Google is making its video generation model, Veo, more accessible to its customers. The model, unveiled in April, can create 1080p clips from images and prompts. Veo understands visual and cinematic styles, including shots of landscapes and time lapses, and makes edits to generated footage.
The private preview launch will enable Quora to integrate Veo into its Poe chatbot platform and Mondelez International to create marketing content with agency partners. To achieve this, Google has augmented the model for enterprise customers on Vertex AI, improving its ability to generate high-definition videos in 720p.
Veo is competitive with leading video-generating models but still faces limitations such as inconsistent object behavior. It was trained on various footage and uses prompt-level filters for violent and explicit content. To mitigate deepfake risks, Google embeds invisible markers into generated frames using SynthID technology.
While Veo gains traction, its applications risk alienating creatives due to job disruptions estimated at over 100,000 U.S.-based jobs by 2026. Google is taking a cautious approach, releasing products in preview first to gather feedback from enterprise customers before wider availability. The flagship image generator Imagen 3 is also now available for all Vertex AI customers without a waitlist.
Source: https://techcrunch.com/2024/12/03/googles-video-generator-comes-to-more-customers