What is AI virtual try-on? A guide for Shopify fashion stores
How AI virtual try-on works, what it can and can't show, which products it suits, and how to add it to a Shopify store without slowing it down.
The Mirava team 4 min read
AI virtual try-on is a store feature that takes a shopper’s photo and an ordinary product photo and generates a new image of the shopper wearing the product, so they can see how it looks on them before they buy.
Online fashion has always had the same gap: a shopper can see the product, but not the product on them. Virtual try-on closes part of that gap. This guide explains what AI virtual try-on is, how it differs from the AR filters you may have seen, and what to look for if you want to add it to a Shopify store.
What virtual try-on means today
Virtual try-on is any tool that shows a product on the shopper before they buy. There are three broad kinds:
- AR try-on overlays a 3D model on a live camera feed. It works well for rigid items with a known shape, like glasses, watches, makeup or sneakers, but needs a 3D model of every product.
- Size and fit recommenders ask for height, weight or past purchases and suggest a size. They answer “which size?” but not “how does it look on me?”
- AI image try-on takes a photo of the shopper and an ordinary product photo, and generates a new image of the shopper wearing the product. No 3D modelling, so it works with the catalog you already have.
This guide is about the third kind, which is what Mirava does.
How AI image try-on works
Modern image models can edit a photo while keeping most of it unchanged. A try-on request gives the model two images, the shopper and the product, and asks it to dress the person in the product. Good results keep the face, pose, body shape and background exactly as they were, and change only the clothing or accessory.
In practice the flow on a product page looks like this:
- The shopper taps Try it on next to Add to cart.
- They upload a photo, or pick a ready-made model if they’d rather not.
- The photo is checked: one adult person, no unsuitable content.
- The AI generates the look. With current models this takes around a minute.
- The shopper compares before and after, and can add the product to the cart.
Because generation takes time, a good widget lets shoppers keep browsing while they wait instead of staring at a spinner.
What it shows well, and what it doesn’t
AI try-on is strongest at answering “does this suit me?”: color against skin tone, the silhouette of a dress, the length of a coat, the scale of a bag against the body.
It is weaker at exact fit. The model has never measured the garment or the shopper, so it can’t tell whether a medium will pull across the shoulders. Be honest with shoppers about this. A short note such as “AI preview: actual fit and color may vary” sets the right expectation, and it belongs in the try-on window itself.
Which products it suits
AI image try-on works for most things worn on the body:
- Tops, shirts, knitwear and dresses
- Pants, skirts and matching sets
- Outerwear such as jackets, coats and blazers
- Accessories: bags, hats, scarves, sunglasses, shoes and statement jewelry
Accessories are a strength of photo-based try-on: the shopper sees the bag, hat or scarf with the rest of their outfit, which a close-up camera filter can’t show. Two caveats: eyewear try-on shows the style of a frame, not whether it fits the width of the face, and shoes need a photo where the feet are visible, or at least room below the knees.
It is less useful for products where the look depends on tiny details (delicate rings, fine chains), or where the shopper can’t judge much from a photo (socks, underwear, gift cards). Most try-on apps let you hide the button on those, for example with a product tag.
What to check before you choose an app
Image quality on your products. Try it on your own best sellers, not the demo. Look for faces that stay recognisable and products that keep their pattern and color.
Speed and the waiting experience. Expect around a minute. The important part is what the shopper can do meanwhile.
Cost control. Every generation costs money. Look for daily limits per shopper, a store-wide daily limit and a monthly cap, so a busy weekend can’t surprise you.
Privacy. Shopper photos are personal data. Check how long they are kept, whether they are used for training, and whether the app handles Shopify’s privacy webhooks.
Measurement. Usage numbers tell you people like the feature. They don’t tell you whether it sells more. Look for a way to compare visitors with and without try-on. We cover this in how to measure if virtual try-on lifts sales.
Theme fit and speed. The button should match your store, and the widget shouldn’t add weight to every page load.
Getting started on Shopify
With Mirava, setup is an app install, one app embed and one block in the theme editor. The installation guide walks through it step by step, and the Free plan includes 30 try-ons a month so you can see results on your own products first.