Google Presents ‘Veo’, Its Most Advanced AI Video Generator For Filmmakers
Video screenshot via Google DeepMind
At Google I/O 2024, the tech giant revealed ‘Veo’, a generative AI model with high-definition video-making talents promising immediacy for content creators with big dreams.
The most advanced video generation model Google has rolled out to date, Veo can produce high-quality, 1080p videos lasting over a minute long. Its control extends beyond basic image generation. Veo is fluent in the language of cinema, understanding terms like “timelapse” or “aerial shots of a landscape,” and simulating real-world physics.
With this knowledge, the tool crafts cohesive footage where people, animals, and objects move realistically throughout the scene, envisioned to open doors for filmmakers and content creators with an efficient way to render their ideas without back-breaking work.
Veo’s development is the culmination of years of research in generative video models by Google AI, building upon projects like Generative Query Network (GQN), DVD-GAN, Imagen-Video, Phenaki, WALT, VideoPoet, and Lumiere. By combining advancements in architecture, scaling techniques, and other cutting-edge methods, Veo produces higher-quality videos with exceptional output resolution.
Examples of Veo, Google's generative video platform #GoogleIO2024 pic.twitter.com/7xVVuwggGx
— Julian Chokkattu (@JulianChokkattu) May 14, 2024
Unveiled alongside Veo were Imagen 3, Google’s latest text-to-image model, and a prototype called ‘AI Teammate’, a virtual coworker designed to integrate into collaborative work environments.
Adding fuel to the fire of excitement surrounding Veo was the revelation that Donald Glover, the multi-talented actor, musician, and director, has been handed the reins to play with Veo’s features and produce a short film.
An aspect that particularly resonated with Glover and his team was its ability to facilitate “making mistakes faster.” It’s a crucial part of the creative process, where rapid iteration and refinement of ideas are invaluable.
Although Veo is still in development, Google AI’s DeepMind lab is pushing the boundaries further, exploring features like storyboarding and extending scene lengths.
The tool is now available to select creators for private testing within VideoFX through a waitlist. Additionally, Google plans to integrate some of Veo’s functionalities into YouTube Shorts and other products in the future.
[via TechCrunch and ZDNET, images via various sources]
