CAVS to BMP Converter

CAVS to BMP: Grabbing Frames Out of a Raw AVS Bitstream

This is a frame-extraction tool, not a file rewrap. A .cavs is a bare AVS1 video bitstream — coded picture data with no container, no audio and no seek index — and BMP is a single-image raster format, so the conversion decodes the stream and writes out one still image per frame you ask for. The Frame Selection section is where you say which ones.

The other thing worth knowing before you start is that the Quality Preset dropdown here is not a quality dial in the usual sense. Its three entries produce three genuinely different BMP encodings, with different bit depths, different palettes and file sizes that differ by more than a factor of ten. None of them is a plain 24-bit bitmap; there is no 24-bit path on this route at all. If what you actually want is a lossless full-colour still, CAVS to PNG is the better destination.

How to Convert CAVS to BMP

1. Upload Your CAVS File: Drag the .cavs onto the page or click "+ Add Files". Ordinary players cannot open a loose elementary stream, but the decoder here reads it directly — you do not need to wrap it in a container first.
2. Choose Which Frames You Want: Frame Selection offers Specific Frame, which takes a time in seconds and opens at 0 for the very first frame, or Multiple Screenshots, which takes a capture rate running from ten frames a second down to one frame every ten seconds, with 1 second per frame as its starting value.
3. Pick a Quality Preset — It Chooses the BMP Encoding: High, Medium and Low are the three entries, and the section below sets out exactly what each writes. High is what you get if you change nothing. Image resolution opens on Keep original, with Resolution Percentage, Preset Resolutions, Width, Height and Width x Height available.
4. Convert and Download: Click Convert and save the bitmaps. Files are uploaded over an encrypted connection, decoded on our servers, and deleted automatically after a few hours — no sign-up, no watermark, never shared or made public.

The Three Presets Are Three Different BMP Files

Every figure below was measured by putting the same decoded 704x576 frame through each of the three presets and reading the resulting file headers.

Two things follow from that table. First, High is not lossless — RGB555 discards three bits per channel, so smooth gradients and skies pick up visible banding that no setting here can avoid. Second, Medium and Low are not "the same picture, smaller": they are palettised images with dithering applied, which looks acceptable on graphics and flat colour but produces obvious noise on photographic content. The jump from 811 KB to 95 KB is a change of format, not a change of compression ratio.

The High size is worth checking against arithmetic if you want to confirm it yourself: 704 x 576 pixels at two bytes each is 811,008 bytes, and the file measured 811,062 — the 54 extra bytes are the BMP file and information headers.

Why High Is What You Get Without Touching Anything

The page's option set marks a "Very High" rung as pre-selected, but Very High is not one of the three entries that render — the dropdown shows only High, Medium and Low. When the interface finds that its pre-selected entry is hidden, it falls back to the first visible one, which here is High. So the effective default is the 16-bit RGB555 encoding, and you have to change the dropdown deliberately to get either of the palettised options.

That fallback is also the reason there is no Highest, Very Low or Lowest on this page even though those rungs exist elsewhere on the site: the BMP path implements exactly the three encodings above, and anything else would have nothing to write.

Where Else a Frame From a Raw AVS Stream Can Go

BMP is one of several still-image targets this pipeline can build from a .cavs, and for most purposes it is not the strongest one.

Frequently Asked Questions

Will I get one BMP or several?

It depends entirely on Frame Selection. Specific Frame produces a single bitmap from the moment you name, and it opens at 0 seconds, which is the first frame of the stream — this is the mode to use when you want a thumbnail or a single reference still. Multiple Screenshots produces a series, one image per interval, at the capture rate you choose: 1 second per frame is the starting value, and the list runs from 0.1 seconds per frame at the dense end to 10 seconds per frame at the sparse end. A long stream at a dense rate produces a great many files, so start conservative.

Why does my grab look banded or blotchy?

Because of which preset you were on. High writes RGB555, which stores only five bits per colour channel instead of eight — enough for most content but not for smooth gradients, where it produces visible stepping in skies, walls and shadows. Medium and Low are worse in a different way: they reduce the frame to 256 or 86 colours respectively and apply dithering to disguise the loss, which scatters visible noise across areas of continuous tone. If the banding matters, take the frame as a PNG instead via CAVS to PNG, which can hold full-colour data.

Can I get a 24-bit BMP out of this page?

No. The three presets implement 16-bit RGB555, a 256-colour palette and an 86-colour palette, and there is no fourth option — a plain 24-bit uncompressed bitmap is not something this route produces. If a piece of software has specifically asked you for a 24-bit BMP, produce the frame as a PNG or a TIFF here and convert that to BMP in an image editor, where you can set the bit depth explicitly.

Why won't my .cavs file open in a normal media player?

Because it is not a media file in the sense a player expects. A .cavs is a raw AVS1 elementary stream: coded video frames with no container wrapped round them, no audio track, no duration field and no seek index. Players are built to read containers — AVI, MP4, MKV, a transport stream — and hand the payload inside to a decoder, and given a loose bitstream they have nothing to parse. The decoder on this page reads the bitstream directly, which is why extraction works here even when your desktop player refuses to open the file at all.

What is AVS1, and is it related to AviSynth?

AVS1 is the first generation of the Chinese Audio Video Standard, adopted as national standard GB/T 20090.2 in 2006 by the AVS Workgroup, which was established in June 2002. A .cavs file holds the AVS1-P2 Jizhun profile bitstream. It is entirely unrelated to AviSynth, whose .avs files are text scripts of frameserving instructions rather than video data — the similar extensions are a coincidence that catches people regularly. It is also distinct from the later AVS2 and AVS3 generations, which use different bitstream syntax and normally arrive inside a real container.

Why does seeking to a specific time land on the wrong frame?

Because a bare elementary stream has no timing metadata for the seek to use. There is no container index mapping timestamps to byte offsets, and no declared frame rate, so a request for "the frame at 12 seconds" has to be resolved by decoding forward from the start and counting — and if the frame rate the decoder infers differs from what the material was actually shot at, the landing point drifts. For a precise grab, take a run of frames with Multiple Screenshots at a dense capture rate and pick the one you want, rather than trying to hit it exactly with a timestamp.

Does the BMP keep the original frame size?

Yes, unless you change it. Image resolution opens on Keep original, so the bitmap is written at whatever size the AVS stream decoded to — typically a standard-definition broadcast size such as 704x576 or 720x576 for this kind of material. Resolution Percentage, Preset Resolutions, Width, Height and Width x Height are all available if you need something specific. Scaling up is rarely worth it: an AVS1 broadcast frame has no hidden detail to reveal, so enlargement just makes the existing softness bigger.

Which is smaller, and does it matter for a still frame?

Low is smallest, then Medium, then High, and the gap is large — 71 KB, 95 KB and 811 KB for the same 704x576 frame. Whether that matters depends on what the still is for. If you are pulling a contact sheet of hundreds of frames out of a long stream, the palettised presets save real disk space and the dithering is often acceptable. If you are grabbing one reference frame, the 811 KB of the High preset is trivial and the extra colour depth is worth having. BMP is an inefficient format in every one of its modes, so if size is genuinely the constraint, a PNG or JPEG target is a better answer than a smaller BMP.

Will the audio come across?

There is no audio to come across, and BMP could not hold it anyway. A .cavs file is video only by construction — in AVS workflows the sound travelled as a separate stream and the two were multiplexed together only when a real container was built. If your goal is the soundtrack rather than the picture, a bare .cavs cannot help you; you need the original container the stream was demuxed out of.