drawer.tools
TOOL

Compress an image to a target size

Hit an exact file-size limit: pick the KB target and get the best-looking image that fits under it. Free, in your browser.

Drop files here or click to browse. JPG / JPEG / PNG / WEBP accepted, batches welcome

Drop an image, set the number, download a JPG that fits under it. Made for every form that says "maximum file size: 100KB" and rejects yours.

How it hits the target: quality first, pixels second. The tool searches for the highest JPG quality that fits under your limit. If even low quality cannot get there, it scales the image down and searches again, repeating until the file fits. You get the best-looking image that obeys the number, and the exact final bytes are shown per file.

Pick a sensible target. 200KB keeps most photos looking good. 100KB is fine for documents and ID-style photos. 20KB is passport-portal territory: expect visible softness, because there is only so much picture that will fit into 20KB. If a target is unreachable without wrecking the image, the tool says so and gives you the closest it could get, rather than pretending.

PNG and WebP in, JPG out. JPG is what size-limited forms expect, and it is the format that compresses photographs hardest. Transparency composites onto white on the way.

Dimensions may shrink when the target demands it, and the output is named for the target you asked for, as in photo-100kb.jpg.

Per-file numbers show as each one lands, with a ZIP for batches.

Frequently asked questions

The form says 50KB. What do I type?

  1. The target box takes the number in KB, and the output is at or under it whenever the target is reachable. A little headroom never hurts: aim for 45 if the form is strict about the boundary.

Why did my image come out smaller in pixels?

Because the byte target demanded it. Quality alone bottoms out around 35% before the image falls apart; past that point the only lever left is fewer pixels. The filename carries the target you set, as in photo-100kb.jpg.

Does the compressed file keep my location data?

No. The output is written fresh from the pixels. GPS position, camera model and timestamps are all dropped. Useful, given these files are usually headed for a form.