Fashion5 min read
AR in fashion retail: four kinds of virtual dressing room
The four technologies sold as AR in fashion retail, from in-store mirrors to phone try-on, what each can and cannot do, and what to ask a vendor.
By CharpstAR · 21 May 2024 · Updated 22 Sept 2026
"Virtual dressing room" is a marketing phrase that covers at least four different technologies, and they do different jobs. One puts a life-size model on a store floor. One puts a pair of shoes on your feet through a phone camera. One pastes a jacket onto a selfie. One tracks your face and fits a frame to it. When AR in fashion goes wrong, it is usually because a retailer bought one of these expecting another. This article covers each one, how it works and where it stops working, since that is where retailers lose money.
1. In-store AR: signs, mirrors and virtual models
Zara's 2018 experiment is the case everyone points to. For two weeks in April 2018, in 120 stores, shoppers pointed the Zara app at in-store signs and saw international models appear life-size, posing in the current collection, with the outfits buyable from the app. It made the news, and Zara did not repeat it. It worked as a promotion, but only for people who had the app installed.
Smart mirrors are the other in-store form. A screen with a camera shows the shopper with a garment overlaid, or shows the same garment in another colour. They work best at the rail for checking colour and style. For fit they are weak, because the overlay does not know the shopper's measurements.
Limit: in-store AR only reaches shoppers who are already in the store, and it needs either an app or a fixed installation. Retailers mostly use it for brand campaigns, and it does little for day-to-day sales.
2. Phone try-on for rigid items: shoes, glasses, watches
Phone try-on works well here because the objects keep their shape. A shoe, a frame, a watch or a bag held at arm's length is rigid. A modern phone tracks feet, faces and wrists well, so the object can be drawn at true scale in the right place. Gucci's 2020 Snapchat campaign put four sneaker styles on shoppers' feet with a shop button, and Snapchat called it the first global shoppable AR try-on. Eyewear try-on is the most mature of the lot, since the face is what a phone camera tracks best.
Limit: it tells a shopper whether something suits them and whether the width is right. For anything soft it says nothing about size. The result also depends on the 3D model, and a badly modelled shoe on a real foot looks worse than a photo.
3. Garment try-on on a selfie or a live body
This is what most people mean by a virtual dressing room, and it needs the most care. The current versions take a photo or a live camera view and drape a garment over the body using a body-pose estimate. The good ones look convincing. None of them measures you. They estimate a silhouette from a single camera and stretch the garment to it, so what you see is the garment on roughly your shape in roughly your pose, at a size the system guessed.
As a styling tool that is fine: does this colour suit me, does the cut work with these trousers. As a fitting tool it shows an estimate that looks more certain than it is. The returns it can prevent are the "I did not like it on" kind. It does little about "it was too small".
Limit: it makes shoppers more confident about how something looks, and it does not measure them. A vendor who says it will end size returns is promising more than the technology can do.
4. 3D models on the product page
The least glamorous form is a 3D model of the item in a viewer on the product page, with an AR button for the phone. For fashion this means a bag you can turn over and open, a shoe you can look at from below, a jacket shown on a model of a stated size with the lining visible. It does not try to put the garment on you. It gives the shopper roughly what they get in a store by picking the item up.
The numbers are steadiest here too. Shopify's merchant data put conversion 94 percent higher on interactions with 3D or AR content, and Rebecca Minkoff, a fashion brand, saw shoppers who used the 3D viewer 44 percent more likely to add to cart and AR users 65 percent more likely to buy. Those shoppers chose to open the viewer themselves, which inflates the figures. Still, the result came from a fashion catalogue on the brand's own site, with no app involved.
Limit: you need proper models of the real products, which means a modelling budget and a way to keep them up to date.
The four forms compared
| Form | Where | Needs an app | Fixes fit | Best for |
|---|---|---|---|---|
| In-store signs and mirrors | Store | Yes or fixed install | No | Brand moments, colour choice |
| Rigid try-on | Phone, social or web | No on the web | Yes for width and suitability | Glasses, shoes, watches, bags |
| Garment try-on | Phone or web | No | No, confidence only | Styling, sharing |
| 3D on product page | Web, any device | No | Partly, via detail and stated sizes | Whole catalogue, everyday selling |
What to ask a vendor
Ask what the system measures, and listen for the word "estimate". Ask to see the 3D model up close, since that is where you can judge the quality. Ask whether it runs from a web link or needs an app. Web versions now open AR on iPhone and Android from a normal link, so an app is optional. Ask how a new season's products get added and what that costs. Finally, ask for return and conversion numbers from a comparable retailer instead of the platform average.
What we build
We build the fourth form, plus rigid try-on for eyewear. We do not build garment try-on. If your catalogue has rigid products where the shopper wants to know whether it suits them, or products a shopper would pick up and turn over in a store, a 3D model on the page is the form of AR in fashion retail that helps sales all year, long after a launch campaign has ended. We make the first product free so you can judge the model before committing to anything. Send us one product.
