Remove Color from Image
Make a chosen color transparent anywhere it appears in an image, with an adjustable match tolerance. 100% client-side.
Loading tools...
How removing a color from an image works
The tool draws the uploaded image onto an offscreen canvas and inspects its raw pixel data via the Canvas ImageDataAPI. For every pixel, it measures the Euclidean distance between that pixel's RGB value and the color you chose to remove. Pixels within the tolerance radius have their alpha channel set to zero, making them fully transparent, while pixels just outside it are feathered — partially faded rather than left with a hard edge — so anti-aliased boundaries (like the edge of text or a logo) knock out cleanly instead of leaving a jagged halo.
Use the eyedropper to click directly on the preview and sample an exact color from the image rather than guessing a hex value — handy for removing a solid background color from a product photo or scanned logo. Raising the tolerance slider widens the net to also catch close shades and compression artifacts around the target color.
Everything runs locally in your browser with the Canvas API — the image is never uploaded to a server. Output is always exported as PNG so the transparency is preserved. To replace a color instead of deleting it, try the Image Color Replacer tool, or use the Color Converter to translate the hex values you find into RGB, HSL, and more.
Built and maintained by Meet Shah · Last updated
What this tool is used for
- Knocking out a flat background to get a transparent PNG for a layout.
- Removing a white background from a logo supplied without transparency.
- Isolating a subject from a solid-colour backdrop.
- Preparing a graphic to sit on a coloured background without a visible box.
- Producing a transparent version of an icon for a dark theme.
Frequently Asked Questions
- What format must the output be?
- One with an alpha channel — PNG or WebP. JPEG has NO transparency, so saving a colour-removed image as JPEG fills the transparent area with black or white and silently undoes the work.
- Why is there a coloured fringe around the subject?
- Because anti-aliased edge pixels are partly the removed colour. This is why green screens are green: it is far from most skin tones, so the spill is easier to key out. A colour close to the subject leaves visible edges.
- How is this different from background removal?
- Colour removal is mechanical — it targets pixels matching a value, wherever they are. Background removal is semantic and uses a model to identify the subject, so it works on photographs where no single background colour exists.
- Why do parts of the subject vanish?
- Because they contain the target colour too. Removing white from a photo takes the white shirt with the background — the tool cannot tell them apart, which is the fundamental limit of colour-based keying.
- Does partial transparency work?
- Yes, and it is what makes edges look right. Pixels partially matching become partially transparent rather than fully, preserving the anti-aliasing instead of producing a hard jagged cut.
Common errors and gotchas
- Exporting as JPEG, which has no alpha channel, so the transparency becomes a solid fill.
- Leaving a halo where anti-aliased edges blended with the removed colour.
- Using a tolerance wide enough to punch holes in the subject where it shares the background's colour.
- Working on a lossy source whose compression has already varied the background colour pixel to pixel.
- Assuming a shadow will survive. Soft shadows blend with the background and usually go with it.