Google DeepMind introduces major updates to its video generation platform Flow, including the new Veo 3.1 model and expanded creative controls. Jess Gallegos and Thomas Iljic write in the official Google Blog.
Flow users have generated over 275 million videos since the platform launched five months ago. The latest updates address user requests for more artistic control and comprehensive audio support across all features.
According to the post, Veo 3.1 delivers richer audio, improved narrative control, and enhanced realism with lifelike textures. The model supposedly shows stronger prompt adherence and better audiovisual quality when converting images to videos compared to its predecessor.
Several existing features now include generated audio. “Ingredients to Video” allows users to combine multiple reference images to control characters, objects, and style. “Frames to Video” generates seamless transitions between a starting and ending image. The “Extend” feature creates longer videos, lasting up to a minute or more, by continuing action from the final second of an original clip.
New editing capabilities give users more precision during the creative process. The “Insert” tool adds new elements to scenes, with Flow automatically handling complex details like shadows and lighting. An upcoming removal feature will allow users to delete unwanted objects or characters, with Flow reconstructing the background seamlessly.
The Veo 3.1 model is available through the Gemini API for developers, Vertex AI for enterprise customers, and the Gemini app.
