter; metallic and roughness are not part of COLLADA 1.4. declaredPath(id:) with model3d:obj>dae —
obj glb — model3d supported structural attributes, coordinateSystem, dimensions, materials, textures Decoded by Assimp, written by ORIX writer, verified by re-reading with ORIX glTF checker + Assimp. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. Only PNG and JPEG textures are carried. glTF is in metres: a known source unit is converted, an unknown one is written unchanged. Wavefront OBJ states no unit: set the source unit, otherwise numbers are written unchanged. declaredPath(id:) with model3d:obj>glb 998 000 (triangles; OBJ → GLB)
obj gltf — model3d supported structural attributes, coordinateSystem, dimensions, materials, textures Decoded by Assimp, written by ORIX writer, verified by re-reading with ORIX glTF checker + Assimp. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. Textures are written as files next to the model; the output is then a folder. Only PNG and JPEG textures are carried. glTF is in metres: a known source unit is converted, an unknown one is written unchanged. Wavefront OBJ states no unit: set the source unit, otherwise numbers are written unchanged. declaredPath(id:) with model3d:obj>gltf —
obj obj model.reprocess model3d supported partial attributes, dimensions, materials, textures Decoded by Assimp, written by ORIX writer, verified by re-reading with Assimp. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. PBR materials are approximated with MTL Phong values (plus Pr/Pm lines). Textures are written as files next to the model; the output is then a folder. Vertex colours are not written. Only the first set of texture coordinates is written. Wavefront OBJ records neither a unit nor an up axis. declaredPath(id:) with model3d:obj>obj#model.reprocess —
obj ply — model3d supported partial attributes, dimensions Decoded by Assimp, written by ORIX writer, verified by re-reading with Assimp. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. PLY has no materials: they are dropped. PLY cannot hold textures: they are dropped. Only the first set of texture coordinates is written. PLY records neither a unit nor an up axis. declaredPath(id:) with model3d:obj>ply —
obj stl — model3d supported partial dimensions Decoded by Assimp, written by ORIX writer, verified by re-reading with Assimp. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. STL has no materials: they are dropped. STL cannot hold textures: they are dropped. Vertex colours are not written. Texture coordinates are not written. STL records neither a unit nor an up axis. Facet normals are recomputed from the triangle winding. declaredPath(id:) with model3d:obj>stl —
obj usd — model3d supported structural attributes, coordinateSystem, dimensions, materials, textures Decoded by Assimp, written by ORIX writer, verified by re-reading with Model I/O. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. Textures are written as files next to the model; the output is then a folder. Only PNG and JPEG textures are carried. Wavefront OBJ states no unit: set the source unit, otherwise numbers are written unchanged. declaredPath(id:) with model3d:obj>usd —
obj usda — model3d supported structural attributes, coordinateSystem, dimensions, materials, textures Decoded by Assimp, written by ORIX writer, verified by re-reading with Model I/O. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. Textures are written as files next to the model; the output is then a folder. Only PNG and JPEG textures are carried. Wavefront OBJ states no unit: set the source unit, otherwise numbers are written unchanged. declaredPath(id:) with model3d:obj>usda —
obj usdc — model3d supported structural attributes, dimensions, materials, textures Decoded by Assimp, written by ORIX writer + Model I/O, verified by re-reading with SceneKit. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. Textures are written as files next to the model; the output is then a folder. Only PNG and JPEG textures are carried. Only the first set of texture coordinates is written. Model I/O cannot record metersPerUnit in a crate file: coordinates are converted to centimetres (USD's default) when the source unit is known. Wavefront OBJ states no unit: set the source unit, otherwise numbers are written unchanged. The crate layer is produced by Model I/O from the text layer ORIX writes: Model I/O re-authors the materials (their names and number can change). declaredPath(id:) with model3d:obj>usdc —
obj usdz — model3d supported structural attributes, coordinateSystem, dimensions, materials, textures Decoded by Assimp, written by ORIX writer, verified by re-reading with Model I/O. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. Only PNG and JPEG textures are carried. Wavefront OBJ states no unit: set the source unit, otherwise numbers are written unchanged. declaredPath(id:) with model3d:obj>usdz —
ply 3mf — model3d supported structural coordinateSystem, dimensions Decoded by ORIX reader, written by ORIX writer, verified by re-reading with Assimp. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. Vertex colours are not written. Texture coordinates are not written. PLY states no unit: set the source unit, otherwise numbers are written unchanged. declaredPath(id:) with model3d:ply>3mf —
ply dae — model3d supported structural attributes, coordinateSystem, dimensions Decoded by ORIX reader, written by Assimp, verified by re-reading with ORIX COLLADA checker. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. Only the first set of texture coordinates is written. PLY states no unit: set the source unit, otherwise numbers are written unchanged. Written by Assimp's COLLADA exporter; metallic and roughness are not part of COLLADA 1.4. declaredPath(id:) with model3d:ply>dae —
ply glb — model3d supported structural attributes, coordinateSystem, dimensions Decoded by ORIX reader, written by ORIX writer, verified by re-reading with ORIX glTF checker + Assimp. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. glTF is in metres: a known source unit is converted, an unknown one is written unchanged. PLY states no unit: set the source unit, otherwise numbers are written unchanged. declaredPath(id:) with model3d:ply>glb —
ply gltf — model3d supported structural attributes, coordinateSystem, dimensions Decoded by ORIX reader, written by ORIX writer, verified by re-reading with ORIX glTF checker + Assimp. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. glTF is in metres: a known source unit is converted, an unknown one is written unchanged. PLY states no unit: set the source unit, otherwise numbers are written unchanged. declaredPath(id:) with model3d:ply>gltf —
ply obj — model3d supported structural attributes, dimensions Decoded by ORIX reader, written by ORIX writer, verified by re-reading with Assimp. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. Vertex colours are not written. Only the first set of texture coordinates is written. Wavefront OBJ records neither a unit nor an up axis. declaredPath(id:) with model3d:ply>obj —
ply ply model.reprocess model3d supported structural attributes, dimensions Decoded by ORIX reader, written by ORIX writer, verified by re-reading with Assimp. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. Only the first set of texture coordinates is written. PLY records neither a unit nor an up axis. declaredPath(id:) with model3d:ply>ply#model.reprocess —
ply stl — model3d supported structural dimensions Decoded by ORIX reader, written by ORIX writer, verified by re-reading with Assimp. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. Vertex colours are not written. Texture coordinates are not written. STL records neither a unit nor an up axis. Facet normals are recomputed from the triangle winding. declaredPath(id:) with model3d:ply>stl —
ply usd — model3d supported structural attributes, coordinateSystem, dimensions Decoded by ORIX reader, written by ORIX writer, verified by re-reading with Model I/O. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. PLY states no unit: set the source unit, otherwise numbers are written unchanged. declaredPath(id:) with model3d:ply>usd —
ply usda — model3d supported structural attributes, coordinateSystem, dimensions Decoded by ORIX reader, written by ORIX writer, verified by re-reading with Model I/O. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. PLY states no unit: set the source unit, otherwise numbers are written unchanged. declaredPath(id:) with model3d:ply>usda —
ply usdc — model3d supported structural attributes, dimensions Decoded by ORIX reader, written by ORIX writer + Model I/O, verified by re-reading with SceneKit. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. Only the first set of texture coordinates is written. Model I/O cannot record metersPerUnit in a crate file: coordinates are converted to centimetres (USD's default) when the source unit is known. PLY states no unit: set the source unit, otherwise numbers are written unchanged. The crate layer is produced by Model I/O from the text layer ORIX writes: Model I/O re-authors the materials (their names and number can change). declaredPath(id:) with model3d:ply>usdc —
ply usdz — model3d supported structural attributes, coordinateSystem, dimensions Decoded by ORIX reader, written by ORIX writer, verified by re-reading with Model I/O. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. PLY states no unit: set the source unit, otherwise numbers are written unchanged. declaredPath(id:) with model3d:ply>usdz —
stl 3mf — model3d supported structural coordinateSystem, dimensions Decoded by ORIX reader, written by ORIX writer, verified by re-reading with Assimp. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. Vertex colours are not written. Texture coordinates are not written. STL states no unit: set the source unit, otherwise numbers are written unchanged. STL attribute bytes (colours) are ignored. declaredPath(id:) with model3d:stl>3mf —
stl dae — model3d supported structural attributes, coordinateSystem, dimensions Decoded by ORIX reader, written by Assimp, verified by re-reading with ORIX COLLADA checker. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. Only the first set of texture coordinates is written. STL states no unit: set the source unit, otherwise numbers are written unchanged. Written by Assimp's COLLADA exporter; metallic and roughness are not part of COLLADA 1.4. STL attribute bytes (colours) are ignored. declaredPath(id:) with model3d:stl>dae —
stl glb — model3d supported structural attributes, coordinateSystem, dimensions Decoded by ORIX reader, written by ORIX writer, verified by re-reading with ORIX glTF checker + Assimp. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. glTF is in metres: a known source unit is converted, an unknown one is written unchanged. STL states no unit: set the source unit, otherwise numbers are written unchanged. STL attribute bytes (colours) are ignored. declaredPath(id:) with model3d:stl>glb —
