Resize Image to KB

JPEG quality and file size are related, but not in a straight line. Dropping quality from 95 to 90 can halve the file, while going from 40 to 35 barely moves it, and where those cliffs fall depends entirely on the individual image. A photo of a plain wall compresses very differently from a busy street scene at the same setting. That is why sliding a quality control until you land under 50 KB takes so many tries, and why the number that worked for one photo is wrong for the next.

This tool runs a binary search instead. It encodes the image at full quality first, and if that already fits it stops there, which is one pass. Otherwise it halves the remaining quality range repeatedly, keeping the largest value that still fits, and converges in about eight passes regardless of how the image happens to compress. The result reports how much of your allowance it used, and it is normally over 95 percent, meaning you are getting close to the best quality the limit permits rather than something needlessly degraded.

When even the lowest quality overshoots, the only thing left to give up is pixels. If you allow it, the tool then searches the scale the same way, at a mid quality rather than the floor, so what comes back is a smaller but still sharp image rather than a full-size smear. If you have locked the dimensions because a form requires them, it does not silently shrink your image. It tells you it could not fit.

There is one more judgement built into this. If you have not asked for particular dimensions and the limit is tight, the tool will shrink the image even when it could technically have squeezed the full pixel count in. A 12 megapixel phone photo forced into 50 KB at its original dimensions comes out at around quality 5, which is a smear; the same 50 KB spent on a quarter of the pixels at quality 50 is perfectly legible. It reports when it has made that trade, and it never makes it against dimensions you specified yourself.

Application forms usually state both, and people routinely satisfy one and fail the other. A 200 by 230 pixel box says nothing about how many kilobytes those pixels take, and a 20 KB limit says nothing about the shape. Set them independently here: the size limit at the top, the dimensions below it.

When a form fixes the dimensions, the aspect ratio rarely matches whatever came out of your camera, so something has to give. Crop to fill centres the image and trims the overflowing edge, which is what you want for a face. Fit inside keeps the whole image and pads the gap with a colour, which is what you want for a signature or a logo where cropping would cut off a stroke. Pick deliberately, because a cropped signature and a padded portrait are both immediately obvious to whoever reviews the form.

Two of the presets here come from physical measurements rather than from any particular form: 600 by 600 is 2 by 2 inches at 300 DPI, the United States passport photo standard, and 413 by 531 is 35 by 45 millimetres at the same resolution, which is the common size on visa and identity documents. The rest are pixel sizes that turn up frequently on application forms. Always read the requirement on your own form rather than trusting a preset list, since portals change their rules without announcing it.

The images people bring to a tool like this are passport photographs, signatures, and scans of identity documents. Sending those to a stranger's server to be processed, and trusting a privacy policy about how long the copy is kept, is a poor trade for something a browser can do on its own.

Everything here uses the canvas API that is already built into your browser. The file is read from disk into memory, redrawn, and re-encoded locally. There is no upload, no queue, no temporary file on someone else's disk, and no size limit imposed by a server. The practical consequence is that the tool also works with no network connection once the page has loaded.