Google’s AI try-on feature: you’re one selfie away from the perfect fit
Google is updating its AI Try-On feature – which now lets users virtually try on clothes and get a better understanding of how they would look wearing them in real life. You will no longer need to upload a full-body picture: a single selfie is enough.
The new system presents an image it gets after a user uploads a selfie, and it’s run by Google’s Nano Banana model – a part of Gemini’s 2.5 Flash Image technology. The result is a full-body digital image of the shopper wearing whatever they are virtually trying on.
Just like in a shop in real life, users can select their usual clothing size, and the tool will generate “studio-like images” to pick from.
One image can then be saved as the default for future try-ons. Also, users who are not fans of taking selfies can still upload a full-body photo. Another option is to use one of Google’s preset models with different body types.
As Google has previously stated, a newly generated AI model of yourself can show how a selected piece of clothing would look on an individual's body, as images are said to “accurately reflect how it would drape, fold, cling, stretch, and form wrinkles and shadows on a diverse set of real models in various poses”.
Google claims that models are based on people who differ in sizes from XXS to 4XL and who represent different skin tones, body shapes, ethnicities, and hair types.
This is not Google’s first attempt to mimic real-life shopping for clothes. Cybernews has previously reported on the company’s experimental app, Doppl.
Having entered the market in June, the app lets users see outfits on an animated digital version of themselves.
Users can upload images of clothing they see on Instagram, Pinterest, a retailer’s website, or other platforms, and Doppl will generate a still image or an AI video that shows how the outfit might look once it’s worn.
However, Google cautions that details such as how the outfit may fit a body may be inaccurate, as the app version at the time was still fresh.
Unlock more exclusive Cybernews content on YouTube.
