You’ve heard it, online shopping for clothes is a terrible idea. You scroll through countless options, picking the right size, fabric, and colour, hoping the shirt or dress will fit and look just like it does in the product photos. Yet too often, what gets delivered doesn’t match what you added to your cart.
According to a survey by Fashion Network, 27% of consumers simply avoid buying clothes online for sizing reasons, and the same article notes that 45% of shoppers prefer to shop in-store instead, specifically to ensure an item fits before purchasing, even though it takes time and effort.
But at a time when generative AI is peaking, this is about to change.
Among Google’s latest AI model launches is a tool that lets you virtually try on clothes using a photo.
Rolled out in Search Labs in the U.S., the new virtual try-on technology shows you how clothes would fit your body, helping you shop with more confidence.
Alongside Google’s new Try-On Technology is an AI shopping assistant that helps you find inspiration, think through what matters most, narrow down options to match your style, and even spot the right time to buy at a good price.
How to use Google Virtual Try-On feature
Simply upload a full-body photo of yourself (preferably with a plain background). Google Shopping’s Try-On AI Model does the rest.
While browsing Google Shopping for shirts, pants, skirts, or dresses, just tap the “Try it On” icon on a product listing to generate images of yourself wearing that clothing, fitting perfectly in the exact photo pose. It’s not just a flat image slapped onto your photo. The AI model understands how listed clothes move, stretch, and drape on different bodies, making it look real.
You can even save the images or share them with friends for a quick “What do you think?” comment.
Google recommends using your own photos for better accuracy and privacy. The feature is still experimental and currently limited to U.S. users through Search Labs. More features are expected in the final version. Generating images in different poses could be a useful addition.
