Google’s latest iteration of its AI model, Veo 3, can now generate eight-second video clips with AI-generated audio from a single still image. The feature is available as a “preview offering” for all Google Cloud customers and partners.
The new capability allows influencers, brands, and creative professionals to create short videos without needing to shoot on-site footage. For example, an influencer could upload a headshot and prompt Veo 3 to generate a video of her walking down a runway, complete with ambient noise and AI-generated audio. Brands can also use the feature to showcase products from different angles.
Veo 3’s image-to-video capability is expected to save time and resources for creative professionals, while providing new materials for social media and other channels. The model has already attracted attention for its ability to seamlessly integrate AI-generated video and audio, simulating real-world physics without technical glitches.
However, the feature also raises concerns about the potential spread of online misinformation and manipulation on social media. Google DeepMind CEO Demis Hassabis hinted at possible future use in generating virtual worlds for video games, sparking debate. The company’s training data sourcing has also raised questions, with some citing YouTube videos as a source.
To try Veo 3, customers can access the Vertex AI Media Studio across 159 countries, following its public release last month. Moonvalley’s Marey is an alternative AI-generated video tool trained on licensed data for those seeking more airtight options.
Source: https://www.zdnet.com/article/googles-veo-3-can-now-create-an-8-second-video-from-a-single-image-how-to-try-it