.webp)
We Love Storytelling
Since ancient times, humans have been storytellers. Our earliest ancestors sat around campfires to tell stories. We created epics like the Iliad, the Odyssey, the Vedas, the Analects, and One Thousand and One Nights. We broadcast pivotal moments to millions on television. We've made masterpieces that have shaped us, and pioneered visual effects that transport us to distant lands. We've created cinematic masterpieces such as Casablanca, 2001: A Space Odyssey, The Godfather, Schindler's List, Jurassic Park, Toy Story, Interstellar, and Dunkirk.
As technology evolves, so do our stories. However, one thing remains constant - great stories are the product of blood, sweat, and tears.
We Need to Decide How Artificial Intelligence Tells Stories
We need to hold artificial intelligence to a higher standard. A brief glance around the web will reveal that we are flooded with low-quality AI-generated content. Content farms, spam bots, and even well-meaning companies are using artificial intelligence to mass-produce text and images, aiming to gamify algorithms and capture your attention. If not handled properly, AI video generation could go in a similar direction, and we could be overwhelmed by random videos without spark or story. Over a long enough timescale, perhaps we will become so desensitized to these junk videos that we forget what high-quality human stories look like. Perhaps humans will only become viewers of stories, rather than storytellers.
At Odyssey, we refuse to accept such a future. Storytelling by humans is too important to our way of life, and professional storytellers have proven they have much to offer. Likewise, if we build powerful artificial intelligence correctly, it can also offer much. We believe that instead of replacing humans with algorithms optimized for clicks, we should put new visual artificial intelligence in the hands of professional storytellers. This visual artificial intelligence will not only allow them to create stunning videos but also to precisely direct them and tell the epic stories they have in mind.
Breakthroughs Needed in Visual Artificial Intelligence
To achieve this goal, we are pioneering Hollywood-level visual AI that can generate and control beautiful landscapes, characters, lighting, and motion. This technology will bring game-changing AI efficiency and creativity, but more importantly, it allows you full, fine-grained control over every element in the scene. From magnificent mountains, lush vegetation, beautiful rivers, epic characters, all the way down to low-level materials, lighting, motion, and more. In short, this is a new way to make movies, TV shows, and video games.
How Our Experience Works
This feature is achieved at a deeper level than traditional text-to-vision models. Instead of training a model that restricts you to a single input and a single uneditable output, we train four powerful generative models that can fine-tune control over each major aspect of visual storytelling. Specifically, these models can generate high-quality geometry, realistic materials, stunning lighting, and controllable motion. Combined, these models allow you to quickly generate scenes and shots that perfectly match your vision. Furthermore, we are building workflows around these models for advanced users and integrating them seamlessly with existing production workflows used in Hollywood, gaming, and other fields.