Background Remover

Drop a photo onto the tool or click to browse — JPG, PNG and WebP all work. The first time you use it, a small AI model (about 50 MB) downloads and is then cached, so later removals are quick. The background is erased automatically in a few seconds and you get a cut-out with a transparent background that you can download as a PNG, or place on a white, black or custom colour before downloading.

It is completely free with no sign-up, and your image is never uploaded. The background removal — including the AI model — runs locally in your browser, so your photo never leaves your device. The only thing fetched from the network is the model itself, which downloads once from a CDN and is then cached; your image is not part of that. That makes it safe for product shots, ID photos and personal pictures, and after the one-time model download it even works offline.

You get a PNG with a transparent background, ready to drop into a design, document or store listing. If you would rather have a solid colour behind the subject, pick white, black or any custom colour and the tool composites the cut-out onto it before you download — handy for passport-style photos, thumbnails or catalog images.

Yes. Once the background is gone you can crop the result two ways. Auto-trim snaps the crop tight to the subject by removing the empty transparent margins, which is great for product shots and stickers. Manual crop lets you drag a box — with corner and edge handles and rule-of-thirds guides — to frame it exactly. The transparent PNG you download is the cropped version, and Reset crop restores the full image at any time.

It uses a modern image-segmentation model (ISNet) that handles people, products, animals and objects well, including most soft edges. Very fine hair, glass or motion blur can leave a few rough pixels, since no automatic tool is perfect — but for the vast majority of photos the cut-out is clean enough to use as-is.

Removing a background is real AI inference running on your own device, so the first run downloads the model and warms it up, and large images take more memory and time. Subsequent images are faster because the model stays cached. If a very large photo struggles, resize it smaller first — the result is the same and it runs quicker.