OdysseyThe startup founded by Oliver Cameron and Jeff Hawke, founded by the Pioneers of self -propellers, has developed the AI model that allows users to “interact” with streaming video.
Available on the Internet in an “early demo”, the model generates and sends a video frame every 40 milliseconds. Through the basic control elements, viewers can explore the areas in the film, as is the rendered 3D.
“Given the current state of the world, the upcoming action and history of states and activities, the model tries to predict the next state of the world” explains Odyssey in the blog post. “This power supply is a new world model, showing possibilities such as generating pixels that seem realistic, maintaining spatial consistency, learning to video actions and displaying coherent video streams for 5 minutes or longer.”
Many startups and large technology companies are chasing global models, including Deepmind, influential researcher AI Fei-Fei Li Labs, Microsoft and Decart. They believe that world models can one day be used to create interactive media, such as games and films, and conducting realistic simulations, such as training environments for robots.
But creations have mixed technology feelings. AND last Wired Investigation showed that games such as Activision Blizzard, which have fired dozens of employees, use artificial intelligence to cut corners and combat. And 2024 test Associated by the animation guild, a relationship representing Hollywood animators and cartoonists, estimated that over 100,000 American films, television and animation would be disturbed by AI in the coming months.
For his part, Odyssey undertakes to cooperate with creative professionals – not replace them.
“Interactive video … opens the door to completely new forms of entertainment, in which stories can be generated and tested on demand, free of restrictions and costs of traditional production,” writes the company in its blog post. “Over time, we believe that everything that is video today – entertainment, advertising, education, training, travel and many more – will evolve in interactive video, all powered by Odyssey.”
Odyssey's demo is a bit rough around the edges that the company recognizes in its position. The environments generated by the model are blurred and distorted, and unstable in the sense that their systems do not always remain the same. Go in one direction for a moment or turn around, and the environment may suddenly look different.
But the company promises to quickly improve the model, which can currently be streaming video at speeds of up to 30 frames per second from the NVIDIA H100 GPU clusters, at a cost of 1 to 2 USD for “user time”.
“Looking to the future, we examine richer teams of the world, which much more faithfully capture dynamics, while increasing the time stability and persistent state,” writes Odyssey in his position. “At the same time, we expand the space of action from movement to interaction in the world, learning open activities from large -scale video.”
Odyssey adopts a different approach than many AI laboratories in the world modeling space. He designed a 360-degree camera system mounted on a backpack to capture real landscapes, which according to Odyssey can serve as a basis for higher quality models than models trained only on publicly available data.
Until now, Odyssey has collected $ 27 million from investors, including EQT Ventures, GV and Air Street Capital. Ed Catmull, one of the co -founders of Pixar and former president of Walt Disney Animation Studios, is a member of the startup board.
In December last year, Odyssey said he was working on software that allows creators to charge scenes generated by his models on tools such as Unreal Engine, Blender and Adobe After Effects so that they can be manually edited.
