Explore Odyssey's AI-Powered Interactive Video Worlds

Imagine stepping into a world that feels like a video game yet looks like a blurry, dreamlike version of reality. This is the promise of Odyssey, an AI startup backed by Pixar cofounder Edwin Catmull, which is pioneering what it calls "interactive video." Unlike traditional games built with polygons, Odyssey’s environments are generated entirely by AI in real-time, creating a unique blend of video and interactivity.

Currently available as a research preview, Odyssey lets users explore several AI-generated worlds—ranging from wooded cabins to shopping malls—using familiar WASD controls. Each session is different because the system regenerates visuals dynamically, resulting in a fuzzy, glitchy experience that feels like wandering through a surreal version of Google Street View.

Powered by clusters of H100 GPUs across the US and Europe, Odyssey’s AI model predicts and streams the next video frame in as little as 40 milliseconds. Despite this impressive speed, the experience is still rough around the edges: collision detection is inconsistent, environments morph unpredictably, and the visuals remain noisy and blurry.

Edwin Catmull acknowledges these imperfections but emphasizes that Odyssey is at the forefront of a rapidly evolving field. The startup is actively integrating advances from the AI community, aiming to refine image quality and stability by applying neural network filters to reduce noise and enhance textures.

While Odyssey’s interactive video isn’t ready to replace traditional video games or films, it opens a fascinating window into the future of AI-generated media. The concept hints at a future where immersive, dynamic worlds can be created on the fly, blending the boundaries between watching and playing.

For developers and creators, Odyssey’s approach signals new opportunities to innovate in entertainment, simulation, and interactive storytelling. As AI models grow more sophisticated, the dream of a Holodeck-like experience—where virtual environments respond fluidly to user input—edges closer to reality.

In the meantime, Odyssey’s interactive video remains a raw but undeniably exciting glimpse into how AI can transform media consumption. It challenges us to rethink what it means to engage with digital content and invites us to imagine a future where the line between viewer and participant blurs.