Google virtual try-on tool lets you select dresses which suit you
Technology
It preserves the wearer's features while seamlessly integrating the virtual dress.
(Web Desk) - Google announced on Thursday that its generative AI-powered virtual try-on tool now supports dresses, offering users the ability to virtually try on thousands of styles from numerous brands, such as Boden, Maje, Sandro, Simkhai, and Staud.
Dresses have emerged as one of the most popular categories among users of the tool. However, Google highlighted in a blog post that incorporating dresses into the existing diffusion technique has proven challenging due to their intricate details and complex designs.
The virtual try-on tool, which launched last year, originally used Google's diffusion technology to create realistic images of tops and blouses, simulating how these garments would drape and fold on a person.
The technology struggled with dresses, particularly in accurately representing detailed patterns like floral or geometric designs. The previous model could handle low-resolution images but required an updated approach to capture the finer details of dresses.
To overcome these challenges, Google developed a new training strategy that begins with lower-resolution images and progressively incorporates higher resolutions to maintain detail.
As dresses cover most of the body and come in various lengths—such as midi, maxi, and mini—there was a risk of obscuring or blurring body features.
To address this, Google introduced a new method known as the VTO-UNet Diffusion Transformer (VTO-UDiT), which preserves the wearer's features while seamlessly integrating the virtual dress.
Virtual try-on technology aims to improve the shopping experience by allowing customers to better gauge how clothing will fit, regardless of body type.
With competitors like Adobe, Amazon, and Walmart also offering virtual try-on solutions, Google’s expanded feature seeks to provide a more sophisticated and accurate experience.
