Background Remover
Click the background colour to remove it. This is a colour-based tool, so it works well on plain backdrops and poorly on busy ones — the page is honest about which is which.
Drop an image here or click to choose
Works best on plain, evenly lit backgrounds
Click the background in the image to remove that colour. Click again elsewhere to remove another shade.
What this tool can and cannot do
This removes a background by colour, not by understanding what is in the picture — and that distinction decides whether it will work for you. Click a background pixel and every pixel of a similar colour becomes transparent. On a product shot against a white sweep, a passport photo against a plain wall, or a logo on a solid field, that is exactly right and takes seconds. On a person in a park, it will fail, because the AI-style removers that handle those run a trained segmentation model — something this page deliberately does not download or upload your photo to.
When it works, and when it does not
| Image | Result |
|---|---|
| Product on a white sweep | Excellent |
| Passport photo on a plain wall | Good, with tolerance tuning |
| Logo or graphic on solid colour | Excellent |
| Screenshot with a flat background | Excellent |
| Person against a green screen | Good |
| Anyone photographed outdoors | Poor — too many background colours |
| Hair, fur or anything wispy | Poor — the edge is semi-transparent by nature |
| Subject a similar colour to the background | Poor — they cannot be told apart |
If the subject and background share a colour, no colour-based method can separate them. That is a limit of the approach, not of the settings.
Getting a clean cut-out
- Start with low tolerance and raise it. Going too high first eats into the subject and you lose the reference for what "too far" looks like.
- Click several times. Even a "plain" wall has shadow and gradient. Removing three or four related shades usually beats one huge tolerance.
- Preview against dark. White fringes around the subject are invisible against white and obvious against a dark background — which is where they will end up.
- Use a little edge softness. One or two removes the staircase edge without blurring the subject.
- Shoot for it if you can. A plain wall, subject a metre away from it, even front lighting. Two minutes of setup beats an hour of correction.
Why there is no AI model here
Background removal services that handle complicated photos work by running a trained segmentation model. That model has to be either downloaded to your device — typically tens of megabytes — or, far more commonly, run on the provider's servers, which means uploading your photo.
For the photos people most often want cut out — product shots, ID photos, headshots — uploading is a real cost, and colour-based removal handles those cases perfectly well.
So this tool does what can be done honestly on your own machine, in a few kilobytes of JavaScript, and tells you plainly where that stops. For a photograph with a complex background, an AI service will do a better job, and you should decide knowingly whether the upload is a fair trade.
How to use the Background Remover
- Drop in an image with a plain background.
- Click the background to remove that colour.
- Adjust tolerance until the background goes but the subject stays.
- Add a little edge softness to avoid a hard jagged outline, then download the PNG.
Frequently asked questions
Is my photo uploaded?
No. Everything happens on a canvas in your browser. There is no server and no model download.
Why is it leaving a coloured fringe around my subject?
Because edge pixels blend subject and background, so they are neither colour exactly. Raise tolerance slightly and add edge softness. A fringe against a dark preview is the clearest way to see it.
Can it remove a busy background?
No. It matches colours, so a background with many colours cannot be selected. Use an AI-based service for those, accepting that it means uploading the photo.
Why did it delete part of my subject?
Tolerance is too high, or the subject contains the background colour. Lower the tolerance and click several background shades individually instead.
What format is the download?
PNG, which supports transparency. JPEG cannot store an alpha channel, so a transparent background would come out white.
Does it work on hair?
Poorly. Hair edges are semi-transparent and fine, which colour thresholding cannot handle. This is the main case where a trained model genuinely outperforms.
Related tools
Last reviewed September 4, 2026. Results are estimates for general information only.