To empower potential customers and help them visualise how items will look on their bodies, people can use a virtual 'try-on' feature to 'test' clothing before they buy it.
How it works
Using AI, Google combines a user-uploaded photo, analyses your product images and generates a new image, showing how different garments might look on the customer's body.
While not necessarily a perfect representation of fit, this feature enhances the discovery process, encouraging exploration of different items and potentially driving engagement with your products among a wider range of potential buyers across the web. The accuracy of the representation depends on the quality of the images that you provide and on the user's photo.
Eligibility and availability
Any merchant with a Shopping feed who is eligible to show free listings is automatically opted in to this feature. The current experience supports the following categories: shoes, tops, bottoms and dresses. Lingerie, swimwear and accessories are not currently supported.
Try-on is currently available within the US, the UK, Australia, Canada, India and Japan for logged-in users on both mobile and desktop versions of Google Shopping and Google Search. Users must be 18 years or older to use this feature, and accounts must have Web & App Activity and Search personalisation turned on.
Product image requirements and best practices
You’ll need to follow the general requirements for product images, as well as these requirements for try-on images in particular:
- Add high-resolution images that are at least 512 x 512 pixels, but ideally 1024 pixels or higher.
In addition, follow these best practices to make sure that you’re showing the highest quality product images:
- The image should contain only the listed garment
- The image should show the entire garment and should be a full-body picture
- Garment is on a model facing directly forward with arms down to the side
- Garments, when laid flat, should avoid excessive folds or wrinkles
- Garment details should not be covered by hands, handbags or accessories
For tops:
- Sleeves should be rolled down
- Hoodies should have the hood down
Shoppers in the US can also generate a digital version of themselves using just a selfie via the 'Try-on' overview page. This module uses the Nano Banana AI model to generate a digital version of the shopper for visualising how clothes look on them. The AI-generated image can be saved as their default image for trying on products.
When the shopper chooses to generate and use a digital version of themselves for virtual try-on, the accuracy of the final generated image is still dependent on the quality of the merchant’s product images.
AI limitations and accuracy
Generated images may include mistakes, such as body shapes, personal features or errors in clothing details. The generated image can help people find out how an item might look, but it does not indicate fit, suggest a size or indicate size availability from the merchant.
We’re committed to continuously improving this experience. If you have any comments or feedback about the generated images, you can click the 'Give feedback' link at the bottom of the Try-on screen and let us know what you think could have been better. If there are issues related to something other than the try-on image itself, you can get in touch here.
Privacy and data handling
Google does not collect or store any biometric data during this experience, nor are people’s photos used for training purposes. You will not have access to any user images or generated looks from this feature.
Frequently asked questions
How do I participate in this feature?
What if I don't want people to use this feature to try on my products?
- Sign in to your Merchant Centre account.
- From the 'Help' menu, select Help Centre and support.
- Select Contact us.
Note that opting out will only remove the 'Try-on' button from your listings but does not affect listings from other merchants selling the same product.
Can I use this feature with my Shopping ads?
How does this feature work for model try-on?
How can I ensure that people get good results?
I keep seeing mistakes in the generated image. What should I do?
Can I block try-on’s access to my domain’s images?