Google's latest virtual try-on tool now allows consumers to preview the stunning effects of wearing their favorite dresses.
On September 5th, the tech giant announced a significant expansion, extending its AI-based virtual try-on (VTO) service to the clothing category, which has long been a popular choice for consumers.
Google proudly announced in its official blog, "This feature upgrade is made possible by our generative AI technology developed specifically for VTO, particularly utilizing cutting-edge diffusion model technology. This technology can finely construct, pixel by pixel, high-definition realistic images of models wearing tops, shirts, and even dresses."
However, Google's team encountered significant challenges in exploring the virtual try-on of dresses. Dress designs are complex and varied, placing higher demands on image generation technology, as if painting delicate pictures within a small space.
Google explained, "Our VTO AI model performs excellently on low-resolution images, but when it comes to dresses, this method often struggles to capture key details. Simply increasing the resolution does not solve the problem."
To address this, Google's research team innovatively adopted a "progressive training strategy," starting from low-resolution images and gradually increasing to higher resolutions to ensure that every detail is accurately presented.
In addition, dresses cover more areas of the body, and achieving "changing clothes" without blurring the character's features or obscuring important details is also a major challenge. Google successfully achieved the precise preservation of character features and the natural integration of dresses through its self-developed VTO-UNet Diffusion Transformer technology.
The release of this news comes at a time when many e-commerce giants are intensifying their virtual try-on projects. For example, Walmart emphasized consumers' strong interest in technologies such as virtual try-on in its July report on adaptive retailing, hoping to enhance the shopping experience and preview the real effects of products.
As industry observations have pointed out, virtual try-on technology can effectively reduce return rates, helping consumers purchase their desired items on the first try and reducing the after-sales burden for retailers.
"Virtual try-on allows shoppers to ensure they choose the most suitable and favorite products, significantly reducing return situations," said Liu Xiaoyan, President and Chief Growth Officer of Perfect Company, a leading company in the field of AI and augmented reality (AR) beauty technology, in a previous interview. With the continued investment and innovation from tech giants like Google, the future of virtual try-on will undoubtedly be even more exciting.
