How to remove background from clothing photos with AI in…

Striped tank and beige shorts laid on a marble table with fur and jewellery

Same striped tank and beige shorts as a ghost mannequin catalog shot on pure white

Removing the background from a clothing photograph is the most basic AI image task in fashion ecommerce, and the most commonly done badly. The cheap path produces hard-edged cutouts with halos around hair, lost shadow detail, and backgrounds that fail Amazon's pure-white #FFFFFF check. The right path produces a catalog-grade isolated garment that ships directly to a marketplace listing. This is the practical 2026 how-to for removing backgrounds from clothing photos with AI.

Why clothing background removal is uniquely tricky

Clothing fights a generic remover. Sheer fabric — chiffon, organza, lace — blends with the background by design. Naive removal either deletes too much, cutting garment with the wall, or too little, leaving the old colour visible through the sheer.

A collar is a fine edge. A cheap cut hardens it into a cartoon outline, or leaves a halo of leftover pixels. The table under a dress is often still in the file. A generic tool deletes half of it and calls the job done.

Sheer white blouse photographed on a hanger against a grey wall

Same blouse cut out with a generic background remover

Step 1: pick the right tool for the job

For a Stories sticker or a one-off transparent PNG, generic tools are fine. remove.bg and Photoroom are fast. We even ship a free browser tool if you only need a PNG and you are not listing on Amazon.

For a marketplace listing the output has to be the final asset. That is Ghost Mannequin when you want the garment alone — hollow neck, body volume, no person. It is White Studio when you want the same garment on a person. Both composite onto #FFFFFF. ChatGPT will not. A prompt that says “white background” will not.

Yellow floral midi dress as a ghost mannequin catalog shot on pure white

Same yellow floral midi dress on a model in White Studio on pure white

The decision is simple. Amazon, eBay, Walmart main image — Ghost Mannequin. A Shopify PDP that needs a face — White Studio. Own-site hero or a sticker — a generic remover is enough.

A free account opens Ghost Mannequin after sign-up, not a blank chat. That is the point of this guide.

Try Ghost Mannequin free

Step 2: prep the input photograph

The quality of the output is bounded by the input. Shoot in even light, with no harsh shadow running into the background. Use a backdrop that contrasts with the garment — not white-on-white, not black-on-black — and keep the garment isolated: no hanger in the collar, no second SKU in the corner. A phone photo is enough if those things are true.

1,500 px on the long edge is enough to start. 3,000×3,000 if you are shipping catalog-grade. Folded piles and crumpled baskets produce melted output. The garment has to be readable as a garment.

Bad input: jeans crumpled on a dark floor in harsh shadow

Good input: jeans laid flat in even light, garment readable as a garment

Brands that ship volume standardise capture: same table, same window, same distance. That discipline pays back on every SKU. If all you have is a supplier photo, start with how to convert supplier flat-lays into catalog imagery — then run the cleaned file through Ghost Mannequin.

Step 3: handle sheer and translucent fabric correctly

Sheer and translucent fabric is the failure mode that catches most generic tools. A chiffon blouse photographed on a coloured background will have the background showing through the fabric where the fabric is sheerest (often the upper sleeve or yoke). Background removal has to delete the background pixels behind the model but preserve the fabric's actual sheerness against the new white background.

The right approach is to use a tool that maintains alpha-channel transparency on sheer fabric rather than forcing a binary cut. Apiway's ghost mannequin template handles this on garment-only inputs. For on-model sheer-fabric photographs, the workflow typically involves rendering the garment fresh through White Studio rather than trying to retouch the existing photograph; the rendered output is cleaner than the retouched one in most cases.

Ivory lace bow blouse as a ghost mannequin shot, fabric still translucent against pure white

Step 4: Amazon pure-white #FFFFFF compliance

Amazon's product image policy requires the main PDP image background to be pure white #FFFFFF for apparel listings. “Almost white” is grounds for suppression. Generic background removal tools often produce a near-white gradient that fails Amazon's automated check. Verify the background hex on every output before submission to Amazon.

Square Amazon-style main image of an orange crochet dress as a ghost mannequin on a pure white frame

Apiway's catalog templates guarantee #FFFFFF background output at the platform level. For brands shipping serious Amazon volume, this is the single most-cited reason to use category-purpose tools rather than generic background removers. The marketplace suppression cost of a single bad upload exceeds the per-image cost difference between the two tool tiers.

Start free with 100 credits

Step 5: batch processing at scale

A catalog of 50 to 200 SKUs needs the same garment-alone shot on every listing: same pose, same white, same crop. That used to mean a studio week of clipping paths. In Apiway it is one pass in Ghost Mannequin. Photoroom will batch a PNG. It will not keep the hollow-neck catalog look.

Open Ghost Mannequin. On Garments, add each SKU. Front of the garment is required. Back if you shot it. You can load several garments in the same run and generate them together.

Step 6: QC the output

Check the hex on a sample. Every tenth frame is enough once the template is stable. Check edges on collar, cuff, sheer, and hair if a model is in frame. Check the crop against the destination: Amazon wants the product at 85% of a 1:1.

Confirm no leftover table, no leftover plant, no leftover hanger. The QC is one job: is this white, and is this the garment. If both hold, ship. If either fails, rerun that SKU.

Getting started with AI background removal

Sign up for a free Apiway account. It opens Ghost Mannequin after registration, not a blank chat. Upload five of your most-listed flat-lays. Look at the collar. If the hollow neck and the #FFFFFF both hold, the rest of the catalog will too.

New accounts get 100 one-time credits — enough for a PDP thumbnail and a couple of extra poses. If you are shipping to Amazon, confirm the hex on one test listing before you batch the season. The three settings that actually decide a ghost shot are in how to ghost-mannequin a garment with AI.