Initializing... drag & drop files here
Supports: CAVS
CAVS is a raw Chinese AVS (AVS1) video stream, so this conversion is a frame grab: one moment out of the clip, saved as a single still image. HEIC is the smallest of the still formats on offer here and the least portable — worth choosing when the picture is heading into an Apple workflow or a size-constrained archive, and worth avoiding when other people have to open it. This page compares the realistic targets first, then walks through the frame-selection controls.
| Property | HEIC | JPG | PNG |
|---|---|---|---|
| Underlying coding | HEVC-coded image data in a HEIF container (ISO/IEC 23008-12) | Baseline DCT (ISO/IEC 10918-1) | Deflate, lossless |
| Compression | Lossy, the most efficient of the three | Lossy, oldest and least efficient | Lossless, largest files |
| Bit depth | 8 or 10-bit | 8-bit in the common baseline | 8 or 16-bit per channel |
| Transparency | Supported | None | Full alpha |
| Native browser display | Safari 17+ only — about 15% of tracked users | Universal | Universal |
| Desktop support | Apple natively; Windows needs the HEIF and HEVC extensions from the Microsoft Store; patchy on Linux | Everywhere | Everywhere |
| Best for | Small archives and Apple Photos libraries | Sharing a frame with anyone | Editing, screenshots, exact pixels |
.cavs file onto the page or click "+ Add Files". Several clips can be queued and each produces its own still.0 — type the offset you want, decimals included, so 12.5 lands twelve and a half seconds in. Switch to "Multiple Screenshots" to sample the whole clip instead..heic image. No sign-up, no watermark.The frame does not go straight from AVS to HEIC. It passes through a lossless intermediate, which is why the only quality-affecting decisions you make are the resolution and the final encode setting.
| Stage | What happens | Adds loss? |
|---|---|---|
| 1. Seek | The decoder is positioned at your Time (seconds) value | No — but a raw stream carries no index, so it lands on the nearest decodable frame |
| 2. Decode | One AVS1-P2 frame is decoded and written out as a lossless PNG | No |
| 3. Resize | Applied only if you move Image resolution off "Keep original" | Yes, when downscaling |
| 4. Encode | The PNG is encoded to HEIC at your chosen quality | Yes — this is the only lossy encode step |
Because step 2 is lossless, "Highest" or "Very High" with "Keep original" gives you the closest thing to the decoded broadcast frame that a lossy still format can hold.
The video part of China's first-generation Audio Video coding Standard (AVS1), published as national standard GB/T 20090.2-2006 in February 2006 by the AVS Working Group, which began work in June 2002. A .cavs file is normally a raw AVS1-P2 "Jizhun" profile elementary stream — coded pictures with no container, no index and no audio — which is why players that expect an MP4-style file structure cannot open it even when they could decode the pictures.
The one at time 0, the very first frame of the clip. On broadcast captures that is frequently a black lead-in, a fade or a channel ident, so it is worth setting the Time (seconds) field to a moment you actually want. The field takes decimals, so 2.25 is two and a quarter seconds.
Because it uses HEVC intra-frame coding, a design roughly two decades newer than JPEG's, with larger and more adaptive transform blocks and far better prediction within the image. For the same visual quality it commonly lands at around half the size of a baseline JPEG, and it can carry 10-bit colour, so gradients band less.
No — this is the format's main cost. Safari 17 and later display HEIC natively, which is about 15% of tracked browser users, and Apple's own apps have handled it since iOS 11 and macOS High Sierra. Windows needs Microsoft's HEIF Image Extensions from the Store, plus HEVC support for the coded data, and many Android galleries and web tools still refuse it. If the still is going anywhere unpredictable, JPG is the safer target.
Yes. Switch Frame Selection to "Multiple Screenshots" and pick a Capture Rate — the list runs from one frame every 0.1 seconds up to one every 10 seconds, and starts on "1 second per frame". Every sampled frame is encoded as its own HEIC, which is the practical way to work when you cannot preview the AVS stream to find the moment you want.
Because broadcast AVS material is often interlaced, and a frame captured mid-movement interleaves two fields taken a fraction of a second apart. Nudging the timestamp to a more static part of the shot usually solves it; so does choosing a moment where the camera and subject are still.
Almost certainly not. Broadcast video is normally coded in limited (TV) range, and displays interpret that with their own transfer handling, so a still lifted out of the stream can look a little less punchy than the same frame on a television. That is a difference of interpretation, not damage to the image.
Not in any useful sense. A raw .cavs elementary stream has no container to hold creation dates, camera tags, GPS or track titles — that information lives in the container the stream was demuxed out of, and it was gone before the file reached you. The output is a bare picture with only the properties the encoder writes.
It is uploaded over an encrypted connection, decoded and encoded on our servers, and deleted automatically a few hours later — no sign-up, no watermark, never shared or made public.