Remove Image Background

This tool is color-based, not AI segmentation. It works well when the background is plain and near-uniform: a white, black, or solid-color wall, a paper scan, or a studio backdrop. Complex or textured scenes with many colors will not cut out cleanly.

On load, ImgSprint samples the four corners of the image and averages them to guess the key color. You can override it with the eyedropper, which reads the exact background pixel you click, and then fine-tune with the tolerance slider.

Tolerance sets how close a pixel must be to the key color before it is removed — higher removes more similar colors but can eat into the subject. Softness feathers the transition at the cut edge over a range of color distances, producing a smoother outline instead of a harsh line.

The checkerboard is a standard way to show transparency. Squares are visible wherever pixels have been removed, so you can judge exactly what is transparent before you download.

PNG and WebP keep real transparency. JPG cannot store transparency, so when you choose JPG the removed area is flattened onto white — useful when you need the smallest possible file for a white-background layout.

On uniform backgrounds, raising softness usually removes the colored fringe around the subject. Because the method keys on color, subjects whose edges fade into the background may still keep a slight outline.

No. The image is read and processed entirely in your browser using the Canvas API. Nothing is sent over the network.