Image Requirements

Viewrobe works from the product photography an apparel business already has. There is no 3D asset to commission per style and no second sample run. This page covers what makes a source image work well, what the shopper supplies, and what comes back.

The garment image

Either a flat-lay or an on-mannequin shot. Both are standard coverage photography, and most catalogues are full of them already. What matters is not production value but legibility: the render can only reproduce what the photograph actually shows.

The whole garment in frame. Cropped hems and sleeves cannot be reconstructed.
Even, diffuse light. Hard shadows across the cloth get read as part of the garment.
A plain, uncluttered background. It makes separating the garment cleaner and more reliable.
Colour you would stand behind. The render reproduces the colour in the source. If the source is warm, the output will be too.
The garment laid or hung as it is worn. Twisted sleeves and folded-under hems carry straight into the result.
Detail resolved. Print, lace, beading, texture and closures need to be visible at the resolution you supply, because they cannot be invented.
More than one angle where the back matters. In bridal and outerwear especially, a style shot only from the front can only be rendered from the front.

The shopper's photo

A single photograph, taken on a phone, uploaded on the product page. There is nothing to install, no account to create and no 3D body scan.

Real shopper photos are imperfect by nature, and a try-on has to cope with that: indoor light, a busy room, a turned torso, an arm across the body. That is the condition to test a system under, not a studio portrait. In practice the results are best when the shopper is facing the camera with the relevant part of the body unobstructed, and you can say so in the upload prompt without making it a requirement.

What comes back

Property	Value
Resolution	Up to 4K
Formats	PNG, WebP, JPEG
Fabric coverage	17 fabric types: cotton, silk, denim, leather, linen, chiffon, velvet, knit, jersey, polyester, nylon, wool, satin, organza, lace, mesh and fleece
Source formats accepted	Flat-lay and on-mannequin product photography
Shopper input	One photograph; no app, account or body scan

Exact file-size ceilings, request limits and turnaround for a given deployment are confirmed during onboarding rather than published here, because they depend on volume and on how the try-on is called.

A sensible way to test

Send a set that represents your catalogue honestly: your best hero flat-lay, a typical one, and the worst one you would still publish. Add a structured piece, something with a print, and something in a difficult colour. Then look at the output against the checks on the what is VTON page. A system that holds up on your awkward images will hold up on the rest.

Reference

Background

Questions

Do we need to re-photograph our catalogue?

Almost never. The try-on works from flat-lay and on-mannequin photography, which most apparel businesses already produce for production, quality control and the web.

Does the garment have to be shot on a model?

No. On-model photography is the output, not the input.

What formats does the try-on return?

PNG, WebP or JPEG, at up to 4K resolution, which covers product pages, wholesale linesheets and print.

What does the shopper have to provide?

One photograph. No app, no account, no 3D body scan.

Our photography is inconsistent across seasons. Is that a problem?

It affects how consistent the output set looks rather than whether it works. If you are producing a catalogue or linesheet where the images sit side by side, consistent source photography is worth the effort; for a product page it matters less.