Background Remover — AI, In Your Browser
A real segmentation model runs on your own device and separates the subject from the background — including soft edges a colour-based tool cannot handle. Export a transparent PNG or drop in a new background colour.
Three quick steps
1 · Add
Drop one photo. On first use the model downloads once, then it is cached.
2 · Let it cut
The model finds the subject and builds a mask, entirely on your device.
3 · Adjust and save
Tune the edge, pick transparent or a colour, then download the PNG.
About background removal
Is this really AI, and is it really private?
Both. It runs U²-Net, a genuine image segmentation neural network, compiled to run inside your browser. The model is downloaded to your device and your photo is processed there. Nothing is uploaded.
Why is the first use slow?
The model and its runtime are about 15 MB and download once. After that your browser caches them, so later images are fast and it even works offline.
How is this different from the white-background tool?
The older approach fills areas of similar colour, so it only works when the background is plain. This model understands what the subject is, so it copes with busy, blurred or gradient backgrounds.
What if the edges look rough?
Raise Edge tightness to trim colour fringing, or lower it and add a little softness to keep fine detail such as hair. Well-lit photos with a clear subject give the best cut.
Can I use the results commercially?
The model is U²-Net, released under the permissive Apache 2.0 licence, which allows commercial use. You keep all rights to your own images.
Good situations for this tool
- A product photo needs a clean white or transparent background.
- You want a headshot cut out for a profile or presentation.
- A design needs the subject isolated from its surroundings.
- A marketplace requires plain backgrounds on listing images.