Google announced early access for the Ask Photos feature and updated its virtual try-on tool

Along with updates to the traditional search feature for Google Photos which enables users to perform searches using conversational language for improved results and to sort results by date or relevance, Google has announced an early access preview in Google Labs of the new Gemini-powered Ask Photos feature for users based in the US. With Ask Photos, users can ask Gemini questions that leverage the assistant's understanding of their photo galleries to uncover details about memories and other information locked in users photo galleries.

For instance, users can ask for details of a trip, such as the places visited or the foods tried. In addition to unveiling details about a picture's contents, Gemini can assist with other photo-related tasks, such as selecting the best photos from an occasion for a shared album or summarizing the most memorable events from a recent vacation. Early access to the feature has already been rolled out for select US users.

Following the success of Google's virtual try-on tool, the company recently added dresses to the variety of garments compatible with the AI-powered shopping feature. A recent blog post details the specific challenges related with teaching the technology to generate high-quality, detailed, and realistic images of users wearing the dresses they are interested in buying. The complexity of dresses, the variety of styles available, and the fact that most dresses cover more skin than tops and blouses make it challenging to obtain pictures that offer a realistic level of detail without blurring out or deleting the features of the person supposed to be wearing them. To solve this issue, Google developed a new technique, called VTO-UNet Diffusion Transformer (VTO-UDiT for short). The updated feature will enable users to virtually try on dresses from hundreds of labels across the internet.