Glossary · AI imaging
Virtual try-on
Virtual try-on is a technique that shows a product, usually clothing or accessories, on a person who was not photographed wearing it, either live for shoppers or as generated on-model images for listings.
The phrase has two meanings. Shopper-facing virtual try-on, sometimes called a virtual dressing room, overlays a product on the shopper's own camera image. Seller-side virtual try-on generates on-model listing photos from a photo of the product alone. Listing pages are concerned with the second.
Generated models bring disclosure and accuracy rules. Amazon asks for the keyword contains-synthetic-performer in the IPTC/XMP dc:subject field of any image showing a photorealistic AI-generated person, written before upload. TikTok Shop's AIGC rules allow generated images only when they accurately represent the real product, and say they must not change its size, colour, shape or features. Walmart says AI-generated images must be truthful, accurate and not misleading.
The usual failure is a generator that redraws the garment instead of dressing the model in it. Patterns change scale, logos move or lose letters, buttons multiply, and the hem moves up or down. The result looks plausible but no longer matches what the buyer receives, which is exactly what the platforms prohibit. Other mistakes are forgetting the Amazon tag, and using try-on for categories where TikTok Shop does not allow child models.
Check every try-on image against the real garment before it goes live: colour, length, print placement and fastenings.
PicBarn's virtual try-on tab keeps the garment's pixels and generates the model and setting around them, and exports carry the Amazon contains-synthetic-performer XMP tag.
See also
- On-model imageAn on-model image shows clothing, jewellery or accessories worn by a person, real or AI-generated, so buyers can judge fit, drape and scale on a body.
- contains-synthetic-performer tagcontains-synthetic-performer is the keyword Amazon asks sellers to write into an image's IPTC/XMP dc:subject field before uploading any photo that shows a photorealistic AI-generated person.
- AI-generated content disclosureAI-generated content disclosure is the requirement to tell buyers or the platform when an image or video was made or significantly changed with AI, through a label, toggle, checkbox or metadata tag.
Related
- AI virtual try-on — put your clothes on a modelVirtual try-on for sellers: upload a flat-lay or ghost-mannequin shot, pick an AI fashion model, and get on-model photos that keep every stitch of the garment. Amazon-, Shopify- and TikTok-ready sizes.
- How to disclose AI-generated models in product listings: Amazon, TikTok Shop, Etsy and ShopifyAmazon requires an XMP keyword on any image with a photorealistic AI-generated person; TikTok labels AI content and requires listing images to match the real product; Etsy requires one unedited photo; Shopify sets no rule. The exact tag, how to add it with exiftool or Preview, and a pre-publish checklist.
- Ghost mannequin vs AI model photos: which one sells the garment?Ghost-mannequin shots show shape; on-model shots show fit. When to use each, what each costs, and how to get both from one photo.