stl gltf — model3d supported structural attributes, coordinateSystem, dimensions Decoded by ORIX reader, written by ORIX writer, verified by re-reading with ORIX glTF checker + Assimp. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. glTF is in metres: a known source unit is converted, an unknown one is written unchanged. STL states no unit: set the source unit, otherwise numbers are written unchanged. STL attribute bytes (colours) are ignored. declaredPath(id:) with model3d:stl>gltf —
stl obj — model3d supported structural attributes, dimensions Decoded by ORIX reader, written by ORIX writer, verified by re-reading with Assimp. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. Vertex colours are not written. Only the first set of texture coordinates is written. Wavefront OBJ records neither a unit nor an up axis. STL attribute bytes (colours) are ignored. declaredPath(id:) with model3d:stl>obj —
stl ply — model3d supported structural attributes, dimensions Decoded by ORIX reader, written by ORIX writer, verified by re-reading with Assimp. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. Only the first set of texture coordinates is written. PLY records neither a unit nor an up axis. STL attribute bytes (colours) are ignored. declaredPath(id:) with model3d:stl>ply 998 000 (triangles; STL → PLY)
stl stl model.reprocess model3d supported structural dimensions Decoded by ORIX reader, written by ORIX writer, verified by re-reading with Assimp. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. Vertex colours are not written. Texture coordinates are not written. STL records neither a unit nor an up axis. Facet normals are recomputed from the triangle winding. STL attribute bytes (colours) are ignored. declaredPath(id:) with model3d:stl>stl#model.reprocess —
stl usd — model3d supported structural attributes, coordinateSystem, dimensions Decoded by ORIX reader, written by ORIX writer, verified by re-reading with Model I/O. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. STL states no unit: set the source unit, otherwise numbers are written unchanged. STL attribute bytes (colours) are ignored. declaredPath(id:) with model3d:stl>usd —
stl usda — model3d supported structural attributes, coordinateSystem, dimensions Decoded by ORIX reader, written by ORIX writer, verified by re-reading with Model I/O. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. STL states no unit: set the source unit, otherwise numbers are written unchanged. STL attribute bytes (colours) are ignored. declaredPath(id:) with model3d:stl>usda —
stl usdc — model3d supported structural attributes, dimensions Decoded by ORIX reader, written by ORIX writer + Model I/O, verified by re-reading with SceneKit. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. Only the first set of texture coordinates is written. Model I/O cannot record metersPerUnit in a crate file: coordinates are converted to centimetres (USD's default) when the source unit is known. STL states no unit: set the source unit, otherwise numbers are written unchanged. The crate layer is produced by Model I/O from the text layer ORIX writes: Model I/O re-authors the materials (their names and number can change). STL attribute bytes (colours) are ignored. declaredPath(id:) with model3d:stl>usdc —
stl usdz — model3d supported structural attributes, coordinateSystem, dimensions Decoded by ORIX reader, written by ORIX writer, verified by re-reading with Model I/O. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. STL states no unit: set the source unit, otherwise numbers are written unchanged. STL attribute bytes (colours) are ignored. declaredPath(id:) with model3d:stl>usdz —
usd 3mf — model3d supported partial coordinateSystem, dimensions, materials Decoded by Model I/O, written by ORIX writer, verified by re-reading with Assimp. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. The node hierarchy is flattened to one placed object per mesh. Only the base colour of each material is kept. 3MF cannot hold textures: they are dropped. Vertex colours are not written. Texture coordinates are not written. USD files that reference other files outside their folder are refused; variants, instancing and subdivision are resolved by Model I/O. declaredPath(id:) with model3d:usd>3mf —
usd dae — model3d supported partial attributes, coordinateSystem, dimensions, materials, textures Decoded by Model I/O, written by Assimp, verified by re-reading with ORIX COLLADA checker. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. PBR materials are approximated with COLLADA Phong values. Textures are written as files next to the model; the output is then a folder. Only the first set of texture coordinates is written. Written by Assimp's COLLADA exporter; metallic and roughness are not part of COLLADA 1.4. USD files that reference other files outside their folder are refused; variants, instancing and subdivision are resolved by Model I/O. declaredPath(id:) with model3d:usd>dae —
usd glb — model3d supported structural attributes, coordinateSystem, dimensions, materials, textures Decoded by Model I/O, written by ORIX writer, verified by re-reading with ORIX glTF checker + Assimp. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. Only PNG and JPEG textures are carried. glTF is in metres: a known source unit is converted, an unknown one is written unchanged. USD files that reference other files outside their folder are refused; variants, instancing and subdivision are resolved by Model I/O. declaredPath(id:) with model3d:usd>glb —
usd gltf — model3d supported structural attributes, coordinateSystem, dimensions, materials, textures Decoded by Model I/O, written by ORIX writer, verified by re-reading with ORIX glTF checker + Assimp. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. Textures are written as files next to the model; the output is then a folder. Only PNG and JPEG textures are carried. glTF is in metres: a known source unit is converted, an unknown one is written unchanged. USD files that reference other files outside their folder are refused; variants, instancing and subdivision are resolved by Model I/O. declaredPath(id:) with model3d:usd>gltf —
usd obj — model3d supported partial attributes, dimensions, materials, textures Decoded by Model I/O, written by ORIX writer, verified by re-reading with Assimp. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. The node hierarchy is flattened: every mesh is written in world coordinates. PBR materials are approximated with MTL Phong values (plus Pr/Pm lines). Textures are written as files next to the model; the output is then a folder. Vertex colours are not written. Only the first set of texture coordinates is written. Wavefront OBJ records neither a unit nor an up axis. USD files that reference other files outside their folder are refused; variants, instancing and subdivision are resolved by Model I/O. declaredPath(id:) with model3d:usd>obj —
usd ply — model3d supported partial attributes, dimensions Decoded by Model I/O, written by ORIX writer, verified by re-reading with Assimp. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. The node hierarchy is flattened: every mesh is written in world coordinates. PLY has no materials: they are dropped. PLY cannot hold textures: they are dropped. Only the first set of texture coordinates is written. PLY records neither a unit nor an up axis. USD files that reference other files outside their folder are refused; variants, instancing and subdivision are resolved by Model I/O. declaredPath(id:) with model3d:usd>ply —
usd stl — model3d supported partial dimensions Decoded by Model I/O, written by ORIX writer, verified by re-reading with Assimp. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. The node hierarchy is flattened: every mesh is written in world coordinates. STL has no materials: they are dropped. STL cannot hold textures: they are dropped. Vertex colours are not written. Texture coordinates are not written. STL records neither a unit nor an up axis. Facet normals are recomputed from the triangle winding. USD files that reference other files outside their folder are refused; variants, instancing and subdivision are resolved by Model I/O. declaredPath(id:) with model3d:usd>stl —
usd usd model.reprocess model3d supported structural attributes, coordinateSystem, dimensions, materials, textures Decoded by Model I/O, written by ORIX writer, verified by re-reading with Model I/O. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. Textures are written as files next to the model; the output is then a folder. Only PNG and JPEG textures are carried. USD files that reference other files outside their folder are refused; variants, instancing and subdivision are resolved by Model I/O. declaredPath(id:) with model3d:usd>usd#model.reprocess —
usd usda — model3d supported structural attributes, coordinateSystem, dimensions, materials, textures Decoded by Model I/O, written by ORIX writer, verified by re-reading with Model I/O. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. Textures are written as files next to the model; the output is then a folder. Only PNG and JPEG textures are carried. USD files that reference other files outside their folder are refused; variants, instancing and subdivision are resolved by Model I/O. declaredPath(id:) with model3d:usd>usda —
usd usdc — model3d supported structural attributes, dimensions, materials, textures Decoded by Model I/O, written by ORIX writer + Model I/O, verified by re-reading with SceneKit. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. Textures are written as files next to the model; the output is then a folder. Only PNG and JPEG textures are carried. Only the first set of texture coordinates is written. Model I/O cannot record metersPerUnit in a crate file: coordinates are converted to centimetres (USD's default) when the source unit is known. The crate layer is produced by Model I/O from the text layer ORIX writes: Model I/O re-authors the materials (their names and number can change). USD files that reference other files outside their folder are refused; variants, instancing and subdivision are resolved by Model I/O. declaredPath(id:) with model3d:usd>usdc —
usd usdz — model3d supported structural attributes, coordinateSystem, dimensions, materials, textures Decoded by Model I/O, written by ORIX writer, verified by re-reading with Model I/O. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. Only PNG and JPEG textures are carried. USD files that reference other files outside their folder are refused; variants, instancing and subdivision are resolved by Model I/O. declaredPath(id:) with model3d:usd>usdz —
