AR Try-On

What is AR try-on?

AR try-on (augmented reality try-on) uses a device's camera to superimpose a 3D product model onto the shopper in real time. The product tracks the user's movements: turning the head shows glasses from a new angle; moving a foot rotates a sneaker.

It is the most established form of virtual try-on for rigid products whose shape does not deform with the body: eyewear, watches, rings, makeup, and footwear.

How it works

1. Tracking: the camera feed is analyzed frame by frame with face tracking or pose estimation to detect anchor landmarks (eyes, wrist, feet).
2. 3D model loading: a pre-built 3D asset of the product (e.g. a .glb file) is fetched.
3. Rendering: the model is anchored to the landmarks and composited into the live video, with lighting and occlusion handling for realism.
4. Interaction: the shopper switches colors or models without leaving the camera view.

AR try-on can run in a native app or directly in the browser (often called WebAR), which removes the app-download step.

AR try-on vs. generative AI try-on

	AR try-on	Generative AI try-on
Output	Live, interactive video overlay	Static generated image
Typical products	Rigid items: glasses, watches, shoes	Apparel: tops, dresses, knitwear
Per-product setup	3D asset per SKU	None (uses existing product photos)
Fabric realism	Limited	High (drape, wrinkles, lighting)

The two approaches are complementary: AR suits fixed-shape products where live interaction matters; garment transfer suits clothing, where fabric must be re-rendered per body and pose.

FAQ

Questions, answered.

Why is apparel hard for AR try-on?↓

Fabric deforms with every body shape and movement. A rigid 3D garment overlay can't reproduce realistic drape in real time, which is why apparel try-on generally uses generative AI on a photo instead.