Drag & drop an image here
or
PNG, JPG or WEBP — processed locally on your device
How the Background Remover works
Guide last reviewed: 28 September 2026
This tool uses an on-device AI segmentation model — the same category of technology behind "portrait mode" on phone cameras — to identify which pixels in your photo belong to the subject and which belong to the background. Unlike older background-removal techniques that rely on a plain color backdrop (green screen, chroma key), this model works on ordinary photos with any background, because it has learned what people, animals, products, and other common subjects generally look like. The entire model runs inside your browser tab using WebAssembly; nothing about your photo, or the fact that you used this tool at all, is ever sent anywhere.
How to remove a background from a photo
- Drag a photo into the drop zone, or click to choose one from your device (PNG, JPG, or WEBP).
- Click Remove Background. The first time you do this on a given visit, your browser downloads the AI model (a one-time, several-dozen-megabyte download); every run after that is instant since the model is cached.
- Use the Erase and Restore brushes to fix any spots the AI got wrong — erase leftover background specks, or restore subject detail (like a stray strand of hair or a finger) that got cut away.
- Go to Review & Export to see the final result against different backgrounds, optionally upscale or sharpen it, and download your transparent PNG.
Tips for a cleaner cutout
- Contrast helps. A subject that's a similar color and brightness to its background is the single biggest cause of a rough edge — the model simply has less information to work with there.
- Zoom in before brushing. Edge mistakes are usually only a few pixels wide; it's much easier to correct them precisely at higher zoom than to guess at full size.
- Check your export against both light and dark backgrounds. A faint halo around hair can be invisible on one and obvious on the other — the Review & Export step lets you preview both before committing.
- Fine hair and semi-transparent material (glass, sheer fabric) are the hardest cases for any segmentation model, AI or otherwise — expect to spend a little more brush time there.
Common uses
People use this tool for product photos on marketplaces and online stores, professional headshots and ID-style photos, swapping or removing a photo's background for design work, isolating a logo or graphic element, and preparing images for presentations or print where a transparent PNG is needed. Because everything runs locally, it's also a reasonable choice for photos you'd rather not upload to a third-party server at all — legal documents, unreleased product shots, or anything else sensitive.
Frequently asked questions
Is this actually free, with no hidden limits?
Yes. There's no sign-up, no watermark, no daily quota, and no paid tier — the processing happens on your own device, so there's no server cost per image for us to recoup. See our General Terms and Conditions for the full terms.
Does it work on phones and tablets?
Yes, in any modern mobile browser. Very old browsers that lack WebAssembly SIMD support (roughly, iOS Safari before 16.4, or old Android browsers) can't run the AI model — the page will tell you clearly if yours is one of them, rather than failing silently.
What image formats can I upload, and what do I get back?
You can upload PNG, JPG, or WEBP. The output is always a PNG, since that's the only common format that can actually store the transparency the tool produces.
Will the AI ever get it completely wrong?
On very difficult images — low contrast, motion blur, very fine or wispy detail — the first pass can miss areas. That's exactly what the Erase and Restore brushes are for: rather than accepting or discarding the AI's guess wholesale, you can correct specific regions by hand in seconds.