Odyssey’s new AI model streams 3D interactive worlds

Odyssey's AI streams realistic 3D worlds, revolutionizing video interaction.

: Odyssey, led by Oliver Cameron and Jeff Hawke, has unveiled an AI model allowing users to interact with streaming video in real-time. The model generates video frames every 40 milliseconds, providing an immersive, game-like experience without a traditional game engine. Despite initial limitations like blurry and unstable environments, Odyssey promises improvements and has already secured $27 million in funding. The startup is focusing on partnerships with creatives and innovations in video technology and cloud-based interactive media.

Odyssey, a startup co-founded by self-driving pioneers Oliver Cameron and Jeff Hawke, has developed an AI model that allows users to interact with streaming video in real-time. This novel technology streams video frames every 40 milliseconds, offering users the ability to explore video areas akin to a 3D-rendered video game environment. The AI model tries to predict the next state of the world, ensures spatial consistency, and can output videos that maintain coherence for over five minutes. This technology is available in an early web demo, though its capabilities are currently rough around the edges as Odyssey works to refine it further.

Big names in the AI industry, such as DeepMind, World Labs of Fei-Fei Li, Microsoft, and Decart, are also exploring world models with the vision of creating interactive media, such as video games and movies, and realistic simulations for training environments. Despite the potential these technological advances hold, the creative industry shows some apprehension. Reports have highlighted that gaming giants like Activision Blizzard are already using AI to reduce labor costs, which has reportedly led to a significant reduction in workforce. A study commissioned by the Animation Guild indicates that AI may soon disrupt upwards of 100,000 jobs in the film, television, and animation sectors in the U.S., including pioneering efforts aiming to support, not replace creative professionals.

Oliver Cameron recently demonstrated one of the scenarios generated by Odyssey’s AI model, which depicted a piece that was, at once, serene and chaotic. While the current demo shows that the generated environments are blurry, distorted, and sometimes fail to maintain layout stability, Odyssey has promised rapid improvement, streaming video from clusters of Nvidia H100 GPUs at the cost of $1 to $2 per user-hour. They plan to expand the “action space” from basic motion to more complex interactions, and refine dynamic representation using a 360-degree, backpack-mounted camera system for capturing real-world landscapes.

The venture has already raised $27 million from prominent investors, including EQT Ventures, GV, and Air Street Capital, and boasts luminary Ed Catmull, co-founder of Pixar, on its board. Odyssey's AI endeavors seek to broaden the current capabilities of video creation, potentially rendering traditional production models obsolete. The aim is to open new avenues in entertainment and other industries, moving all current video applications, such as ads, education, and training, towards interactive media powered by their technology.

A recent blog post by Odyssey emphasized their focus on collaboration with the creative sector, expressing that interactive video could generate and explore stories on-demand, devoid of traditional production constraints. Looking ahead, Odyssey’s research aims to enrich world representations, improve temporal stability and ensure persistent states, enhancing the efficiency and appeal of their AI-generated interactive content.

Sources: TechCrunch, Wired, Animation Guild, Odyssey Blog

AIstartupsGenerative AI