✨ Remove Background

Step-by-step

1. Pick an image with a flat background White product shots, green-screen footage, and portraits on plain backdrops all suit this. The simpler the background and the greater its contrast with the subject, the cleaner the result.
2. Confirm the background colour By default the four corners are averaged, which gets white and green screens right first time. If the subject fills the frame and the corners are not background, turn detection off and pick the colour yourself.
3. Adjust the tolerance This is the value that matters most. Raise it if a halo remains around the edges; lower it if the subject starts losing pieces. Around 30 suits most clean white backgrounds; green screens usually want 50–70.
4. Set the edge softness With none, edges come out jagged. Around 20 lets the cut-out settle naturally onto a new background. Raise it for hair and fuzzy edges, but too much makes the whole outline semi-transparent.
5. Run it and download The output is a PNG with real transparency — the checkerboard behind the thumbnail is where it is see-through. To put it on a solid colour, run the result through Convert to JPG.

When you would use this

How it decides what is background

The approach is direct: take a background colour as reference and measure how far each pixel's colour sits from it. Very close becomes fully transparent, far away stays fully opaque, and the range between gets proportional partial transparency — that is what "edge softness" controls, and why edges do not come out jagged.

Colour alone has one obvious failure mode, though: anything inside the subject that shares the background colour gets removed too. A white shirt on white, a highlight on a white product, the whites of someone's eyes — all become holes.

So by default one more step runs: a flood fill inward from all four edges. Only transparent regions reachable from an edge actually count as background. White areas enclosed by the subject cannot be reached, so they survive. That single option usually decides whether the result is usable.

What material works

Works well: white product shots, green screens, studio portraits on plain backdrops, existing cut-outs, line art. What they share is a single background colour with clear separation from the subject.

Does not work: outdoor scenes, textured or gradient backgrounds, fine hair detail, translucent things like glass or smoke. Those need a model that understands what a person is and what a background is — not something colour comparison can solve.

Why there is no AI mode

It is technically possible, but the costs are concrete: mainstream in-browser removal models require tens of megabytes of weights to download, and inference occupies the visitor's CPU or GPU for several seconds.

Our judgement is that for this site's most common need — cutting a product out of a white background — colour comparison already gives a clean result instantly, with nothing downloaded. Rather than making everyone pay that download cost for a minority of hard cases, we would rather state the limits plainly. For genuinely complex material, a dedicated service is the better tool.

Frequently asked questions

What you might need next

Image work rarely ends in one step. These pair up with Remove Background most often:

• 🪪 ID Photo Sizes — Passport and ID sizes, print sheet included
• 🎨 JPG to PNG — Convert to lossless PNG or WebP
• ✂️ Crop Image — Drag to select the area to keep

Want to share the result? Upload turns images, photos, or videos into a short link with optional password, expiry, and view limits. Or head back to all 12 tools.