What actually shrinks a GIF
GIF has no quality slider. There is no equivalent of dropping a JPG from 90 to 75, because a GIF frame is not a compressed photograph — it is a grid of indices into a palette of at most 256 colours, run through lossless LZW. So the levers are different from the ones people expect, and there are only three that matter:
- Colours per frame. The biggest by a distance. Fewer palette entries means fewer distinct indices, which means the LZW stream finds longer repeats. Going from 256 to 64 typically takes around half the file away on UI and screen-recording footage.
- Frame count. Close to linear, because every frame carries its own pixels. Dropping every second frame is usually invisible on interface footage and obvious on anything with fast motion.
- Dimensions. Third, and it works on pixel area — halving the width quarters the pixel count, so a modest scale-down goes further than it sounds.
When a GIF is the wrong answer
If the clip has photographic detail, gradients or sound, no amount of GIF compression will catch up with a video codec, which can describe what moved between frames instead of storing every frame whole, and is not capped at 256 colours. Converting to MP4 is usually a much larger saving than anything on this page. GIF earns its place when you need something that plays automatically and inline somewhere a video will not run.
Frequently Asked Questions
How do I compress a GIF without losing the animation?
Drop it in and reduce the colour count. Every frame is decoded, re-quantised to a smaller palette and re-encoded with its original delay, so the animation and its timing survive. Most general image compressors keep only the first frame, which is why a compressed GIF so often comes back as a still.
What actually makes a GIF smaller?
Three things, in order of effect. Colours per frame is the biggest: a GIF stores palette indices, so halving the palette cuts what has to be encoded. Frame count is next, and it is close to linear. Pixel dimensions are third, because file size tracks pixel area rather than width. Quality sliders of the kind JPG has do not exist here — GIF has no lossy mode in that sense.
How many colours should I use?
64 is a good default for screen recordings, UI captures, text and line art, which is most of what people turn into GIFs. Photographic or gradient-heavy footage starts to show visible banding below about 128. Going to 32 or 16 is worth trying on flat-colour animation, where it can be close to invisible and cuts the file hard.
Why did my GIF get bigger?
Because it was already well optimised. This encoder writes every frame in full rather than storing only the pixels that changed since the previous one, so a file that already used that technique can come back larger. The result panel always shows the before and after, and says so when that happens. Keep the original in that case.
Does dropping frames speed the animation up?
No. Each dropped frame gives its display time back to the frame that replaces it, so the animation still runs for the same duration at the same pace — it just moves in bigger steps. Keeping one frame in two roughly halves the file and is usually unnoticeable on UI footage.
Is my GIF uploaded to a server?
No. Decoding, re-quantising and re-encoding all run in your own browser. You can confirm it by opening your browser Network tab while compressing — no request carries the file.
Should I use MP4 instead?
Usually, if you can. A video codec describes motion between frames and is not limited to 256 colours, so the same clip as MP4 is routinely a fraction of the size and looks better. Use GIF when you need something that plays automatically and inline somewhere a video will not run.
Related tools
- GIF Resizer — change dimensions only
- GIF to MP4 — usually a much bigger saving
- Video to GIF — make one from a clip
- Image Compressor — for PNG, JPG and WebP