What is background removal (flood fill)?
Background removal converts a solid background area to alpha 0 (fully transparent) and saves the result as a transparent PNG. This tool uses a flood-fill algorithm that recursively detects connected near-matching colors starting from the four corners or a chosen color, and turns them transparent.
When to use it
Useful when you need to make a logo or product photo's white background transparent for a website or design layout, remove a scanned document's background to keep only the text, or make icon and illustration backgrounds transparent so they sit cleanly on any color.
Practical Limitations
Browser-based Canvas processing has fundamental limits. This tool uses color-similarity flood fill, so when the subject and background colors are close (e.g., white clothing on a white background), accurate separation is difficult. Semi-transparent objects (glass, smoke, fine hair) cannot be handled with per-pixel color tests on principle. Compared to server-side AI background removal (e.g., Segment Anything), accuracy has clear limits.
Trade-offs
To protect privacy, all processing runs in the browser, which means GPU/CPU resources depend on the client. Processing high-resolution images (4000 px and above) can risk out-of-memory errors or browser crashes. A high tolerance produces jagged edges, while a low tolerance leaves background remnants. The right balance varies per image, so trial and error is needed. Output is always PNG; direct JPEG export is not supported.
Common Q&A
Q: What about photos with non-uniform/gradient backgrounds? Raise the tolerance gradually, or run the tool multiple times clicking different positions. Q: There is a halo (white outline) around the subject. The chosen color is slightly off from the actual background. Use the color picker to grab the exact pixel color and fine-tune the tolerance. Because no image data is ever uploaded, you can safely use it for internal documents and personal photos.
Trends in background removal
Before 2020, Photoshop's Magic Eraser and GIMP's "Select by Color" were the standard tools. Since 2023, AI-based segmentation (Meta SAM, Remove.bg, etc.) has become mainstream. Those services rely on cloud processing, which is unsuitable for sensitive images. With WebGPU, ML models will eventually run quickly in the browser, enabling offline AI background removal. Until then, color-based tools like this one remain a practical, privacy-respecting option.