PicBarn

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.