Simple Background Remover

How this background remover actually works

Unlike AI-powered tools that try to guess where a subject ends using a trained neural network, this remover uses a simple, transparent, and completely predictable technique: color-key removal, the same idea as a weather-forecast green screen. You click a pixel in the background, the tool reads its exact color, and then it makes every pixel "close enough" to that color transparent. There is no model, no guessing, and no upload — the entire process happens on an HTML canvas inside your browser, so your image never leaves your device.

The color-distance math

"Close enough" is measured as the straight-line distance between two colors in RGB space. Each pixel and the sampled color are points with three coordinates — red, green, and blue, each from 0 to 255. The distance between them is the familiar Euclidean formula:

distance = √((R₁−R₂)² + (G₁−G₂)² + (B₁−B₂)²)

If that distance is within your tolerance, the pixel's alpha (opacity) is set to 0, making it transparent. The Tolerance slider runs from 0 to 100 and is scaled internally up to about 221 — the maximum possible distance between two colors in this space. A tolerance of 0 erases only pixels that are an exact match; higher values forgive more variation in shade and lighting.

Flood fill versus global threshold

The "Flood fill (connected only)" checkbox changes which matching pixels get removed, and the difference matters enormously:

• Flood fill on (the default): Removal starts from the four edges of the image and spreads inward only through connected matching pixels. A white pixel inside your subject's eye is left alone because it is not connected to the edge background. This is what you want almost every time.
• Flood fill off (global threshold): Every pixel matching the sampled color is erased, anywhere in the image. A white shirt against a white background would vanish along with the background. Use this only when the color you want gone appears throughout the image.

A worked example

Suppose you have a product photo of a dark mug on a plain light-gray background, roughly rgb(235, 235, 235). You click the gray area; the tool samples that color. With Flood fill on and the Tolerance slider at 30, it walks in from the edges, erasing every connected pixel within a color distance of about 66 of that gray. The mug — which is far away in color — stays fully opaque, and the result canvas shows it floating on a checkerboard pattern that represents transparency. If a faint gray halo remains around the mug's edge, nudge Tolerance up to 40 or 45; if parts of a light shadow on the mug start disappearing, drop it back down. When it looks right, click Download PNG to save the image with a real transparent background.

When this technique shines — and when it doesn't

• Great for: products shot on a solid backdrop, logos and icons on a flat color, screenshots, scanned documents, and anything with a clean, uniform background and a clear color gap between subject and background.
• Struggles with: busy or gradient backgrounds, soft hair and fur, motion blur, and any case where the subject contains the same colors as the background. For those, an AI-based remover that understands shapes will do better — but it cannot match the privacy and predictability of this approach.

Practical tips

• Shoot for the tool. If you can control the photo, use a solid, evenly lit background that contrasts with your subject. A wrinkle-free sheet of paper works well.
• Sample from a representative spot. Click an average area of the background, not a bright glare or a dark shadow corner, so the sampled color sits in the middle of the background's range.
• Adjust tolerance live. Moving the slider re-runs the removal instantly using the same sampled color, so dial it in by eye.
• Re-sample for uneven backgrounds. If lighting makes one side of the background a different shade, sample that side too — though uneven backgrounds are exactly where a higher tolerance helps.

Common mistakes

• Cranking tolerance too high. Push it far enough and the distance threshold swallows part of your subject. Raise it gradually.
• Leaving flood fill off by accident. If holes appear inside your subject, you are using the global threshold — turn flood fill back on.
• Expecting clean hair edges. Color keying produces hard edges; fine wispy detail will look ragged. This is a limitation of the method, not a setting to fix.
• Saving as JPEG elsewhere. The tool exports PNG because JPEG cannot store transparency — re-save your final file as PNG to keep the see-through background.