How small can your image get?

How it works

Photos are stored with lossy compression: the file keeps what the eye notices and drops fine detail it hardly sees. The quality setting decides how much is dropped. Your browser re-saves the picture on your own device, so nothing is sent anywhere.

The saving is how much smaller the new file is: saving = (original size − new size) ÷ original size. A 1.4 MB photo that becomes 200 kB saved about 86 %. JPEG works everywhere; WebP and AVIF usually make the same picture smaller still, but only appear when your browser can write them. Shrinking the longest side saves even more, because there are fewer pixels to store.

With a target file size the tool tries qualities for you: a higher quality makes a bigger file, so it halves the gap between a quality that fits and one that does not until it finds the highest that fits, in about seven tries. Only when even a low quality is too big does it shrink the picture, by about the square root of how much too big it is, because the file grows roughly with the number of pixels. Sizes count 1 kB as 1000 bytes.

Compressing is a one-way step. Each lossy save costs a little quality, so keep your original, and expect almost no gain from a picture that is already small or compressed. Quality 70–85 is a good start for photos; sharp text and screenshots look better as PNG, which has no quality setting.