Compress an image

A form that names the cap has its own page: 100 KB, 200 KB, or 1 MB. Discord’s upload ceiling is a different problem — compress for Discord. If the picture is the wrong shape rather than the wrong weight, resize it. A HEIC or WebP that an app will not open should be converted first, then compressed only if the JPG is still too big.

Photographs are already compressed. A JPG from a phone is a lossy file: the camera kept the look and dropped the bytes it thought you would not miss. Compressing it again means encoding those pixels once more, usually as a new JPG at a lower quality, and sometimes at fewer pixels when you asked for a hard cap. That can make a file small enough for a form. It cannot make the picture sharper, and a very low setting adds blocks.

PNG is different. Re-saving a PNG through a canvas does not perform the kind of lossless squeeze a dedicated PNG optimizer does. For a photo stored as PNG, this page will usually offer a JPG or, when the image has real transparency, a WebP, because that is what actually gets smaller. A flat logo with transparency is a poor candidate for a photographic JPEG. The result says when the original was already smaller and was left alone.

Nothing is uploaded. The decoder and encoder run on your device, which is the point of doing this instead of dropping the same photo into a site that stores it “for up to 48 hours.” If that ever changed, the line under the tool would have to change with it.

The blank target is the lighter touch: one re-encode around the quality you set, and the original back if that did not help. A filled-in target is a search. The page tries higher and lower JPEG quality and, if quality alone cannot hit the number, a smaller pixel grid, and it stops rather than shrinking the picture into a thumbnail you cannot read. Caps that show up constantly — 100 KB, 200 KB, 1 MB, and Discord — have their own pages so the number starts filled in and the explanation matches the form you are fighting.