Virtual “Fitting Room” with a variety of skin tones and body types
According to Google’s Product Director, Lilian Ricon, the idea of this feature is to make consumers feel confident in buying clothes online, avoiding disappointments and allowing the user to visualize how the clothes will look on their body with more detail, even before they buy it.
According to Google, 42% of online shoppers don’t feel represented by model images, and 51% feel dissatisfied with an item they purchased online because it looked different than expected.
Virtual clothing fitting uses a diffusion-based generative AI model, which is trained by adding Gaussian noise to an image (essentially random pixels) that the model learns to remove to generate realistic images. This makes Google’s AI model represent images more realistically, no matter what angle or pose they are in.
It’s a bit wild how all of this still feels straight out of sci-fi, even in 2025. Sometimes it hits you when you’re scrolling through endless photos, and suddenly one of these virtual try-ons pops up—convincingly close to reality, but with just enough polish that you know it’s AI at work. There’s still this subtle disconnect between the algorithm’s ideal and the actual messy way clothes fit people, but it’s closing fast. If you compare what we had just a couple of years ago, the progress is honestly dizzying. And yet, for anyone who’s ever bought a shirt online only to have it fit like a potato sack, these tools are overdue.
What doesn’t always make the headlines is the sheer challenge of capturing nuance—like how fabric drapes on different postures, or how lighting changes the look of a color against your skin tone in a real room. Google’s model aims to simulate all that, but even the best tech can’t fully account for the surprises of real life (not yet, anyway). I’m guessing we’ll still see frustrated customers posting awkward try-on photos for a while, but maybe a little less often. People are already reporting fewer returns and less buyer’s remorse with these updates, according to scattered posts on Reddit and some early user surveys. If the trend holds, maybe we’re actually headed for a point where “what you see is what you get” isn’t just a marketing phrase.
Refine a product and find what you really want
Google has also launched a feature that helps you find products based on other options you’ve tried. For example, you like a shirt but want a cheaper version? Or did you find a pair of jeans but want a different pattern? In this refinement created by Google, shoppers can tweak products until they find the perfect piece.
This is thanks to machine enhancements and new visual matching algorithms, allowing you to refine parts using inputs such as color, style and pattern.This feature is also available for beginners directly from the product listings.
The future of AI
What we previously only saw in movies – like “Clueless” with the virtual closet of the protagonist Cher – has become more than real and now AI technology is present during different moments of our day.
With that continual boom in Artificial intelligence usage day by day, the concern regarding its regulation has also become a point to be discussed. The EU, for example, has this week approved a bill to regulate AI.
One of the points raised during the discussion of the bill was the prohibition of the use of real-time facial recognition in public spaces, except in cases of criminal activity and court authorization. It is possible that other countries will also start creating this type of law, aimed at Artificial Intelligence usage, as a way to regulate the market and protect consumer rights.
It is important that large technology companies remain attentive to this movement so that their ideas do not exceed the limits imposed by the jurisdictions of each country, and do not interfere with the development of a technology that can be beneficial to all.
Furthermore, we are looking forward to testing these new Google Shopping features and making online shopping more democratic and realistic.
