Target-size compression
Type a KB or MB ceiling. A binary search lands under it precisely in about seven re-encodes — no trial and error.
Capabilities
Target-size image compression — and everything around it
The core is target-file-size compression. The rest exists so you never need a second tool for the same image.
Format conversion
Inputs: JPG, PNG, WebP, GIF, AVIF, HEIC. Outputs: WebP, JPEG, PNG, and ICO favicons up to 256×256.
Crop, paint & overlays
Aspect-ratio presets, click-to-place text with font controls, drag-to-resize image overlays, unified undo/redo.
AI background removal
A segmentation model runs in your browser. Edge feathering, optional solid fill, manual eraser for cleanup.
Batch & ZIP
Drop dozens of files, set a per-image target, download the whole set as a single ZIP.
10 languages & offline
English, Spanish, Portuguese, French, German, Russian, Arabic, Hindi, Japanese, Chinese — and it works without internet once loaded.
Why we built it
Most free online image compressors are just a server with a web page on top
They upload your file, process it remotely on a shared ImageMagick install, and hand back the result. It works. But you pay for it in privacy, speed, and the ads and watermarks stitched into the result page.
The browser already has everything needed. The Canvas API is fast, JPEG and WebP encoders are built in, and HEIC and AVIF decoders run cleanly in WebAssembly. Sukat is what you get building from that side — and adding the one thing those tools keep getting wrong: hitting an exact target on the first try.
Typical online compressor
Your file → their server → back to you
A stranger's machine now has a copy. Upload + queue + download is slower than local. Rate limits, watermarks, ads.
Sukat
Your file → your browser → back to you
Nothing transmitted. No queue. No watermark. No account. Works offline once loaded.
Privacy by design
Your image is never uploaded — all compression runs in your browser
Every image is read into a Canvas, re-encoded with the browser's built-in encoders, and downloaded back to your device. There is no Sukat backend that handles image content — there's no machine we control that could receive it.
The only network call tied to a compression is a decorative counter ping that carries no image data.
How it works
How Sukat compresses an image to an exact size, in four steps
A fully static site delivered from Cloudflare's edge. No backend touches your images at any point.
Read into Canvas
The image you pick is decoded locally and drawn to an off-screen Canvas. HEIC and AVIF use in-browser WebAssembly decoders.
Binary-search the quality
The algorithm re-encodes across the JPEG/WebP quality scale, converging in roughly seven passes on the highest quality that fits under your ceiling.
Downscale only if forced
If even quality 1 is too large for the target, dimensions are reduced and the search runs again. Dimensions are the last resort, not the first move.
Download & (optionally) ZIP
The result is handed back to your device. Batch jobs are packaged by an in-browser ZIP builder. Nothing leaves at any step.
For the deeper walkthrough, see How It Works →
Explore
Image compression tools and size references
For what Sukat does and doesn't do with your data, see the Privacy Policy. License and acceptable-use rules are in the Terms of Use. Bug reports, feature requests, and partnerships go through Contact.