Virtual Try-on

Virtual try-on literally means “virtually trying something on.” It refers to software that places a piece of clothing, glasses, or a lipstick onto an image of you. The result looks like a photo in which you’re actually wearing the product. Some providers work with an uploaded photo, others with the live image from your phone camera. The purpose is always the same: you’re supposed to see how something looks on you before buying it, without having to order it first. The term appears mainly in online shops and in news about major retail corporations.

What retailers gain from the fitting room in the browser

In online fashion retail, a large share of ordered goods gets sent back. For clothing, return rates often lie between 30 and 50 percent, depending on the provider and country. Every return costs postage, staff time for inspection, and sometimes the entire item if it’s no longer sellable. Anyone who lowers that rate by even a few percentage points saves considerable sums.

The second reason is the sale itself. Many people abandon a purchase because they’re unsure whether something will suit them. A preview on their own body removes some of that uncertainty. This works especially well for glasses and cosmetics, since fit barely matters there and only appearance counts.

There is criticism, too. The images often look flattering, because the software smooths out wrinkles and lets fabric fall ideally. What fits perfectly on screen can pinch in reality. Consumer advocates also point out that trying things on this way produces photos of your body, meaning particularly sensitive data.

How the garment gets onto the body

The first step is always analyzing your image. A program searches it for distinctive points: shoulders, elbows, hips, and for glasses, the eyes and bridge of the nose. From these points, a rough description of your posture and proportions is created. You can picture it as a stick figure overlaid on your photo.

Then the product comes into play. Older systems simply distort a product photo so that it matches the stick figure. Modern systems, by contrast, use image generators — AI models that recreate images pixel by pixel. They receive two inputs: your photo and the photo of the garment. From these, they paint a new image in which both are merged together.

The difficult part is the details. The fabric has to form creases exactly where the body bends, and arms need to appear in front of the sweater, not behind it. The pattern must not shift either — a striped shirt has to stay striped. This is exactly where weaker systems fail, and you can spot it from smeared hands or warped lettering on T-shirts.

From shopping apps to filters on social networks

You most often encounter the technology in online shops and their apps. Among eyewear retailers it has been standard for years, as it has among cosmetics brands, where you can test lipstick shades directly in the camera image. Large platforms like Amazon, Google, or Zalando have built their own features for shoes and clothing. Furniture apps work on the same principle too, except there the cabinet ends up in your living room instead of on your body.

A close relative is augmented reality, or AR for short: this refers to camera images into which the computer inserts additional objects. Face filters on social networks use the same underlying technology as a virtual glasses try-on. The only difference lies in the purpose — entertainment on one side, sales on the other.

In business news, virtual try-on usually comes up in connection with return rates, or when a corporation acquires a startup for it. A common misconception is that the software determines the correct size. That’s a different task — size recommendation — and it’s based on measurements and purchase history, not on a generated image.