usda 3mf — model3d supported partial coordinateSystem, dimensions, materials Decoded by Model I/O, written by ORIX writer, verified by re-reading with Assimp. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. The node hierarchy is flattened to one placed object per mesh. Only the base colour of each material is kept. 3MF cannot hold textures: they are dropped. Vertex colours are not written. Texture coordinates are not written. USD files that reference other files outside their folder are refused; variants, instancing and subdivision are resolved by Model I/O. declaredPath(id:) with model3d:usda>3mf —
usda dae — model3d supported partial attributes, coordinateSystem, dimensions, materials, textures Decoded by Model I/O, written by Assimp, verified by re-reading with ORIX COLLADA checker. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. PBR materials are approximated with COLLADA Phong values. Textures are written as files next to the model; the output is then a folder. Only the first set of texture coordinates is written. Written by Assimp's COLLADA exporter; metallic and roughness are not part of COLLADA 1.4. USD files that reference other files outside their folder are refused; variants, instancing and subdivision are resolved by Model I/O. declaredPath(id:) with model3d:usda>dae —
usda glb — model3d supported structural attributes, coordinateSystem, dimensions, materials, textures Decoded by Model I/O, written by ORIX writer, verified by re-reading with ORIX glTF checker + Assimp. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. Only PNG and JPEG textures are carried. glTF is in metres: a known source unit is converted, an unknown one is written unchanged. USD files that reference other files outside their folder are refused; variants, instancing and subdivision are resolved by Model I/O. declaredPath(id:) with model3d:usda>glb —
usda gltf — model3d supported structural attributes, coordinateSystem, dimensions, materials, textures Decoded by Model I/O, written by ORIX writer, verified by re-reading with ORIX glTF checker + Assimp. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. Textures are written as files next to the model; the output is then a folder. Only PNG and JPEG textures are carried. glTF is in metres: a known source unit is converted, an unknown one is written unchanged. USD files that reference other files outside their folder are refused; variants, instancing and subdivision are resolved by Model I/O. declaredPath(id:) with model3d:usda>gltf —
usda obj — model3d supported partial attributes, dimensions, materials, textures Decoded by Model I/O, written by ORIX writer, verified by re-reading with Assimp. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. The node hierarchy is flattened: every mesh is written in world coordinates. PBR materials are approximated with MTL Phong values (plus Pr/Pm lines). Textures are written as files next to the model; the output is then a folder. Vertex colours are not written. Only the first set of texture coordinates is written. Wavefront OBJ records neither a unit nor an up axis. USD files that reference other files outside their folder are refused; variants, instancing and subdivision are resolved by Model I/O. declaredPath(id:) with model3d:usda>obj —
usda ply — model3d supported partial attributes, dimensions Decoded by Model I/O, written by ORIX writer, verified by re-reading with Assimp. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. The node hierarchy is flattened: every mesh is written in world coordinates. PLY has no materials: they are dropped. PLY cannot hold textures: they are dropped. Only the first set of texture coordinates is written. PLY records neither a unit nor an up axis. USD files that reference other files outside their folder are refused; variants, instancing and subdivision are resolved by Model I/O. declaredPath(id:) with model3d:usda>ply —
usda stl — model3d supported partial dimensions Decoded by Model I/O, written by ORIX writer, verified by re-reading with Assimp. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. The node hierarchy is flattened: every mesh is written in world coordinates. STL has no materials: they are dropped. STL cannot hold textures: they are dropped. Vertex colours are not written. Texture coordinates are not written. STL records neither a unit nor an up axis. Facet normals are recomputed from the triangle winding. USD files that reference other files outside their folder are refused; variants, instancing and subdivision are resolved by Model I/O. declaredPath(id:) with model3d:usda>stl —
usda usd — model3d supported structural attributes, coordinateSystem, dimensions, materials, textures Decoded by Model I/O, written by ORIX writer, verified by re-reading with Model I/O. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. Textures are written as files next to the model; the output is then a folder. Only PNG and JPEG textures are carried. USD files that reference other files outside their folder are refused; variants, instancing and subdivision are resolved by Model I/O. declaredPath(id:) with model3d:usda>usd —
usda usda model.reprocess model3d supported structural attributes, coordinateSystem, dimensions, materials, textures Decoded by Model I/O, written by ORIX writer, verified by re-reading with Model I/O. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. Textures are written as files next to the model; the output is then a folder. Only PNG and JPEG textures are carried. USD files that reference other files outside their folder are refused; variants, instancing and subdivision are resolved by Model I/O. declaredPath(id:) with model3d:usda>usda#model.reprocess —
usda usdc — model3d supported structural attributes, dimensions, materials, textures Decoded by Model I/O, written by ORIX writer + Model I/O, verified by re-reading with SceneKit. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. Textures are written as files next to the model; the output is then a folder. Only PNG and JPEG textures are carried. Only the first set of texture coordinates is written. Model I/O cannot record metersPerUnit in a crate file: coordinates are converted to centimetres (USD's default) when the source unit is known. The crate layer is produced by Model I/O from the text layer ORIX writes: Model I/O re-authors the materials (their names and number can change). USD files that reference other files outside their folder are refused; variants, instancing and subdivision are resolved by Model I/O. declaredPath(id:) with model3d:usda>usdc —
usda usdz — model3d supported structural attributes, coordinateSystem, dimensions, materials, textures Decoded by Model I/O, written by ORIX writer, verified by re-reading with Model I/O. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. Only PNG and JPEG textures are carried. USD files that reference other files outside their folder are refused; variants, instancing and subdivision are resolved by Model I/O. declaredPath(id:) with model3d:usda>usdz —
usdc 3mf — model3d supported partial coordinateSystem, dimensions, materials Decoded by Model I/O, written by ORIX writer, verified by re-reading with Assimp. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. The node hierarchy is flattened to one placed object per mesh. Only the base colour of each material is kept. 3MF cannot hold textures: they are dropped. Vertex colours are not written. Texture coordinates are not written. USD files that reference other files outside their folder are refused; variants, instancing and subdivision are resolved by Model I/O. declaredPath(id:) with model3d:usdc>3mf —
usdc dae — model3d supported partial attributes, coordinateSystem, dimensions, materials, textures Decoded by Model I/O, written by Assimp, verified by re-reading with ORIX COLLADA checker. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. PBR materials are approximated with COLLADA Phong values. Textures are written as files next to the model; the output is then a folder. Only the first set of texture coordinates is written. Written by Assimp's COLLADA exporter; metallic and roughness are not part of COLLADA 1.4. USD files that reference other files outside their folder are refused; variants, instancing and subdivision are resolved by Model I/O. declaredPath(id:) with model3d:usdc>dae —
usdc glb — model3d supported structural attributes, coordinateSystem, dimensions, materials, textures Decoded by Model I/O, written by ORIX writer, verified by re-reading with ORIX glTF checker + Assimp. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. Only PNG and JPEG textures are carried. glTF is in metres: a known source unit is converted, an unknown one is written unchanged. USD files that reference other files outside their folder are refused; variants, instancing and subdivision are resolved by Model I/O. declaredPath(id:) with model3d:usdc>glb —
usdc gltf — model3d supported structural attributes, coordinateSystem, dimensions, materials, textures Decoded by Model I/O, written by ORIX writer, verified by re-reading with ORIX glTF checker + Assimp. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. Textures are written as files next to the model; the output is then a folder. Only PNG and JPEG textures are carried. glTF is in metres: a known source unit is converted, an unknown one is written unchanged. USD files that reference other files outside their folder are refused; variants, instancing and subdivision are resolved by Model I/O. declaredPath(id:) with model3d:usdc>gltf —
usdc obj — model3d supported partial attributes, dimensions, materials, textures Decoded by Model I/O, written by ORIX writer, verified by re-reading with Assimp. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. The node hierarchy is flattened: every mesh is written in world coordinates. PBR materials are approximated with MTL Phong values (plus Pr/Pm lines). Textures are written as files next to the model; the output is then a folder. Vertex colours are not written. Only the first set of texture coordinates is written. Wavefront OBJ records neither a unit nor an up axis. USD files that reference other files outside their folder are refused; variants, instancing and subdivision are resolved by Model I/O. declaredPath(id:) with model3d:usdc>obj —
usdc ply — model3d supported partial attributes, dimensions Decoded by Model I/O, written by ORIX writer, verified by re-reading with Assimp. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. The node hierarchy is flattened: every mesh is written in world coordinates. PLY has no materials: they are dropped. PLY cannot hold textures: they are dropped. Only the first set of texture coordinates is written. PLY records neither a unit nor an up axis. USD files that reference other files outside their folder are refused; variants, instancing and subdivision are resolved by Model I/O. declaredPath(id:) with model3d:usdc>ply —
