Comparison
AI background removal vs manual cut-out for white backgrounds
Every marketplace main image starts with a clean cut-out on white. You can let a segmentation model do it in seconds or trace it by hand with a clipping path. Here is the trade-off.
| Category | AI background removal | Manual cut-out |
|---|---|---|
| Cost | In PicBarn the main image is 1 credit per image, about US$0.25, with failed runs refunded. | Retoucher time for every image; complex products take far longer than simple boxes. |
| Time | Seconds per image, and batches are no slower per image. | Minutes for a box, much longer for a chair, a bicycle or a fluffy toy. |
| Edge quality | Good on solid products; can soften or chew fine detail such as hair, fur, mesh, fringing and thin cables. | Exact on hard edges with a clipping path; hair and fur need separate masking and skill. |
| Transparent and reflective products | Glass, clear plastic and chrome confuse segmentation; check for background showing through or halos. | Handled deliberately, with the retoucher deciding what stays visible through the product. |
| Hitting pure white | Must output exactly 255, 255, 255; check that soft edges and shadows do not leave grey near the frame. | Same check: a hand cut-out placed on 254 white still fails Amazon's 255, 255, 255 and Walmart's 255/255/255. |
| Framing | Can be automated to Amazon's 85% fill, square crops and Walmart's "crop close" rule. | Set by hand per image, which is slow but easy to adjust. |
| Platform rules | A background change is an edit: allowed on Amazon, Walmart and Etsy gallery photos, but Etsy still needs one unedited photo, and transparent PNGs turn black on Etsy. | Exactly the same rules; the method is not judged, only the result. |
| Best for | Catalogues, solid products, and refreshing many main images at once. | Hero products, glassware and jewellery, and anything AI got wrong. |
Verdict
For most products AI background removal is good enough and far cheaper, and the main thing to check is the numbers: the background must measure 255, 255, 255 and the product must fill about 85% of the frame for Amazon. PicBarn's main image tab does both in one run, from your own photo.
Keep a manual cut-out for the products AI handles badly: glass, chrome, fine jewellery chains, hair and fur, and anything where a chewed edge would be visible at zoom. Whichever method you use, check the edges at full size and the corner pixels before upload.
Questions
Why does my white background fail Amazon's check?
Usually because it is not exactly 255, 255, 255. Soft shadows, anti-aliased edges and JPEG compression can leave near-white pixels. Measure the background, not how it looks on screen.
Should I export the cut-out as a transparent PNG?
Not for marketplaces. Amazon and Walmart main images need a white background, and Etsy turns transparent areas black. Flatten the cut-out onto pure white.
Can AI remove backgrounds from glass products?
Often not cleanly: transparent and reflective surfaces confuse segmentation. Check for background showing through or grey halos, and use a manual cut-out if the result is wrong.
Related
- Amazon main image requirementsAmazon's main image rules from Seller Central: pure white RGB 255, 255, 255, the product filling 85% of the frame, no text or mannequins, and the file specs behind them.
- Amazon main image: how to get a true pure-white background (RGB 255) without PhotoshopAmazon wants the main image on pure white, RGB 255, 255, 255, with the product filling 85% of the frame. Why a white sweep is never white enough, how to check the pixels, three free ways to fix it, and the edge faults that still get an image removed.
- Amazon image requirements checkerFree: upload an image, see which marketplace rules it passes.
- Walmart Marketplace image requirementsWalmart Marketplace image rules: seamless white 255/255/255 main image, 2200 x 2200 px recommended, 1500 x 1500 for zoom, 1:1, JPEG/PNG/BMP under 5 MB, and truthful AI images.