Removal quality
How to Improve Background Removal Edges
Better edges usually come from a better source photo, a sensible Edge feather setting, and a quick review of difficult areas before downloading the final transparent PNG.
Edge review
Start with the source image
Clipo can estimate a subject from JPEG, PNG, and single-frame WebP inputs, but it cannot recover detail that is missing from the source. A sharp subject with visible separation from the background gives the model more useful edge information.
Avoid tiny subjects, heavy compression, motion blur, severe crops, and backgrounds that closely match the subject color when you can. If a result looks uncertain, try another source photo before spending time on manual cleanup.
Choose an Edge feather setting
0-1
Crisp products, packaging, logos, and hard objects.
2-4
Default everyday portraits, clothing, and mixed edges.
5-8
Softer hair, fabric, or edges that look too cut out.
The web tool offers Edge feather values from 0 to 8 so the most useful cutout adjustments stay simple and predictable.
Use reliability warnings as a review signal
A reliability warning does not always mean the result is unusable. It means Clipo detected conditions that deserve a closer look, such as multiple foreground regions, uncertain subject boundaries, or difficult transparent and reflective areas. Inspect the comparison before publishing the image.
Common difficult edges
Hair, fur, and fabric
Fine strands and fuzzy textiles often need softer edges. Try a clear source photo and inspect partial transparency before publishing the result.
Glass, reflections, and shadows
Transparent or reflective areas may contain both subject and background pixels. Use the alpha mask for review and expect manual cleanup for important assets.
Low contrast or blur
If the subject and background share similar colors, or the source is blurry, the model has less edge information. A sharper, higher-contrast source usually helps.
When to use the alpha mask
A successful result can include an alpha mask. White mask pixels are kept, black pixels are removed, and gray pixels are partially transparent. Use the mask in an image editor when important edges need manual touch-up.
Edge guidance updated: July 15, 2026