usdc stl — model3d supported partial dimensions Decoded by Model I/O, written by ORIX writer, verified by re-reading with Assimp. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. The node hierarchy is flattened: every mesh is written in world coordinates. STL has no materials: they are dropped. STL cannot hold textures: they are dropped. Vertex colours are not written. Texture coordinates are not written. STL records neither a unit nor an up axis. Facet normals are recomputed from the triangle winding. USD files that reference other files outside their folder are refused; variants, instancing and subdivision are resolved by Model I/O. declaredPath(id:) with model3d:usdc>stl —
usdc usd — model3d supported structural attributes, coordinateSystem, dimensions, materials, textures Decoded by Model I/O, written by ORIX writer, verified by re-reading with Model I/O. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. Textures are written as files next to the model; the output is then a folder. Only PNG and JPEG textures are carried. USD files that reference other files outside their folder are refused; variants, instancing and subdivision are resolved by Model I/O. declaredPath(id:) with model3d:usdc>usd —
usdc usda — model3d supported structural attributes, coordinateSystem, dimensions, materials, textures Decoded by Model I/O, written by ORIX writer, verified by re-reading with Model I/O. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. Textures are written as files next to the model; the output is then a folder. Only PNG and JPEG textures are carried. USD files that reference other files outside their folder are refused; variants, instancing and subdivision are resolved by Model I/O. declaredPath(id:) with model3d:usdc>usda —
usdc usdc model.reprocess model3d supported structural attributes, dimensions, materials, textures Decoded by Model I/O, written by ORIX writer + Model I/O, verified by re-reading with SceneKit. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. Textures are written as files next to the model; the output is then a folder. Only PNG and JPEG textures are carried. Only the first set of texture coordinates is written. Model I/O cannot record metersPerUnit in a crate file: coordinates are converted to centimetres (USD's default) when the source unit is known. The crate layer is produced by Model I/O from the text layer ORIX writes: Model I/O re-authors the materials (their names and number can change). USD files that reference other files outside their folder are refused; variants, instancing and subdivision are resolved by Model I/O. declaredPath(id:) with model3d:usdc>usdc#model.reprocess —
usdc usdz — model3d supported structural attributes, coordinateSystem, dimensions, materials, textures Decoded by Model I/O, written by ORIX writer, verified by re-reading with Model I/O. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. Only PNG and JPEG textures are carried. USD files that reference other files outside their folder are refused; variants, instancing and subdivision are resolved by Model I/O. declaredPath(id:) with model3d:usdc>usdz —
usdz 3mf — model3d supported partial coordinateSystem, dimensions, materials Decoded by Model I/O, written by ORIX writer, verified by re-reading with Assimp. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. The node hierarchy is flattened to one placed object per mesh. Only the base colour of each material is kept. 3MF cannot hold textures: they are dropped. Vertex colours are not written. Texture coordinates are not written. USD files that reference other files outside their folder are refused; variants, instancing and subdivision are resolved by Model I/O. declaredPath(id:) with model3d:usdz>3mf —
usdz dae — model3d supported partial attributes, coordinateSystem, dimensions, materials, textures Decoded by Model I/O, written by Assimp, verified by re-reading with ORIX COLLADA checker. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. PBR materials are approximated with COLLADA Phong values. Textures are written as files next to the model; the output is then a folder. Only the first set of texture coordinates is written. Written by Assimp's COLLADA exporter; metallic and roughness are not part of COLLADA 1.4. USD files that reference other files outside their folder are refused; variants, instancing and subdivision are resolved by Model I/O. declaredPath(id:) with model3d:usdz>dae —
usdz glb — model3d supported structural attributes, coordinateSystem, dimensions, materials, textures Decoded by Model I/O, written by ORIX writer, verified by re-reading with ORIX glTF checker + Assimp. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. Only PNG and JPEG textures are carried. glTF is in metres: a known source unit is converted, an unknown one is written unchanged. USD files that reference other files outside their folder are refused; variants, instancing and subdivision are resolved by Model I/O. declaredPath(id:) with model3d:usdz>glb —
usdz gltf — model3d supported structural attributes, coordinateSystem, dimensions, materials, textures Decoded by Model I/O, written by ORIX writer, verified by re-reading with ORIX glTF checker + Assimp. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. Textures are written as files next to the model; the output is then a folder. Only PNG and JPEG textures are carried. glTF is in metres: a known source unit is converted, an unknown one is written unchanged. USD files that reference other files outside their folder are refused; variants, instancing and subdivision are resolved by Model I/O. declaredPath(id:) with model3d:usdz>gltf —
usdz obj — model3d supported partial attributes, dimensions, materials, textures Decoded by Model I/O, written by ORIX writer, verified by re-reading with Assimp. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. The node hierarchy is flattened: every mesh is written in world coordinates. PBR materials are approximated with MTL Phong values (plus Pr/Pm lines). Textures are written as files next to the model; the output is then a folder. Vertex colours are not written. Only the first set of texture coordinates is written. Wavefront OBJ records neither a unit nor an up axis. USD files that reference other files outside their folder are refused; variants, instancing and subdivision are resolved by Model I/O. declaredPath(id:) with model3d:usdz>obj —
usdz ply — model3d supported partial attributes, dimensions Decoded by Model I/O, written by ORIX writer, verified by re-reading with Assimp. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. The node hierarchy is flattened: every mesh is written in world coordinates. PLY has no materials: they are dropped. PLY cannot hold textures: they are dropped. Only the first set of texture coordinates is written. PLY records neither a unit nor an up axis. USD files that reference other files outside their folder are refused; variants, instancing and subdivision are resolved by Model I/O. declaredPath(id:) with model3d:usdz>ply —
usdz stl — model3d supported partial dimensions Decoded by Model I/O, written by ORIX writer, verified by re-reading with Assimp. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. The node hierarchy is flattened: every mesh is written in world coordinates. STL has no materials: they are dropped. STL cannot hold textures: they are dropped. Vertex colours are not written. Texture coordinates are not written. STL records neither a unit nor an up axis. Facet normals are recomputed from the triangle winding. USD files that reference other files outside their folder are refused; variants, instancing and subdivision are resolved by Model I/O. declaredPath(id:) with model3d:usdz>stl —
usdz usd — model3d supported structural attributes, coordinateSystem, dimensions, materials, textures Decoded by Model I/O, written by ORIX writer, verified by re-reading with Model I/O. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. Textures are written as files next to the model; the output is then a folder. Only PNG and JPEG textures are carried. USD files that reference other files outside their folder are refused; variants, instancing and subdivision are resolved by Model I/O. declaredPath(id:) with model3d:usdz>usd —
usdz usda — model3d supported structural attributes, coordinateSystem, dimensions, materials, textures Decoded by Model I/O, written by ORIX writer, verified by re-reading with Model I/O. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. Textures are written as files next to the model; the output is then a folder. Only PNG and JPEG textures are carried. USD files that reference other files outside their folder are refused; variants, instancing and subdivision are resolved by Model I/O. declaredPath(id:) with model3d:usdz>usda —
usdz usdc — model3d supported structural attributes, dimensions, materials, textures Decoded by Model I/O, written by ORIX writer + Model I/O, verified by re-reading with SceneKit. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. Textures are written as files next to the model; the output is then a folder. Only PNG and JPEG textures are carried. Only the first set of texture coordinates is written. Model I/O cannot record metersPerUnit in a crate file: coordinates are converted to centimetres (USD's default) when the source unit is known. The crate layer is produced by Model I/O from the text layer ORIX writes: Model I/O re-authors the materials (their names and number can change). USD files that reference other files outside their folder are refused; variants, instancing and subdivision are resolved by Model I/O. declaredPath(id:) with model3d:usdz>usdc —
usdz usdz model.reprocess model3d supported structural attributes, coordinateSystem, dimensions, materials, textures Decoded by Model I/O, written by ORIX writer, verified by re-reading with Model I/O. Animations, skins, morph targets, cameras and lights are not converted. Polygons are triangulated; point and line primitives are dropped. Only PNG and JPEG textures are carried. USD files that reference other files outside their folder are refused; variants, instancing and subdivision are resolved by Model I/O. declaredPath(id:) with model3d:usdz>usdz#model.reprocess —

# font sources (11 paths)

Source Destination Operation Engine Status Fidelity Preserves Limitations Tests Performance
otf woff — font supported lossless metadata, vectorGeometry From a font collection (TTC / OTC), one font is converted — the first unless you choose another. The font's embedding permissions (OS/2 fsType) and licence are shown before converting and are never altered: you are responsible for having the right to use the font on the web. MatrixTests.everyDeclaredPathConvertsAndValidates(path:) —
