Two veterans of self-driving car companies are trying to break into Hollywood, not as actors, but as designers of the AI software that will power the blockbuster movies of the future.
They launched a San Francisco-based startup called Odyssey that competes with OpenAI, Runway, Kuaishou, and Metaphysic to provide tools for professional filmmakers and animators to create professional-quality films using generative AI.
Today, Odyssey emerged from nearly a year of operating in “stealth mode,” with $9 million in seed funding from GV (Google Ventures, the venture capital arm of Alphabet), with participation from DCVC, Air Street Capital, and more than a dozen early-stage individual investors and angel investors.
The company’s AI software is in some ways similar to OpenAI’s Sora or Runway’s Gen 3, which turn a text description of a scene or sequence of shots into a short, high-quality, cinematic video.
Sora, which has only been rolled out to a handful of users so far but which OpenAI revealed in a splashy demo in February, has put Hollywood on edge. Film producer Tyler Perry said after seeing Sora’s demo that he was indefinitely shelving plans for an $800 million expansion of an Atlanta movie studio because Sora made him doubt whether traditional studios had a place in the future.
Actor and tech investor Ashton Kutcher has also said he believes AI will soon allow anyone to create a movie, without the need for actors, film crews or big Hollywood budgets. Last month, another text-to-video model called Kling, from Chinese AI company Kuaishou, also impressed AI developers.
Fine grain control
But so far, these models only generate short video sequences of less than a minute and can’t guarantee that characters and backgrounds will be consistently represented throughout the sequence. According to Oliver Cameron, Odyssey’s CEO and one of its two co-founders, Odyssey wants to give filmmakers much more precise control over the shots generated by the AI model and ensure that consistency, as well as more realistic lighting and visual effects, over much longer sequences of images than other AI models allow.
Odyssey is designed as a “Hollywood-quality” visual AI, intended to empower “the world’s most maniacal creators,” Cameron said. FortuneHe cited the example of director James Cameron and his Avatar films, which took more than a decade to produce, involved hundreds of special effects artists and cost hundreds of millions of dollars. Odyssey wants to reduce that time and cost — down to a five-person crew spending $50,000 and working for six months — without sacrificing visual quality.
To realize Odyssey’s vision of “Hollywood-quality” visual effects, Cameron says the software Odyssey has created goes beyond a simple text-to-video model. Instead, it’s a family of models that specialize in four distinct tasks: generating three-dimensional graphics renderings; generating material effects, such as water or fabric; generating motion; and finally, generating lighting effects. These models work together to produce the frames in the video. What’s more, the input to each of these models goes beyond text, he says. It can be a human creator drawing with a stylus or still digital images.
This will allow filmmakers and professional visual effects artists, who are the intended users of Odyssey’s software, to have complete creative control over the final result, something not possible with current text-to-video AI models.
Self-driving car veterans
Cameron and co-founder Jeff Hawke, who is Odyssey’s CTO, both come from the self-driving car world. Cameron co-founded self-driving car company Voyage, which was acquired by Cruise in 2021 for an undisclosed sum. He then served as Cruise’s vice president of product for two years. Hawke, meanwhile, was part of the founding research and engineering team at Wayve, a British AI company that builds the software brains of self-driving cars and recently attracted a billion-dollar investment from SoftBank, Nvidia and Microsoft.
While the leap from self-driving cars to Hollywood movies may not seem like a straightforward one, Cameron explains that the software for self-driving cars relies heavily on artificial intelligence learning to create a three-dimensional world from two-dimensional input. The problem with visual effects is essentially the same: take a three-dimensional world and condense it into two-dimensional moving images.
At Wayve, Hawke worked on an AI model, called FIERY, that preceded another model called GAIA and could take an initial video frame and then use it to create a realistic, full-frame video of a street-level world, which the model generates on the fly. Text prompts can also be used to shape what is represented in this simulated world. Wayve uses this technology to create synthetic data to train its autonomous decision-making AI software. But Hawke thought similar ideas could be used to create high-quality footage for movies.
Cameron says another benefit of getting out of the self-driving car business is that the two co-founders understand the importance of gathering their own real-world data sets to train their models.
Unlike other text-to-video companies that have relied on the internet to mine video data to power their AI software, Odyssey collects its own real-world, three-dimensional datasets and uses them to power its models. Cameron says there simply aren’t enough three-dimensional datasets available on the internet to build AI models of the size and capabilities that Odyssey wants to create. And his and Hawke’s experience with self-driving cars has given them some clever ideas about how to collect the real-world data Odyssey needs, though Cameron declines to say on the record what some of those methods are.
Odyssey plans to use the money from its seed funding round to expand its current staff of 13 full-time employees and also to build larger AI models, an expensive proposition that requires the company to have access to large clusters of graphics processing units, the type of computer chips most often used in AI applications.
Correction, July 8: An earlier version of this article incorrectly stated that Jeff Hawke worked on Wayve’s GAIA AI model. He worked on a predecessor to GAIA called FIERY.