Movie Encoder
Windows 10 & 11 Buy license · $79.95
Archive & pro

FFV1 Lossless Video Archiving: Why Archives Use It and the Settings to Choose

Why FFV1 in MKV is a preservation standard, the recommended FFmpeg settings for archiving, how it compares with ProRes and HuffYUV, and how to verify it.

Updated 5 minute read The Movie Encoder team

FFV1 is a free, open, mathematically lossless video codec that is widely used by archives, libraries and film preservation projects, usually stored in an MKV (Matroska) container. Every frame decodes to exactly the same pixels as the source, while files are typically far smaller than uncompressed video. A good archival encode in FFmpeg is ffmpeg -i input.mov -map 0 -dn -c:v ffv1 -level 3 -g 1 -slicecrc 1 -slices 16 -c:a flac output.mkv.

This guide explains why FFV1 is popular for preservation, what each setting does, how it compares with other options, and how to confirm an archive really is lossless.

Why archives use FFV1

Preservation formats need to be lossless, well documented, free of licensing restrictions and likely to remain readable for decades. FFV1 meets all of these:

  • Truly lossless. Decoding produces bit-for-bit identical video frames to the input. No quality is ever traded for size.
  • Open and standardised. FFV1 versions 0, 1 and 3 are documented in RFC 9043, published by the IETF. Matroska, its usual container, is also an IETF standard (RFC 9559). Anyone can write a decoder from the specifications.
  • No patents or licence fees. It is implemented in FFmpeg and other open-source software.
  • Built-in error detection. Version 3 can store a CRC checksum in every slice, so damage to a file can be detected and located.
  • Wide format support. It handles 8 to 16 bits per sample, YUV in 4:2:0, 4:2:2 and 4:4:4, RGB, greyscale and alpha channels, preserving the source's pixel format exactly.

As a result, FFV1 in Matroska has been adopted by many national archives, libraries and broadcasters for digitised tape and film collections.

How it compares

OptionLosslessTypical sizeOpen standardMain use
Uncompressed (v210, raw)YesLargestVariesCapture, interchange
FFV1YesOften around half or less of uncompressedYes (RFC 9043)Long-term preservation
HuffYUVYesLarger than FFV1InformalFast lossless intermediates
x264 / x265 lossless modeYesCan be smaller than FFV1Standard codecs, complexLossless delivery, not a usual archive choice
ProRes, DNxHRNo - visually lossless onlyMediumVendor specificationsEditing and mastering

Compression varies a great deal with content. Clean, static digital video compresses well; grainy film scans and noisy analogue tape compress much less because noise is, by definition, unpredictable. Expect a broad range rather than a fixed ratio.

ProRes and DNxHR are excellent editing codecs, but they are lossy: they discard information, even if you cannot see it. They are not substitutes for a lossless preservation master. For more on those codecs, see ProRes vs DNxHR.

The command from the introduction uses settings commonly recommended for preservation:

  • -map 0 -dn - includes every stream from the source (all audio tracks, subtitles and so on), not just one video and one audio stream. -dn leaves out data tracks such as QuickTime timecode, which Matroska cannot store and which would otherwise stop the encode.
  • -c:v ffv1 -level 3 - uses FFV1 version 3, which adds multithreaded slices and per-slice checksums.
  • -g 1 - makes every frame a keyframe (intra-only). Each frame is self-contained, so damage to one frame does not spread to others, and any frame can be accessed directly.
  • -slices 16 - splits each frame into slices that can be encoded and decoded in parallel. Common choices are 4, 6, 9, 12, 16, 24 or 30; more slices are faster on multi-core PCs but compress very slightly less.
  • -slicecrc 1 - stores a CRC checksum for each slice, enabling corruption detection.
  • -c:a flac - encodes audio losslessly with FLAC. Copying PCM audio (-c:a copy) is also lossless.

One more option is worth knowing: -context 1 uses a larger context model that often compresses a little better at the cost of speed.

Keep the source pixel format

An archive is only lossless if nothing is converted on the way in. By default, FFmpeg keeps the source's pixel format when FFV1 supports it, so 10-bit 4:2:2 video stays 10-bit 4:2:2. Avoid adding scale, colour-space or frame rate filters, and do not force a different -pix_fmt. Changing 4:2:2 to 4:2:0 or 10-bit to 8-bit permanently discards information, even though the FFV1 encode itself is lossless.

Interlaced material should also be archived as it is, without deinterlacing. Deinterlacing is a lossy, irreversible process best applied to access copies, as explained in how to deinterlace video. Metadata such as field order and colour information should be carried across too.

Verifying the archive is lossless

FFmpeg can produce an MD5 hash of each decoded frame. If the source and the FFV1 file give identical hashes, every frame matches exactly:

ffmpeg -i input.mov -map 0:v -f framemd5 source.framemd5

Run the same command on the MKV, then compare the two output files. Any difference points to the exact frame that changed. (Timestamps can differ slightly between containers, so compare the hash column rather than the whole line.) A single whole-stream hash with -f md5 is a quicker overall check.

Separately, generating a SHA-256 checksum of each finished file lets you detect later damage on disk or during copying - see verifying video files with SHA-256 checksums. Frame hashes prove the encode was lossless; file checksums prove the file has not changed since.

Practical archiving tips

  • Keep access copies separate. Make H.264 or HEVC copies for everyday viewing; FFV1 files are large and not every player handles them smoothly.
  • Store more than one copy, ideally on different media and in different places.
  • Record what you did. Keep a note of the source, software version, settings and checksums alongside each file.
  • Do not re-encode lossy sources to "rescue" them. Converting an MP4 to FFV1 preserves it exactly as it is but cannot restore quality already lost - and makes it much larger.

Doing it with Movie Encoder

Movie Encoder includes lossless FFV1 presets in MKV, and its Archive and Verify tools create SHA-256 checksums of sources and outputs, compare streams, duration, resolution and metadata, and write preservation manifests and reports in JSON, CSV and plain text. It never calls a lossy encode lossless.

The Workflow Builder has a lossless archive template, and batch verification lets you check a whole collection. Originals are never overwritten or deleted, and outputs are only moved into place after the encode finishes and passes a check.

Frequently asked questions

Is FFV1 really lossless?

Yes. FFV1 is mathematically lossless, so decoded frames are bit-for-bit identical to the source, provided no scaling or pixel format conversion happens during the encode. You can confirm this by comparing frame hashes.

How much smaller is FFV1 than uncompressed video?

It depends heavily on the content. Clean digital footage compresses well, while grainy film or noisy tape compresses much less, so expect a wide range rather than a fixed ratio.

Should I convert my MP4 files to FFV1 for archiving?

There is little benefit. FFV1 preserves an MP4 exactly as it is but cannot restore quality already lost, and the file becomes much larger. Keep and checksum the original MP4 instead.

Get three guides like this every month, plus software update notices.Subscribe →