otf woff2 — font supported lossless metadata, vectorGeometry WOFF2 stores the glyph tables (glyf, loca) in a transformed form: outlines and metrics are identical, but those tables and the font's checksum fields are not byte-identical to the original, and a digital signature (DSIG) is removed. From a font collection (TTC / OTC), one font is converted — the first unless you choose another. The font's embedding permissions (OS/2 fsType) and licence are shown before converting and are never altered: you are responsible for having the right to use the font on the web. MatrixTests.everyDeclaredPathConvertsAndValidates(path:) —
ttf folder font.split-collection font supported lossless metadata, vectorGeometry Applies to font collections (.ttc / .otc): each member becomes a standalone .ttf or .otf file; tables shared by several members are copied into each file. A file that holds a single font is refused. EngineFontTests/MatrixTests (suite) —
ttf woff — font supported lossless metadata, vectorGeometry From a font collection (TTC / OTC), one font is converted — the first unless you choose another. The font's embedding permissions (OS/2 fsType) and licence are shown before converting and are never altered: you are responsible for having the right to use the font on the web. MatrixTests.everyDeclaredPathConvertsAndValidates(path:) —
ttf woff2 — font supported lossless metadata, vectorGeometry WOFF2 stores the glyph tables (glyf, loca) in a transformed form: outlines and metrics are identical, but those tables and the font's checksum fields are not byte-identical to the original, and a digital signature (DSIG) is removed. From a font collection (TTC / OTC), one font is converted — the first unless you choose another. The font's embedding permissions (OS/2 fsType) and licence are shown before converting and are never altered: you are responsible for having the right to use the font on the web. MatrixTests.everyDeclaredPathConvertsAndValidates(path:) —
woff otf — font supported lossless metadata, vectorGeometry The outlines are not converted: a font with TrueType outlines is saved as .ttf, a font with PostScript (CFF) outlines as .otf. When you ask for the other one, the file gets the extension that matches the font, with a notice. The WOFF metadata and private-data blocks, when present, are not part of the font and are not carried over. MatrixTests.everyDeclaredPathConvertsAndValidates(path:) —
woff ttf — font supported lossless metadata, vectorGeometry The outlines are not converted: a font with TrueType outlines is saved as .ttf, a font with PostScript (CFF) outlines as .otf. When you ask for the other one, the file gets the extension that matches the font, with a notice. The WOFF metadata and private-data blocks, when present, are not part of the font and are not carried over. MatrixTests.everyDeclaredPathConvertsAndValidates(path:) —
woff woff2 — font supported lossless metadata, vectorGeometry WOFF2 stores the glyph tables (glyf, loca) in a transformed form: outlines and metrics are identical, but those tables and the font's checksum fields are not byte-identical to the original, and a digital signature (DSIG) is removed. The WOFF private-data block, when present, is not carried over (the metadata block is). The font's embedding permissions (OS/2 fsType) and licence are shown before converting and are never altered: you are responsible for having the right to use the font on the web. MatrixTests.everyDeclaredPathConvertsAndValidates(path:) —
woff2 otf — font supported lossless metadata, vectorGeometry The outlines are not converted: a font with TrueType outlines is saved as .ttf, a font with PostScript (CFF) outlines as .otf. When you ask for the other one, the file gets the extension that matches the font, with a notice. WOFF2 stores the glyph tables (glyf, loca) in a transformed form: outlines and metrics are identical, but those tables and the font's checksum fields are not byte-identical to the original, and a digital signature (DSIG) is removed. From a WOFF2 collection, one font is extracted — the first unless you choose another. The WOFF metadata and private-data blocks, when present, are not part of the font and are not carried over. MatrixTests.everyDeclaredPathConvertsAndValidates(path:) —
woff2 ttf — font supported lossless metadata, vectorGeometry The outlines are not converted: a font with TrueType outlines is saved as .ttf, a font with PostScript (CFF) outlines as .otf. When you ask for the other one, the file gets the extension that matches the font, with a notice. WOFF2 stores the glyph tables (glyf, loca) in a transformed form: outlines and metrics are identical, but those tables and the font's checksum fields are not byte-identical to the original, and a digital signature (DSIG) is removed. From a WOFF2 collection, one font is extracted — the first unless you choose another. The WOFF metadata and private-data blocks, when present, are not part of the font and are not carried over. MatrixTests.everyDeclaredPathConvertsAndValidates(path:) —
woff2 woff — font supported lossless metadata, vectorGeometry WOFF2 stores the glyph tables (glyf, loca) in a transformed form: outlines and metrics are identical, but those tables and the font's checksum fields are not byte-identical to the original, and a digital signature (DSIG) is removed. WOFF cannot hold a collection: one font is converted. The WOFF metadata and private-data blocks, when present, are not part of the font and are not carried over. MatrixTests.everyDeclaredPathConvertsAndValidates(path:) —

# subtitle sources (30 paths)

Source Destination Operation Engine Status Fidelity Preserves Limitations Tests Performance
ass ass subtitle.clean subtitle supported structural textStyles, timing Clean-up edits the text: review the result. Cues whose text was edited lose markup this app does not model (kept for untouched cues). Non-dialogue events (Picture, Sound, Movie, Command) are not kept. EngineSubtitleTests/MatrixTests (suite) —
ass ass subtitle.retime subtitle supported structural textStyles, timing Only the timing is changed; text and markup are written back as they were read. Times are stored in hundredths of a second; the script header, styles and embedded files are kept, non-dialogue events (Picture, Sound, Movie, Command) are not. EngineSubtitleTests/MatrixTests (suite) —
ass srt — subtitle supported structural textStyles, timing SubRip holds text, timing and simple bold / italic / underline / colour tags only. Style definitions, positions, karaoke, effects, drawings and embedded fonts of the script are not carried over. MatrixTests.everyDeclaredPathConvertsAndValidates(path:) —
ass ssa — subtitle supported structural textStyles, timing Times are stored in hundredths of a second: milliseconds are rounded to the nearest hundredth (halves up). SSA has no underline or strike-through, and no scale, spacing or rotation in its styles. Override tags are kept as written, except alignment tags, which are translated; tags unknown to the other dialect are ignored by players. MatrixTests.everyDeclaredPathConvertsAndValidates(path:) —
ass sub — subtitle supported structural timing MicroDVD counts frames: times are rounded to the nearest frame of the frame rate you give (23.976 fps is assumed otherwise). MicroDVD styles whole lines; positions, karaoke and effects are not carried over. Style definitions, drawings and embedded fonts of the script are not carried over. MatrixTests.everyDeclaredPathConvertsAndValidates(path:) —
ass vtt — subtitle supported structural textStyles, timing WebVTT has no font tag and no strike-through; colours are written as cue classes with a STYLE block. Style definitions, karaoke, effects, drawings and embedded fonts are not carried over; alignment is kept and exact positions are approximated. MatrixTests.everyDeclaredPathConvertsAndValidates(path:) —
srt ass — subtitle supported structural textStyles, timing A single default style is written (font, size and resolution are settings); the result is not a designed script. Times are stored in hundredths of a second: milliseconds are rounded to the nearest hundredth (halves up). MatrixTests.everyDeclaredPathConvertsAndValidates(path:) —
srt srt subtitle.clean subtitle supported structural textStyles, timing Clean-up edits the text: review the result. Cues whose text was edited lose markup this app does not model (kept for untouched cues). EngineSubtitleTests/MatrixTests (suite) —
srt srt subtitle.retime subtitle supported structural textStyles, timing Only the timing is changed; text and markup are written back as they were read. EngineSubtitleTests/MatrixTests (suite) —
srt ssa — subtitle supported structural textStyles, timing A single default style is written (font, size and resolution are settings); the result is not a designed script. Times are stored in hundredths of a second: milliseconds are rounded to the nearest hundredth (halves up). SSA has no underline or strike-through, and no scale, spacing or rotation in its styles. MatrixTests.everyDeclaredPathConvertsAndValidates(path:) —
srt sub — subtitle supported structural timing MicroDVD counts frames: times are rounded to the nearest frame of the frame rate you give (23.976 fps is assumed otherwise). MicroDVD styles whole lines; positions, karaoke and effects are not carried over. MatrixTests.everyDeclaredPathConvertsAndValidates(path:) —
srt vtt — subtitle supported structural textStyles, timing WebVTT has no font tag and no strike-through; colours are written as cue classes with a STYLE block. MatrixTests.everyDeclaredPathConvertsAndValidates(path:) —
ssa ass — subtitle supported structural textStyles, timing Times are stored in hundredths of a second: milliseconds are rounded to the nearest hundredth (halves up). Override tags are kept as written, except alignment tags, which are translated; tags unknown to the other dialect are ignored by players. MatrixTests.everyDeclaredPathConvertsAndValidates(path:) —
ssa srt — subtitle supported structural textStyles, timing SubRip holds text, timing and simple bold / italic / underline / colour tags only. Style definitions, positions, karaoke, effects, drawings and embedded fonts of the script are not carried over. MatrixTests.everyDeclaredPathConvertsAndValidates(path:) —
ssa ssa subtitle.clean subtitle supported structural textStyles, timing Clean-up edits the text: review the result. Cues whose text was edited lose markup this app does not model (kept for untouched cues). Non-dialogue events (Picture, Sound, Movie, Command) are not kept. EngineSubtitleTests/MatrixTests (suite) —
ssa ssa subtitle.retime subtitle supported structural textStyles, timing Only the timing is changed; text and markup are written back as they were read. Times are stored in hundredths of a second; the script header, styles and embedded files are kept, non-dialogue events (Picture, Sound, Movie, Command) are not. EngineSubtitleTests/MatrixTests (suite) —
