Alphabet (GOOGL, Financial), Google's parent company, has announced the rollout of a new "photo-to-video" feature for its paid users. Initially tested on a limited scale earlier this year, this AI tool is now integrated into the Gemini AI assistant. Subscribers to Google AI Ultra and Pro plans in select regions can access this feature via the Gemini web platform, with mobile app updates expected soon.
The new feature allows users to create 8-second videos with sound from a single photo and text description. The videos are generated in MP4 format with 720p resolution and a 16:9 aspect ratio. This integration brings Google in line with competitors like OpenAI and Runway AI Inc. in the AI video sector, while also competing with Chinese firms such as Alibaba, Manus, and Kuaishou Technology, which have recently launched upgraded video tools.
The tool is powered by Google's latest video generation model, Veo 3, unveiled at their developer conference in May. Previously, it was only available through the standalone paid tool, Flow. Google emphasizes that it has implemented significant measures to ensure the generated videos adhere to guidelines, including prohibitions on using public figures' images and content that incites harmful behavior.
Despite these advancements, tests reveal some technical limitations. Users found that the tool sometimes alters facial features or even ethnicities when generating videos from personal photos. While it performs well with simple commands, such as animating plants or static images, it struggles with complex tasks like making a photo subject breakdance, often resulting in simpler animations like waving.
A Google spokesperson acknowledged these issues, explaining that the AI model lacks directives to alter appearances. The photo-to-video and facial animation features are still emerging technologies, and discrepancies between the generated and original content may occur. The company plans to continue refining these capabilities, particularly facial animations, in future updates.
