Practical image workflow
Remove Background from Product Photo
Prepare a product cutout from an image with a simple backdrop. This workflow compares tolerance settings and checks the downloaded PNG on contrasting surfaces before you use it in a listing.
Open background removerReproduce the examples
Download each source and compare your result with the actual tool output. The colored display surface is behind the PNG; it is not added to the downloadable file.

Synthetic bottle source
1024 × 1024 known diagnostic source, not a real product photograph.
Download Synthetic bottle source
Tolerance 20, feather 0
Actual PNG downloaded from the local remover.
Download Tolerance 20, feather 0
Tolerance 50, feather 0
Same source; only tolerance changed.
Download Tolerance 50, feather 0
Tolerance 90, feather 0
Compare subject retention as well as removed area.
Download Tolerance 90, feather 01. Decide whether the photo fits this tool
Use a product clearly separated from a fairly uniform background, with background visible in the corners. This browser tool samples corner colors and removes similar regions connected to the image boundary. It does not recognize the product semantically. Busy scenes, white products on white, glass, mesh, and reflective metal need careful masking or a different source photo; a clean preview alone does not prove those subjects survived.
2. Keep a source and a reference
Save an untouched original before processing. Identify label text, cap edges, pale highlights, and the contact shadow you want to keep. The downloadable bottle below is a synthetic diagnostic drawing, not a customer photo or proof of photographic accuracy. It lets you reproduce the controls on a known source, then apply the same inspection steps to your own photograph.
3. Establish a low-tolerance baseline
Open the remover and load the source. Set tolerance to 20 and feather to 0, process, and download the PNG. A low tolerance can leave background shading behind; that is a useful baseline for seeing what more permissive settings change. Inspect the outside edge and the label separately. An enclosed pale label can remain because the remover follows connected pixels rather than deleting every matching color everywhere.
4. Compare tolerance before feathering
Process the same source at tolerance 50, then 90, keeping feather at 0. Compare each actual output below at equal size. Higher tolerance accepts a wider color distance and may remove more backdrop, but it can also reach pale product regions connected to an edge. Do not choose the highest value just because it produces the largest transparent area. Return to the lowest setting that clears the required background while preserving the subject.
5. Check the shadow and fringe
A contact shadow may remain, soften, or disappear as settings change. Decide whether you want the original shadow before removing it; this tool does not create a replacement. Open the PNG checker and view the cutout over dark and light surfaces. Pale edge pixels that still contain background color can show a halo even when their alpha is partly transparent. Feather controls an alpha transition near the tolerance threshold; it is not a brush or a general edge-decontamination tool.
6. Know when to stop adjusting
If a cap, label, highlight, or pale side of the product disappears, revert to the original and lower tolerance. If no setting separates the product from the backdrop, use a manual layer mask or reshoot against a contrasting background. Raising tolerance cannot supply object understanding. For a logo with enclosed white areas, follow the separate white-logo guide because the connectivity test and desired treatment of white artwork differ.
7. Deliver the right background
Download PNG to retain transparency and keep it as a master. A shop or marketplace may require an opaque white background; composite the approved cutout in your layout editor and inspect the final delivery file against that requirement. Check the destination’s current rules yourself. Reopen the exported file after any resizing or conversion, since a transparent source does not ensure every later format or service will preserve its alpha.
Questions before exporting
Does this remover identify products with AI?
No. It removes boundary-connected colors based on corner samples and your tolerance setting.
Why is part of my white product missing?
Similar subject and background colors can connect at the boundary. Lower tolerance, use a manual mask, or use a better separated source.
Does transparency mean the edge is clean?
No. Inspect over dark and light surfaces. Remaining background color in partly transparent pixels can cause a fringe.
Further reading: Adobe: why fringe pixels remain around a selection.