ssa sub — subtitle supported structural timing MicroDVD counts frames: times are rounded to the nearest frame of the frame rate you give (23.976 fps is assumed otherwise). MicroDVD styles whole lines; positions, karaoke and effects are not carried over. Style definitions, drawings and embedded fonts of the script are not carried over. MatrixTests.everyDeclaredPathConvertsAndValidates(path:) —
ssa vtt — subtitle supported structural textStyles, timing WebVTT has no font tag and no strike-through; colours are written as cue classes with a STYLE block. Style definitions, karaoke, effects, drawings and embedded fonts are not carried over; alignment is kept and exact positions are approximated. MatrixTests.everyDeclaredPathConvertsAndValidates(path:) —
sub ass — subtitle supported structural textStyles, timing A single default style is written (font, size and resolution are settings); the result is not a designed script. Times are stored in hundredths of a second: milliseconds are rounded to the nearest hundredth (halves up). MicroDVD stores frame numbers: without a frame-rate header or setting, 23.976 fps is assumed to compute times. MatrixTests.everyDeclaredPathConvertsAndValidates(path:) —
sub srt — subtitle supported structural textStyles, timing SubRip holds text, timing and simple bold / italic / underline / colour tags only. MicroDVD stores frame numbers: without a frame-rate header or setting, 23.976 fps is assumed to compute times. MatrixTests.everyDeclaredPathConvertsAndValidates(path:) —
sub ssa — subtitle supported structural textStyles, timing A single default style is written (font, size and resolution are settings); the result is not a designed script. Times are stored in hundredths of a second: milliseconds are rounded to the nearest hundredth (halves up). SSA has no underline or strike-through, and no scale, spacing or rotation in its styles. MicroDVD stores frame numbers: without a frame-rate header or setting, 23.976 fps is assumed to compute times. MatrixTests.everyDeclaredPathConvertsAndValidates(path:) —
sub sub subtitle.clean subtitle supported structural textStyles, timing Clean-up edits the text: review the result. Cues whose text was edited lose markup this app does not model (kept for untouched cues). EngineSubtitleTests/MatrixTests (suite) —
sub sub subtitle.retime subtitle supported structural textStyles, timing Only the timing is changed; text and markup are written back as they were read. Times are rounded to whole frames of the chosen frame rate. EngineSubtitleTests/MatrixTests (suite) —
sub vtt — subtitle supported structural textStyles, timing WebVTT has no font tag and no strike-through; colours are written as cue classes with a STYLE block. MicroDVD stores frame numbers: without a frame-rate header or setting, 23.976 fps is assumed to compute times. MatrixTests.everyDeclaredPathConvertsAndValidates(path:) —
vtt ass — subtitle supported structural textStyles, timing A single default style is written (font, size and resolution are settings); the result is not a designed script. Times are stored in hundredths of a second: milliseconds are rounded to the nearest hundredth (halves up). WebVTT regions, STYLE blocks and speaker annotations are not carried over. MatrixTests.everyDeclaredPathConvertsAndValidates(path:) —
vtt srt — subtitle supported structural textStyles, timing SubRip holds text, timing and simple bold / italic / underline / colour tags only. WebVTT positions, regions, STYLE blocks and speaker annotations are not carried over. MatrixTests.everyDeclaredPathConvertsAndValidates(path:) —
vtt ssa — subtitle supported structural textStyles, timing A single default style is written (font, size and resolution are settings); the result is not a designed script. Times are stored in hundredths of a second: milliseconds are rounded to the nearest hundredth (halves up). SSA has no underline or strike-through, and no scale, spacing or rotation in its styles. WebVTT regions, STYLE blocks and speaker annotations are not carried over. MatrixTests.everyDeclaredPathConvertsAndValidates(path:) —
vtt sub — subtitle supported structural timing MicroDVD counts frames: times are rounded to the nearest frame of the frame rate you give (23.976 fps is assumed otherwise). MicroDVD styles whole lines; positions, karaoke and effects are not carried over. MatrixTests.everyDeclaredPathConvertsAndValidates(path:) —
vtt vtt subtitle.clean subtitle supported structural textStyles, timing Clean-up edits the text: review the result. Cues whose text was edited lose markup this app does not model (kept for untouched cues). NOTE comment blocks are not kept. EngineSubtitleTests/MatrixTests (suite) —
vtt vtt subtitle.retime subtitle supported structural textStyles, timing Only the timing is changed; text and markup are written back as they were read. NOTE comment blocks are not kept. EngineSubtitleTests/MatrixTests (suite) —

# ebook sources (14 paths)

Source Destination Operation Engine Status Fidelity Preserves Limitations Tests Performance
epub doc — ebooks supported structural images, links, metadata, tables, textStyles The document is rebuilt from its structure (headings, lists, tables, images, links, notes); the source's visual styling is replaced by default styles. Word 97–2004 is a legacy format: newer features (modern text effects, content controls) have no equivalent. DRM-protected books are refused. Obfuscated embedded fonts are not carried over. DocumentPathTests.declaredPath —
epub docx — ebooks supported structural chapters, images, links, metadata, tables, textStyles The document is rebuilt from its structure (headings, lists, tables, images, links, notes); the source's visual styling is replaced by default styles. DRM-protected books are refused. Obfuscated embedded fonts are not carried over. DocumentPathTests.declaredPath —
epub folder epub.chapters ebooks supported structural chapters, images, links, tables, textStyles One file per spine item, in reading order (a spine item is not always one chapter). HTML chapters embed their images; Markdown chapters reference images written to a media folder. The book's stylesheets and fonts are not carried. DRM-protected books are refused. EbookTests.chapters —
epub folder media.extract ebooks supported lossless images Images, audio and video listed in the book's manifest are copied byte for byte. DRM-protected books are refused. EbookTests.coverAndMedia —
epub html — ebooks supported structural chapters, images, links, metadata, tables, textStyles Page layout, headers and footers are dropped. Images are embedded in the HTML file as data URIs. DRM-protected books are refused. Obfuscated embedded fonts are not carried over. DocumentPathTests.declaredPath 0.40 s (time; Pandoc EPUB→HTML)
epub jpeg epub.cover ebooks supported lossy dimensions Works only when the book declares a cover image. The file is copied unchanged when it already is a JPEG; otherwise it is re-encoded. EbookTests.coverAndMedia —
epub latex — ebooks supported structural chapters, links, metadata, tables, textStyles Images are kept as references only; the image files are not written (use “Extract media” on the source). Page layout, fonts, colours, headers and footers are dropped; complex or merged-cell tables are simplified. The result is LaTeX source; it is not compiled to PDF. DRM-protected books are refused. Obfuscated embedded fonts are not carried over. DocumentPathTests.declaredPath —
epub markdown — ebooks supported structural chapters, links, metadata, tables, textStyles Images are kept as references only; the image files are not written (use “Extract media” on the source). Page layout, fonts, colours, headers and footers are dropped; complex or merged-cell tables are simplified. DRM-protected books are refused. Obfuscated embedded fonts are not carried over. DocumentPathTests.declaredPath —
epub odt — ebooks supported structural chapters, images, links, metadata, tables, textStyles The document is rebuilt from its structure (headings, lists, tables, images, links, notes); the source's visual styling is replaced by default styles. DRM-protected books are refused. Obfuscated embedded fonts are not carried over. DocumentPathTests.declaredPath —
epub pdf — ebooks supported structural images, links, metadata, textStyles Rendered by WebKit with JavaScript disabled and without any network access: remote images, fonts and stylesheets are missing. Typeset with ORIX Convert's built-in print stylesheet; this is a reflowed rendering, not the source's own page design. The book's own CSS, embedded fonts and page breaks are not used. No table structure, bookmarks or tagged-PDF accessibility information. DRM-protected books are refused. Obfuscated embedded fonts are not carried over. DocumentPathTests.declaredPath —
epub png epub.cover ebooks supported lossless dimensions Works only when the book declares a cover image. The file is copied unchanged when it already is a PNG; otherwise it is re-encoded. EbookTests.coverAndMedia —
epub rst — ebooks supported structural chapters, links, tables, textStyles Images are kept as references only; the image files are not written (use “Extract media” on the source). Page layout, fonts, colours, headers and footers are dropped; complex or merged-cell tables are simplified. DRM-protected books are refused. Obfuscated embedded fonts are not carried over. DocumentPathTests.declaredPath —
epub rtf — ebooks supported structural images, links, tables, textStyles The document is rebuilt from its structure (headings, lists, tables, images, links, notes); the source's visual styling is replaced by default styles. Footnotes may be flattened into the text flow; document properties are not written. DRM-protected books are refused. Obfuscated embedded fonts are not carried over. DocumentPathTests.declaredPath —
epub txt — ebooks supported partial — Text only: fonts, emphasis, images, table grids, links and page layout are dropped. DRM-protected books are refused. Obfuscated embedded fonts are not carried over. DocumentPathTests.declaredPath —

# geospatial sources (31 paths)

Source Destination Operation Engine Status Fidelity Preserves Limitations Tests Performance
geojson csv — geo supported structural attributes, vectorGeometry CSV cannot record the reference system; point layers get coordinate columns, other layers a WKT column. PathTests.convertsCities(_😃 0.98 s (convert; GeoJSON → CSV)
geojson gpkg — geo supported structural attributes, coordinateSystem, vectorGeometry — PathTests.convertsCities(_😃 1.98 s (convert; GeoJSON → GeoPackage)
geojson gpx — geo supported partial vectorGeometry GPX is always WGS 84 and holds only points and lines: polygons are refused unless written as closed tracks. Only the fixed GPX fields (name, desc, cmt, src, sym, type, time, link…) are kept; other attributes are dropped and counted. PathTests.convertsCities(_😃 0.96 s (convert; GeoJSON → GPX (polygons as tracks))
