Amazon's main-image policy requires a true RGB 255/255/255 white background, the product covering at least 85% of the frame, no logos, text, or graphics. The penalty for non-compliance is suppression. Here is the exact step-by-step workflow to ship policy-compliant white-background fashion photos with AI in under five minutes per garment.
Step 1: prep the garment file
A clean source file makes the rest of the pipeline reliable. Lay the garment flat on any surface, shoot it on a phone in daylight without flash, and crop tightly. Front and back if the garment has both relevant. JPEG or PNG, under 50 MB. The garment does not need to be on a perfect background — it needs to be in focus and recognisable.
Step 2: pick the right Apiway template
Two templates handle Amazon main images, and the choice depends on whether you want a model.
- Hanger / no-model main image: Ghost mannequin. Cleaner, faster, perfect for Amazon's strict on-pure-white rule. Use this when the listing is on a brand-agnostic Amazon page where the product is the entire subject.
- Model on main image: White Studio. Use this when the brand has a styled-on-model look and Amazon allows on-model main imagery for the category (apparel does).
Step 3: aspect ratio and framing
Amazon's main image accepts square (1:1) up to 2000×2000 pixels minimum recommended. In White Studio, pick the 1:1 aspect ratio. Pick a framing that places the product at 85%+ of the frame — full-body framing usually wastes vertical space; mid-shot or close-up performs better here.
For Ghost mannequin the framing is essentially fixed (the garment fills the frame), and the only setting that matters is whether to include front-and-back or front-only.
Step 4: the pure-white pipeline does the rest
Both templates run Apiway's post-processing pipeline that guarantees a true #FFFFFF background — subject segmentation, recomposite, edge refinement, tone correction. (Detail: why we re-composite onto pure white.) The output is Amazon-compliant out of the box. No Photoshop pass required.
Step 5: verify the corner pixel before upload
One final sanity check that takes ten seconds. Open the result in any image editor (Preview on Mac, Paint on Windows). Use the eyedropper on the top-left corner. The colour should read 255, 255, 255 in RGB. If it does not, regenerate — do not retouch by hand.
Step 6: file naming and Seller Central upload
Amazon ingests product images under predictable naming. Use the SKU as the filename. PNG or JPEG. RGB colour space. Upload through Seller Central or your inventory tool of choice. Keep the original garment file in a folder so future regenerations stay consistent with the same source.
Real cost and time per SKU
Five minutes of operator time per SKU once the workflow is established (most of which is the upload step). Compute cost is approximately one credit per shot — one credit equals one US cent. For a 200-SKU launch, the all-in image cost is in the single-digit-dollar range plus operator hours.
Try it on a real listing
Pick one current Amazon listing where the main image looks slightly grey or slightly off. Open a free account, run the same garment through Ghost mannequin or White Studio, and replace the listing image. Amazon's suppression checks loosen and the conversion baseline often lifts within the first week of fresh imagery.