Product background cleanup

Task tips

Use a plain background that reaches the image corners. Lower tolerance preserves more similar-colored edges; higher tolerance removes a wider color range. Objects touching the edge or matching the background may be erased. This is color-based cleanup, not AI object segmentation.

A complete workflow

Choose a product image with a plain background touching the corners. Start with a low tolerance, choose transparent or white output and process. Increase tolerance gradually only if background remains. Compare the product’s edges with the original before downloading PNG.

If the result is wrong

If part of the product disappears, lower tolerance or use a source with stronger background separation. A white object on white cannot be reliably separated by color alone. Enclosed background holes may remain because cleanup follows edge-connected regions.

What the export contains

This is conventional color-based cleanup, not AI object segmentation. No model download is needed for this operation. Transparent areas are visible against a checkerboard in many editors; open the saved PNG over a contrasting background to check stray pixels.