geojson kml — geo supported structural attributes, vectorGeometry KML is always WGS 84 longitude / latitude: sources in WGS 84 or NAD83 geographic, Web Mercator, UTM or MTM coordinates are reprojected; any other reference system is refused. Measure (M) values are dropped. Dates are written as text. PathTests.convertsCities(_😃 0.87 s (convert; GeoJSON → KML)
geojson shp — geo supported structural attributes, coordinateSystem, vectorGeometry The result is a folder holding the .shp, .shx, .dbf, .cpg and .prj files. dBASE limits apply: attribute names of 10 characters, text of 254 bytes, no date-and-time type, no nested values. One geometry type per shapefile: mixed layers are split. PathTests.convertsCities(_😃 1.18 s (convert; GeoJSON → Shapefile (2 shapefiles))
gpkg csv — geo supported structural attributes, vectorGeometry Only vector layers are converted: tiles, rasters and binary (BLOB) attributes are reported and left out. The feature id column is not exported as an attribute. CSV cannot record the reference system; point layers get coordinate columns, other layers a WKT column. A source with several layers gives one output per layer. PathTests.convertsCities(_😃 —
gpkg geojson — geo supported structural attributes, vectorGeometry Only vector layers are converted: tiles, rasters and binary (BLOB) attributes are reported and left out. The feature id column is not exported as an attribute. GeoJSON is always WGS 84 longitude / latitude: sources in WGS 84 or NAD83 geographic, Web Mercator, UTM or MTM coordinates are reprojected; any other reference system is refused. Measure (M) values are dropped. A source with several layers gives one output per layer. PathTests.convertsCities(_😃 0.51 s (convert; GeoPackage → GeoJSON)
gpkg gpx — geo supported partial vectorGeometry Only vector layers are converted: tiles, rasters and binary (BLOB) attributes are reported and left out. The feature id column is not exported as an attribute. GPX is always WGS 84 and holds only points and lines: polygons are refused unless written as closed tracks. Only the fixed GPX fields (name, desc, cmt, src, sym, type, time, link…) are kept; other attributes are dropped and counted. PathTests.convertsCities(_😃 —
gpkg kml — geo supported structural attributes, vectorGeometry Only vector layers are converted: tiles, rasters and binary (BLOB) attributes are reported and left out. The feature id column is not exported as an attribute. KML is always WGS 84 longitude / latitude: sources in WGS 84 or NAD83 geographic, Web Mercator, UTM or MTM coordinates are reprojected; any other reference system is refused. Measure (M) values are dropped. Dates are written as text. PathTests.convertsCities(_😃 —
gpkg shp — geo supported structural attributes, coordinateSystem, vectorGeometry Only vector layers are converted: tiles, rasters and binary (BLOB) attributes are reported and left out. The feature id column is not exported as an attribute. The result is a folder holding the .shp, .shx, .dbf, .cpg and .prj files. dBASE limits apply: attribute names of 10 characters, text of 254 bytes, no date-and-time type, no nested values. One geometry type per shapefile: mixed layers are split. A source with several layers gives one output per layer. PathTests.convertsCities(_😃 0.96 s (convert; GeoPackage → Shapefile)
gpx csv — geo supported structural attributes, vectorGeometry Per-point timestamps and sensor data of GPX tracks are dropped unless “Track points as features” is on. CSV cannot record the reference system; point layers get coordinate columns, other layers a WKT column. A source with several layers gives one output per layer. PathTests.convertsCities(_😃 —
gpx geojson — geo supported structural attributes, vectorGeometry Per-point timestamps and sensor data of GPX tracks are dropped unless “Track points as features” is on. GeoJSON is always WGS 84 longitude / latitude: sources in WGS 84 or NAD83 geographic, Web Mercator, UTM or MTM coordinates are reprojected; any other reference system is refused. Measure (M) values are dropped. A source with several layers gives one output per layer. PathTests.convertsCities(_😃 —
gpx gpkg — geo supported structural attributes, coordinateSystem, vectorGeometry Per-point timestamps and sensor data of GPX tracks are dropped unless “Track points as features” is on. PathTests.convertsCities(_😃 —
gpx kml — geo supported structural attributes, vectorGeometry Per-point timestamps and sensor data of GPX tracks are dropped unless “Track points as features” is on. KML is always WGS 84 longitude / latitude: sources in WGS 84 or NAD83 geographic, Web Mercator, UTM or MTM coordinates are reprojected; any other reference system is refused. Measure (M) values are dropped. Dates are written as text. PathTests.convertsCities(_😃 —
gpx shp — geo supported structural attributes, coordinateSystem, vectorGeometry Per-point timestamps and sensor data of GPX tracks are dropped unless “Track points as features” is on. The result is a folder holding the .shp, .shx, .dbf, .cpg and .prj files. dBASE limits apply: attribute names of 10 characters, text of 254 bytes, no date-and-time type, no nested values. One geometry type per shapefile: mixed layers are split. A source with several layers gives one output per layer. PathTests.convertsCities(_😃 —
kml csv — geo supported partial attributes, vectorGeometry KML styles, icons, overlays, tours, 3D models and network links are not carried over (network links are never fetched). CSV cannot record the reference system; point layers get coordinate columns, other layers a WKT column. A source with several layers gives one output per layer. PathTests.convertsCities(_😃 —
kml geojson — geo supported partial attributes, vectorGeometry KML styles, icons, overlays, tours, 3D models and network links are not carried over (network links are never fetched). GeoJSON is always WGS 84 longitude / latitude: sources in WGS 84 or NAD83 geographic, Web Mercator, UTM or MTM coordinates are reprojected; any other reference system is refused. Measure (M) values are dropped. A source with several layers gives one output per layer. PathTests.convertsCities(_😃 —
kml gpkg — geo supported partial attributes, coordinateSystem, vectorGeometry KML styles, icons, overlays, tours, 3D models and network links are not carried over (network links are never fetched). PathTests.convertsCities(_😃 —
kml gpx — geo supported partial vectorGeometry KML styles, icons, overlays, tours, 3D models and network links are not carried over (network links are never fetched). GPX is always WGS 84 and holds only points and lines: polygons are refused unless written as closed tracks. Only the fixed GPX fields (name, desc, cmt, src, sym, type, time, link…) are kept; other attributes are dropped and counted. PathTests.convertsCities(_😃 —
kml shp — geo supported partial attributes, coordinateSystem, vectorGeometry KML styles, icons, overlays, tours, 3D models and network links are not carried over (network links are never fetched). The result is a folder holding the .shp, .shx, .dbf, .cpg and .prj files. dBASE limits apply: attribute names of 10 characters, text of 254 bytes, no date-and-time type, no nested values. One geometry type per shapefile: mixed layers are split. A source with several layers gives one output per layer. PathTests.convertsCities(_😃 —
kmz csv — geo supported partial attributes, vectorGeometry Only the KML document of the KMZ archive is read: the images, icons and models packed with it are not carried over. KML styles, icons, overlays, tours, 3D models and network links are not carried over (network links are never fetched). CSV cannot record the reference system; point layers get coordinate columns, other layers a WKT column. A source with several layers gives one output per layer. KMZSourceTests —
kmz geojson — geo supported partial attributes, vectorGeometry Only the KML document of the KMZ archive is read: the images, icons and models packed with it are not carried over. KML styles, icons, overlays, tours, 3D models and network links are not carried over (network links are never fetched). GeoJSON is always WGS 84 longitude / latitude: sources in WGS 84 or NAD83 geographic, Web Mercator, UTM or MTM coordinates are reprojected; any other reference system is refused. Measure (M) values are dropped. A source with several layers gives one output per layer. KMZSourceTests —
kmz gpkg — geo supported partial attributes, coordinateSystem, vectorGeometry Only the KML document of the KMZ archive is read: the images, icons and models packed with it are not carried over. KML styles, icons, overlays, tours, 3D models and network links are not carried over (network links are never fetched). KMZSourceTests —
kmz gpx — geo supported partial vectorGeometry Only the KML document of the KMZ archive is read: the images, icons and models packed with it are not carried over. KML styles, icons, overlays, tours, 3D models and network links are not carried over (network links are never fetched). GPX is always WGS 84 and holds only points and lines: polygons are refused unless written as closed tracks. Only the fixed GPX fields (name, desc, cmt, src, sym, type, time, link…) are kept; other attributes are dropped and counted. KMZSourceTests —
kmz kml — geo supported partial attributes, vectorGeometry Only the KML document of the KMZ archive is read: the images, icons and models packed with it are not carried over. KML styles, icons, overlays, tours, 3D models and network links are not carried over (network links are never fetched). KML is always WGS 84 longitude / latitude: sources in WGS 84 or NAD83 geographic, Web Mercator, UTM or MTM coordinates are reprojected; any other reference system is refused. Measure (M) values are dropped. Dates are written as text. KMZSourceTests —
kmz shp — geo supported partial attributes, coordinateSystem, vectorGeometry Only the KML document of the KMZ archive is read: the images, icons and models packed with it are not carried over. KML styles, icons, overlays, tours, 3D models and network links are not carried over (network links are never fetched). The result is a folder holding the .shp, .shx, .dbf, .cpg and .prj files. dBASE limits apply: attribute names of 10 characters, text of 254 bytes, no date-and-time type, no nested values. One geometry type per shapefile: mixed layers are split. A source with several layers gives one output per layer. KMZSourceTests —
shp csv — geo supported structural attributes, vectorGeometry A shapefile without a .prj file does not say which reference system it uses: state it in the settings. CSV cannot record the reference system; point layers get coordinate columns, other layers a WKT column. PathTests.convertsCities(_😃 —
