Google Veo 3 incorporates sound with video

Currently only available in the US, Veo 3 can generate anything from background traffic noises to birds singing in a park, or even dialogue between characters. The audio feature differentiates it from many other video-only genAI tools like Sora.

Google has also added new capabilities to its Veo 2 model, including a reference tool that allows users to feed the tool images of characters, scenes, objects and styles for better creative control and consistency.

Camera controls can also help Veo 2 users to incorporate camera movements such as rotations, dollies and zooms.

An outpainting function allows users to broaden their frame, turning video from portrait to landscape to fit any screen size. An object add and remove tool lets users add or erase objects from videos.

Meanwhile, Flow is a new AI filmmaking tool launched for Google’s Veo, Imagen and Gemini models. Using prompts, budding filmmakers are able to “create cinematic clips and scenes for their stories.”

Google also announced a new Imagen 4 model for image creation. Google talked up Imagen 4’s clarity in fine details like fabrics, water droplets, and animal fur. It can also create images in a range of aspect ratios and up to 2k resolution.

The models were announced by Eli Collins, VP, Google DeepMind, in a blog post. Collins said: “We’ve partnered closely with the creative industries — filmmakers, musicians, artists, YouTube creators — to help shape these models and products responsibly and to give creators new tools to realise the possibilities of AI in their art.”