Helpful Guide

Image Compression

Compress WebP Images

WebP already compresses better than JPEG and PNG, but the quality level chosen when a file was exported is often higher than a web page needs. Re-encoding lets you find the point where the file is smallest and the image still looks right.

You can also drop JPG or PNG files here; everything is written out as WebP.

Drag images here

JPG, PNG or WebP · up to 20 files · processed inside your browser

How to use this tool

  1. 1Add WebP, JPG or PNG files.
  2. 2Set the quality level. Around 75–80 is a good starting point for photos.
  3. 3Compress and read the size comparison for each image.
  4. 4Download the results one by one or as a ZIP.

How WebP saves space

WebP borrows prediction techniques from video coding. Instead of encoding each block independently, it predicts a block from its neighbours and stores only the difference. On photographs this typically yields files 25–35% smaller than JPEG at matching perceived quality, and considerably smaller than PNG on graphics.

Finding the right quality

Compress the same image at 90, 80 and 70 and compare. Most photographic content holds up well at 75. Illustrations with flat colour and hard edges usually need 85 or above, because the encoder spends its budget smoothing exactly the edges you want kept.

Support and fallbacks

All current browsers display WebP. Older desktop applications, some email clients and a few content management systems still do not, so keep a JPEG copy if your files will be opened outside a browser.

Frequently Asked Questions

Can I compress a WebP that is already compressed?

Yes, but each pass discards more detail. Start from the original source when you can.

Does WebP support transparency?

Yes. Alpha is supported in both lossy and lossless WebP, unlike JPEG.

Is animated WebP supported here?

No. Only the first frame of an animated file is processed, because the browser canvas reads a single frame.

Which is smaller, WebP or AVIF?

AVIF is usually smaller again, but encoding is slower and support is narrower. Try the convert tool if you want to compare.