shp geojson — geo supported structural attributes, vectorGeometry A shapefile without a .prj file does not say which reference system it uses: state it in the settings. GeoJSON is always WGS 84 longitude / latitude: sources in WGS 84 or NAD83 geographic, Web Mercator, UTM or MTM coordinates are reprojected; any other reference system is refused. Measure (M) values are dropped. PathTests.convertsCities(_😃 0.34 s (convert; Shapefile (points) → GeoJSON)
shp gpkg — geo supported structural attributes, coordinateSystem, vectorGeometry A shapefile without a .prj file does not say which reference system it uses: state it in the settings. PathTests.convertsCities(_😃 0.95 s (convert; Shapefile (points) → GeoPackage)
shp gpx — geo supported partial vectorGeometry A shapefile without a .prj file does not say which reference system it uses: state it in the settings. GPX is always WGS 84 and holds only points and lines: polygons are refused unless written as closed tracks. Only the fixed GPX fields (name, desc, cmt, src, sym, type, time, link…) are kept; other attributes are dropped and counted. PathTests.convertsCities(_😃 —
shp kml — geo supported structural attributes, vectorGeometry A shapefile without a .prj file does not say which reference system it uses: state it in the settings. KML is always WGS 84 longitude / latitude: sources in WGS 84 or NAD83 geographic, Web Mercator, UTM or MTM coordinates are reprojected; any other reference system is refused. Measure (M) values are dropped. Dates are written as text. PathTests.convertsCities(_😃 —

# scientific sources (10 paths)

Source Destination Operation Engine Status Fidelity Preserves Limitations Tests Performance
hdf5 csv — scientific supported partial — The output is a flattened table of the variables you select, not the file's structure: groups, other variables and the multidimensional layout are not in it. The transformation applied is recorded in the output. CSV has no place for attributes: they are written to a side file only when you ask for it. External links, external raw-data files and virtual datasets are never followed; datasets compressed with szip or third-party filters cannot be read. FlattenCSVTests (9 tests) —
hdf5 hdf5 hdf5.repack scientific supported structural attributes, dimensions, metadata Datasets holding references are left out and reported; soft and external links are not recreated. Filters other than deflate, shuffle and Fletcher-32 are not re-applied (the values are kept). External links, external raw-data files and virtual datasets are never followed; datasets compressed with szip or third-party filters cannot be read. RepackTests.repackKeepsEverythingItCan, cancellationAndProgress —
hdf5 json — scientific supported partial metadata The output is a flattened table of the variables you select, not the file's structure: groups, other variables and the multidimensional layout are not in it. The transformation applied is recorded in the output. NaN and infinite numbers become null (counted). Large selections are refused above the row limit. External links, external raw-data files and virtual datasets are never followed; datasets compressed with szip or third-party filters cannot be read. FlattenJSONTests.records, nestedArrays —
hdf5 netcdf — scientific supported structural attributes, dimensions, metadata Datasets holding object or region references, bitfields, half-precision floats or the HDF5 time class have no NetCDF equivalent: they are left out and reported. Dimensions without a dimension scale get generated names (phony_dim_N), as NetCDF itself does. NetCDF classic output is refused for files with groups, strings, compound, enumerated or variable-length types. External links, external raw-data files and virtual datasets are never followed; datasets compressed with szip or third-party filters cannot be read. HDF5ToNetCDFTests.everythingToNetCDF4, classicVariantsFromHDF5, classicRefusals —
hdf5 parquet — scientific supported partial metadata The output is a flattened table of the variables you select, not the file's structure: groups, other variables and the multidimensional layout are not in it. The transformation applied is recorded in the output. Attributes and the transformation are stored as JSON text in the Parquet key/value metadata. Pages are PLAIN-encoded, uncompressed unless gzip is chosen. External links, external raw-data files and virtual datasets are never followed; datasets compressed with szip or third-party filters cannot be read. FlattenParquetTests.typesNullsAndMetadata, severalRowGroups —
netcdf csv — scientific supported partial — The output is a flattened table of the variables you select, not the file's structure: groups, other variables and the multidimensional layout are not in it. The transformation applied is recorded in the output. CSV has no place for attributes: they are written to a side file only when you ask for it. External links, external raw-data files and virtual datasets are never followed; datasets compressed with szip or third-party filters cannot be read. FlattenCSVTests.netCDFSources (classic and NetCDF-4) —
netcdf hdf5 — scientific supported structural attributes, dimensions, metadata The HDF5 file keeps the NetCDF-4 dimension scales, so it stays readable as NetCDF-4. External links, external raw-data files and virtual datasets are never followed; datasets compressed with szip or third-party filters cannot be read. NetCDFConversionTests.classicToNetCDF4AndBack, netCDF4ToHDF5AndRecompression —
netcdf json — scientific supported partial metadata The output is a flattened table of the variables you select, not the file's structure: groups, other variables and the multidimensional layout are not in it. The transformation applied is recorded in the output. NaN and infinite numbers become null (counted). Large selections are refused above the row limit. External links, external raw-data files and virtual datasets are never followed; datasets compressed with szip or third-party filters cannot be read. FlattenJSONTests.nestedArrays (classic) —
netcdf netcdf netcdf.reformat scientific supported structural attributes, dimensions, metadata Converting NetCDF-4 to a classic variant is refused when the file uses groups, strings or user-defined types; only one unlimited dimension survives (the others become fixed, with a warning). Chunking and compression exist only in NetCDF-4. External links, external raw-data files and virtual datasets are never followed; datasets compressed with szip or third-party filters cannot be read. classicWriterReproducesIndependentBytes, classicToNetCDF4AndBack, netCDF4ToHDF5AndRecompression, severalUnlimitedDimensionsToClassic —
netcdf parquet — scientific supported partial metadata The output is a flattened table of the variables you select, not the file's structure: groups, other variables and the multidimensional layout are not in it. The transformation applied is recorded in the output. Attributes and the transformation are stored as JSON text in the Parquet key/value metadata. Pages are PLAIN-encoded, uncompressed unless gzip is chosen. External links, external raw-data files and virtual datasets are never followed; datasets compressed with szip or third-party filters cannot be read. EngineSurfaceTests.netCDFToParquet (classic and NetCDF-4) —

# other sources (11 paths)

Source Destination Operation Engine Status Fidelity Preserves Limitations Tests Performance
folder 7z archive.create archive supported lossless attributes, metadata Owner and group are not stored; set-ID and sticky bits are not stored; hard links become separate files. Extended attributes are not stored. 7z archives cannot be encrypted by this app. Compression runs on one processor core and is slow at Normal and Maximum (a few MB per second); ZIP and tar.zst are many times faster. EngineArchiveTests/MatrixTests (suite) —
folder bz2 archive.compress archive supported lossless — A Bzip2 stream holds one file and stores neither its name nor its permissions: the name comes from the archive's own file name. One file at a time: a folder cannot become a single stream (use a TAR archive for folders). EngineArchiveTests/MatrixTests (suite) —
folder gz archive.compress archive supported lossless — A Gzip stream holds one file and stores neither its name nor its permissions: the name comes from the archive's own file name. One file at a time: a folder cannot become a single stream (use a TAR archive for folders). EngineArchiveTests/MatrixTests (suite) —
folder tar archive.create archive supported lossless attributes, metadata Owner and group are not stored; set-ID and sticky bits are not stored; hard links become separate files. Extended attributes are stored only when you ask for them. EngineArchiveTests/MatrixTests (suite) —
folder tbz2 archive.create archive supported lossless attributes, metadata Owner and group are not stored; set-ID and sticky bits are not stored; hard links become separate files. Extended attributes are stored only when you ask for them. EngineArchiveTests/MatrixTests (suite) —
folder tgz archive.create archive supported lossless attributes, metadata Owner and group are not stored; set-ID and sticky bits are not stored; hard links become separate files. Extended attributes are stored only when you ask for them. EngineArchiveTests/MatrixTests (suite) —
folder txz archive.create archive supported lossless attributes, metadata Owner and group are not stored; set-ID and sticky bits are not stored; hard links become separate files. Extended attributes are stored only when you ask for them. EngineArchiveTests/MatrixTests (suite) —
folder tzst archive.create archive supported lossless attributes, metadata Owner and group are not stored; set-ID and sticky bits are not stored; hard links become separate files. Extended attributes are stored only when you ask for them. EngineArchiveTests/MatrixTests (suite) —
folder xz archive.compress archive supported lossless — A XZ stream holds one file and stores neither its name nor its permissions: the name comes from the archive's own file name. One file at a time: a folder cannot become a single stream (use a TAR archive for folders). EngineArchiveTests/MatrixTests (suite) —
folder zip archive.create archive supported lossless attributes, metadata Owner and group are not stored; set-ID and sticky bits are not stored; hard links become separate files. Extended attributes and resource forks are not stored. Encryption (optional) protects file contents only: file names, sizes and the targets of symbolic links stay readable. EngineArchiveTests/MatrixTests (suite) —
folder zst archive.compress archive supported lossless — A Zstandard stream holds one file and stores neither its name nor its permissions: the name comes from the archive's own file name. One file at a time: a folder cannot become a single stream (use a TAR archive for folders). EngineArchiveTests/MatrixTests (suite) —
Files main
Menu

Theme