Why we built it

Every mainstream background remover works the same way: your photo gets uploaded to someone's server, queued, processed, and returned. That's slow, it costs the provider money per image (which is why free tiers are tiny and watermarked), and it means your photos — sometimes of you, your kids, or your customers — sit on machines you don't control.

Modern AI segmentation models have gotten small enough to run inside a web browser. So instead of building another upload pipeline, CutOutLab ships the model to you: the first time you use the tool, a few megabytes of neural network download into your browser and every cutout after that runs on your own device.

What this means for you

Privacy is structural, not a policy. In Auto, People and Precision modes your images never leave your computer — there is no upload code on that path. We can't leak, sell, or train on photos we never receive.

Free is sustainable. Browser processing costs us nothing per image, so the free tier has no credits, no watermark and no sign-up wall. The optional HD cloud engine is what the Pro subscription pays for.

Speed is local. No queue, no upload wait — most cutouts finish in a few seconds, even on modest hardware and slow connections.

How it's made

The on-device engine runs open-weight, Apache-licensed segmentation models (u2net family and IS-Net) via WebAssembly and WebGPU. HD mode sends your image over an encrypted connection to a studio-grade cloud model and discards it immediately after — our processing provider operates under a no-retention policy. The site itself is served as static pages, with no third-party ad networks or trackers beyond privacy-friendly analytics.

Get in touch

Questions, bug reports, or feature requests: email [email protected] — a real person reads it. See also the privacy policy and terms of service.