Image compressor
Nothing is uploaded: every image is processed in this browser
Compression re-encodes a picture at a lower quality setting, discarding detail the eye is least likely to notice. Dropping a JPEG from quality 95 to 78 typically halves the file with no visible difference, and saving the same photograph as WebP is another 25–30% smaller again. Capping the longest side saves far more than either.
How to compress images
Resolution usually matters more than quality. A 4000-pixel photograph displayed at 800 pixels wide is carrying twenty-five times more data than it can show, and capping the longest side removes that waste before the quality setting does anything at all. Compress first by dimension, then by quality. Doing it the other way round means throwing away detail from pixels you were about to discard anyway.
Picking the output format
The three formats fail in different places, and the right one depends on what the picture is rather than on what it currently happens to be. JPEG is lossy and made for photographs; it has no alpha channel, so any transparency becomes solid. PNG is lossless, which means the quality slider is inert on it and a different set of moves applies — the PNG compressor covers those. WebP takes on both: lossy mode competes with JPEG on photographs at roughly 25–30% smaller for matching quality, and it keeps transparency, which PNG only manages losslessly. Every current browser reads it, so the old compatibility argument has gone; the remaining friction is outside the browser, where some desktop software and a few email clients still handle it badly. AVIF is smaller again, but slower to encode and less universally supported, which is why WebP is still the safer default.
The quality number itself is not a percentage of anything and is not comparable between encoders. It sets how coarsely the encoder quantises each 8×8 block of frequency coefficients, so quality 78 in one JPEG library and 78 in another are different files. What is comparable is the shape of the curve: the file size falls steeply from 95 down to about 80 while almost nothing visible goes, then flattens, while the artefacts — ringing along hard edges, blocking in flat gradients like a clear sky — arrive quickly below 60. That is why the useful range is narrow and why anything outside 60 to 90 is usually a mistake in one direction or the other.
Compressing the same file twice
JPEG loss compounds. A file re-encoded five times looks visibly worse than the same file compressed once to the same final quality, because each pass quantises what the previous pass already threw away. Always work from the original rather than from a copy that has been through a chat app, and keep that original if you might need a different size later. Generation loss is exactly why a picture reposted repeatedly ends up looking the way it does.
What people use it for
- Getting a batch of photos under an upload limit
- Preparing images for a website
- Shrinking a folder of screenshots before emailing
- Reducing a phone camera roll before archiving
- Re-encoding camera JPEGs at a lower quality
- Converting a website image set to WebP
Questions
As many as your machine will hold. There is no server, so there is no upload limit and no queue. The only constraint is memory.
No. Every image is decoded, resized and re-encoded by your own browser. Nothing leaves the device.
Around 78 for the web and 85 or higher for anything printed. Below about 60 the artefacts become visible on flat areas and edges.
Typically about half for a camera JPEG at quality 75 to 78. Capping the dimensions as well can take 90% or more off it.
Usually 25–30% smaller at matching visual quality, and it keeps transparency, which JPEG cannot.
Yes, every current browser does. It has not been a practical concern for several years.
For a website, generally yes. For a file you are sending to a person, JPEG or PNG still travel more easily through desktop software and email clients.
Smaller again, but slower to encode and less universally supported. WebP is the safer default today.
PNG is lossless, so the slider is inert on it. Cap its dimensions or switch the output format; the PNG compressor goes through what does and does not work on one.
Yes. JPEG has no alpha channel, so transparent areas become solid. Use WebP if you need transparency and a small file.
Yes. JPEG is lossy and the losses compound, so always work from the original rather than from an already-compressed copy.
No. The number sets how coarsely the encoder quantises, and encoders differ. Judge it by the resulting file rather than by the number.
JPEG for photographs, PNG for flat colour and sharp edges, WebP when you want either of those smaller.
Browsers block repeated automatic downloads, so a loop would silently drop most of a batch. One ZIP always arrives complete.