Free AI background remover.

Yes. No accounts, no daily caps, no per-photo size limit, no watermark. AI runs entirely in your browser. We never see your photo.

We use @imgly/background-removal, a WASM-compiled neural network model that runs entirely inside your browser. The first time you remove a background, a ~40 MB model downloads to your device. After that, every subsequent run is instant because the model is cached. Nothing about your photo is sent to a server.

JPG, PNG, WebP, and HEIC. iPhone HEIC photos are decoded in the browser automatically before the AI runs. The output is always a clean cutout regardless of input format.

Three options. PNG keeps full transparency (alpha channel) and is what most editing workflows want. WebP also supports transparency and is roughly 30% smaller than PNG; modern browsers and tools handle it well. JPEG does not support transparency, so the JPEG option is disabled when you choose 'Transparent' background mode. For JPEG, pick 'Solid colour' or 'Image' background instead.

Yes. Pick the 'Solid colour' tab and choose a preset (white, black, red, blue, green) or any custom hex colour. Useful for product photos on white, ID-style red backgrounds, or design comps.

Yes. Pick the 'Image' tab, drop a background image (your studio wallpaper, a beach scene, a brand pattern), and the cutout layers cleanly on top. Three fit modes: Cover fills the canvas (cropping the background if needed), Contain shows the whole background letterboxed, Stretch distorts the background to fit exactly.

Yes. Drop multiple files (or a folder) and every photo is processed in turn with the same background settings, then bundled into a ZIP for one-click download. The AI model loads once and reuses across the batch, so 20 photos takes only marginally longer than 1.

Very good for portraits, ID photos, e-commerce product shots, and pet photos. The underlying model (BRIA's RMBG family, quantised for browser use) is specifically trained on portrait and product cutout tasks. Fine fly-aways, fishing line, and translucent objects (smoke, glass) can still trip it up; for those, a desktop tool with manual brushing (Photoshop, Affinity, GIMP) is the right answer.

The first run downloads the ~40 MB AI model to your browser cache. On a fast connection this takes 5-10 seconds; on slow connections it can be a minute. Every subsequent photo (this session and future sessions) is instant because the model is cached. We deliberately lazy-load this only when you start using the tool, so visitors who don't need it pay no bandwidth cost.

No. When the background is removed and the image is re-encoded, EXIF (date, GPS, camera info) is stripped. This is usually what you want for a cutout intended for design or product use. If you need to preserve metadata on a different workflow, our EXIF Editor lets you copy specific tags back onto the output JPEG.