Odyssey builds ‘Hollywood-quality’ AI model to compete with Sora
Odyssey, an artificial intelligence (AI) startup founded last year, on Monday shared details about its first AI product. The company revealed that it is building an AI video model that can create Hollywood-quality visual effects, similar to OpenAI’s yet-to-be-released Sora tool. Odyssey’s co-founder says that the AI model will allow users to edit and control the output at a granular level, adding that the company works with multiple large language models (LLMs) to generate different layers of the output video, which can be controlled separately.
How Odyssey’s AI Visual Model Works
In a series of posts on X (formerly Twitter), Oliver Cameron, CEO and co-founder of Odyssey, said that the AI startup had raised $9 million (approximately Rs 75.1 crore) in seed funding led by Google Ventures and that the startup was currently working on building a tool that would deliver high-quality videos that could be customized and edited by users.
Cameron also shared details about Odyssey’s AI technology, claiming that it is designed to generate “Hollywood-grade” video. The exec also said that the startup trained four generative models for users to “take full control of the core layers of visual storytelling.”
Each model individually allows you to precisely configure the details of your scene.
Combined, these models generate video or scenes exactly the way you want them.
Additionally, the outputs from our models can be integrated into existing pipelines used in Hollywood and beyond. photo.twitter.com/jHZoevLV9n
—Oliver Cameron (@olivercameron) July 8, 2024
Cameron highlighted the problem with existing AI text-to-video models, namely the lack of tools or options to control or edit the output. “As a storyteller, you have little ability to control your environment or your characters, or to iterate over the finer details of your shot until it’s just right. More powerful models are needed,” he added.
To solve the problem, the company uses multiple AI models that generate a single layer of the composite video. According to Cameron, four models independently generate geometry, materials, lighting, and motion. These four layers are generated simultaneously from a single text prompt and then combined to present the final video.
The company claims that users will have the option to configure each layer separately for more control over the output. Odyssey will also integrate its video outputs into existing Hollywood tools and systems used to generate visual effects.
Currently, the AI video model is in the early stages of development. There is no launch date for the AI model yet. However, the company has stressed that it will continue to share regular updates on its progress. Notably, Cameron previously worked for Cruise and Voyage, two startups working on self-driving vehicles.
Jeff Hawke, the company’s other co-founder and CTO, previously served as Vice President of Technology at Wayve, an AI company developing autonomous driving systems.