Photo-based virtual try-on: how it works

On the product page, shoppers see the garment on a model, on a mannequin or laid flat on a white background. The question in their head is a different one: how would it look on me? Until they get an answer, doubt remains, and doubt leads to putting off the purchase or ordering with a return already in mind.

Virtual try-on exists to answer that question. There are several approaches, and they do not all ask the same effort of the shopper. In this article we look at how photo-based try-on works, what it needs to give good results, which privacy questions to ask and when it makes sense to introduce it.

What shoppers want to see before they buy

People buying fashion online are not only looking for technical details. They are trying to picture themselves wearing the item. The questions they ask are simple and very personal:

Does the colour suit me? A shade that works on the model may not work with their own complexion.
How does it look on my figure? The same jacket gives a different impression on different body shapes.
Does it go with my style? Shoppers want to know whether the item fits the way they usually dress.
Can I see myself in it? Seeing the garment on someone very different makes the decision harder.

Catalogue photos, even good ones, only partly answer these questions. They show the item at its best, but on someone else. Virtual try-on tries to close that gap.

Photo-based try-on and other approaches

The term virtual try-on covers solutions that are quite different from one another. Broadly speaking, there are three routes.

Avatars built from measurements. The shopper enters their details or picks a body shape, and the garment is shown on a reconstructed figure. It takes some setup, and the result is a representation, not the real person.
Live camera try-on. The garment is overlaid on the shopper's camera view. It needs a camera, good light and often an app or specific permissions on the device.
Photo-based try-on. The shopper uploads a photo of themselves and receives an image of themselves wearing the item. No measuring, no setup and no standing in front of a camera.

The strength of photo-based try-on is simplicity. Shoppers do something they already know how to do, which is choosing a photo of themselves, and they see the result on their own image. For many shops it is the most natural starting point, because it asks shoppers to learn nothing new.

It is worth being clear about what it does not do: a generated image shows how the garment looks when worn, it does not replace a size recommendation. Size needs a dedicated tool that works from the garment's measurements.

Which product images work best

The quality of the result depends a lot on the source images. You do not need special shoots, but a few precautions help a great deal.

Clean packshots. A plain background, no props, the garment clearly separated from the backdrop.
The whole garment in frame. Sleeves, collar, hem and fastenings should all be visible, with nothing cropped.
One item per image. If several garments appear in the photo, it is harder to tell which one should be tried on.
Neutral light and accurate colours. The colour shoppers see on themselves should match the real product.
Good resolution. Fabrics, prints and details read better when the original image is sharp.
Consistency across the catalogue. When a collection's photos follow the same rules, the results are more consistent too.

It is also worth giving shoppers a few tips for their own photo: full length or a clearly visible upper body, good light, a front-facing pose and clothing that is not too bulky. Two lines next to the upload button are enough.

Privacy: what to ask your provider

With photo-based try-on, shoppers hand you an image of themselves. That is personal data and needs careful handling. Before choosing any solution, ours included, make sure you have clear answers to these questions:

How long is the shopper's photo kept, and where?
Are the photo or the generated image used to train models?
Can the shopper ask for deletion, and how?
Who has access to the images, and with what controls?
What are shoppers told before uploading, and how is consent collected?
How are your privacy notice and your contract with the provider kept up to date?

Get the answers in writing and pass them on to shoppers in plain language. A clear explanation next to the photo upload builds trust and removes doubts.

When it pays off

Photo-based try-on is not equally useful for every catalogue. It makes most sense in certain situations:

Items where appearance matters a lot, such as dresses, outerwear, knitwear and pieces with distinctive colours or prints.
Collections with many colourways, where shoppers struggle to choose from product photos alone.
Shoppers who buy mainly on their phones, for whom uploading a photo is a natural gesture.
Catalogues with well organised product images, because the preparation work is small.

As with any tool on the product page, the effect should be measured. Our advice is to start with an A/B test on part of the catalogue or traffic and look at conversions, time on page and returns together, keeping in mind that returns arrive weeks after purchase.

How We Wear does it with FitFlow

FitFlow is the We Wear photo-based virtual try-on. It needs only two images: the product image and the shopper's own photo. The AI generates a photorealistic image of the shopper wearing the garment. There are no measurements to enter, no avatar to configure and no AR camera to switch on.

FitFlow runs in the browser, on mobile too, with no app to install. It works from the product feed you already have, such as your Google Shopping or Meta feed or a CSV file, so there is no separate catalogue to prepare.

It can be used alongside SizeFlow, our size recommendation tool: FitFlow shows how the item looks, SizeFlow says which size to choose. Together the two answers cover the shopper's main doubts on the product page.

If you would like to see FitFlow on your own products, book a 30-minute demo.

Frequently asked questions

What is photo-based virtual try-on?

It is a system where shoppers upload a photo of themselves and receive an image of themselves wearing the chosen item. They do not enter measurements, configure an avatar or use a live camera.

Which product images work best for photo-based try-on?

Clean packshots on a plain background work best, with the whole garment in frame, one item per image, neutral light, accurate colours and good resolution.

How does FitFlow work?

FitFlow needs only two images, the product image and the shopper's own photo, and the AI generates a photorealistic image of the shopper wearing the garment. It runs in the browser on mobile too, with no app, works from the product feed (Google Shopping, Meta, CSV) and can be used alongside SizeFlow for size.

Photo-based virtual try-on: how it works, when it pays off