A startup is training an AI system that it claims will allow developers to create cinematic worlds with complete control over scenery, characters, lighting and movement. How? By strapping cameras to people's backs and wandering world wide.
Odysseyfounded by self-driving pioneers Oliver Cameron and Jeff Hawke (Cameron was previously product VP at Cruise), says it has developed an “advanced camera capture system” that may collect data from almost anywhere an individual can go. The system weighs about 25 kilos and has six cameras, two lidar sensors and a Inertial measurement unit.
Has a similarity to Google's Street View TrekkerThe system can capture its surroundings in “3.5K resolution” and 360 degrees, with “physically accurate” depth information metadata attached.
So what's the purpose? Well, Odyssey says it takes data from the system and runs it through algorithms to “capture the high-quality details that make up our world.” Essentially, the corporate creates digital reconstructions of real-world scenes within the form of Meta's Hyperscape project – scenes with forests, caves, trails, beaches, glaciers, parks, buildings, and so forth.
Now, it's not entirely clear how these reconstructions will lead to raised generative tools for creatives. Cameron and Hawke have done it before said that Odyssey has developed multiple generative AI models that create layers of visual details, including object geometry, lighting and movement, after which mix them right into a single virtual “world” to create desired scenes.
However, even today's best “world models” have limitations – and Odyssey doesn’t claim to have solved all of them. Still, it ensures liquidity to maneuver forward.
Odyssey today announced that it has raised $18 million in a Series A funding round led by EQT Ventures with participation from GV and Air Street Capital. The latest money, which brings the corporate's total revenue to $27 million, might be used to expand Odyssey's data collection operations in California.
Odyssey plans to expand its data collection to other states and countries in the long run – hopefully with privacy protections. (Google's Street View team, for instance, has been in a position to do that Crosshair by regulatory authorities for taking images of public places that violated the privacy of bystanders.)
“We imagine it would be unimaginable for generative models to provide Hollywood-level worlds that feel alive without training on a large amount of wealthy, multimodal real-world 3D data,” the corporate wrote in a post on his website Blog. “We imagine a sophisticated generative world-building model will enable a greater technique to create movies, games and more.”
