MP4 to PNG Converter
Extract lossless PNG frames from MP4 videos in your browser. Free, private, no upload.
How to use
- 1Upload your MP4 file.
- 2Set the number of PNG frames to extract.
- 3Select PNG as the format for lossless output.
- 4Download your PNG images as a ZIP.
What lossless actually gets you
PNG stores exactly the pixels the video decoder produced, with no further compression loss. It is worth being precise about what that means: the frame was already compressed when the video was encoded, and PNG cannot recover detail discarded at that stage. What it guarantees is that no additional degradation happens on top.
That guarantee matters in three situations. When frames will be edited and re-saved multiple times, PNG prevents generational loss. When frames feed a machine learning pipeline, it keeps compression artifacts out of the training signal. And when the source is a screen recording or animation, PNG often produces both a smaller and a visually cleaner file than JPEG, because flat colour regions compress extremely well losslessly and extremely badly with JPEG.
The transparency question
PNG supports an alpha channel, so a reasonable assumption is that extracted frames might be transparent somewhere. For MP4 they never are.
H.264 and H.265 have no mechanism for storing per-pixel transparency. Every frame is fully opaque, and the PNG files you get will have a solid alpha channel throughout. If you need transparency you have to key it in afterwards, or start from a format that supports it - VP8 and VP9 inside a WebM container can carry alpha, which is why transparent web video is almost always WebM.
File size in practice
PNG files are substantially larger than JPEG, and how much larger depends heavily on content. Photographic frames with lots of fine detail compress poorly. Screen recordings, slides and animation compress very well because large regions share identical pixel values.
Plan storage before extracting a large batch. Five hundred PNG frames from 4K footage can approach 4 GB.
| Content | Resolution | PNG per frame | JPEG q92 per frame |
|---|---|---|---|
| Live-action video | 1920x1080 | ~2.5 MB | ~320 KB |
| Live-action video | 3840x2160 | ~8 MB | ~1.1 MB |
| Screen recording | 1920x1080 | ~250 KB | ~180 KB |
| Slide presentation | 1920x1080 | ~90 KB | ~140 KB |
PNG for training datasets
If you are building an image dataset from video, PNG is usually the right default. JPEG artifacts are structured rather than random - they follow the 8x8 block grid of the compression algorithm - and models can and do learn to key on them. A classifier trained on JPEG frames and evaluated on PNG frames, or the reverse, can lose measurable accuracy for reasons unrelated to the actual task.
The tradeoff is storage and load time. If your dataset is large enough that PNG becomes impractical, the important thing is consistency: use one format throughout, at one quality setting, for both training and inference.
FAQ
Why PNG instead of JPG?
PNG is lossless, so no compression artifacts are added on top of what the video codec already did. Use it when frames will be edited repeatedly, fed to a model, or when the source is a screen recording or animation.
How much larger are PNG files?
Typically three to five times larger for live-action footage. For screen recordings and slides the gap narrows sharply, and PNG can actually be smaller than JPEG because flat colour regions compress extremely well.
Can extracted frames have transparency?
Not from an MP4. H.264 and H.265 cannot store an alpha channel, so every frame is fully opaque. Transparent video generally uses VP8 or VP9 inside a WebM container instead.
Does the quality slider affect PNG?
No. PNG is lossless by definition, so the slider is ignored when PNG is selected. It applies only to JPEG and WebP.