Odyssey-2 transforms AI video into instantaneous communication

Odyssey has introduced Odyssey-2, a new interactive video model that creates streaming AI footage at 20 frames per second. As viewers explore the setting, they may manipulate and modify multi-minute recordings using text prompts. The specifics: Odyssey-2 broadcasts footage instantly, with fresh frames appearing every 50 milliseconds, in contrast to conventional video models that require minutes to create brief snippets. Without any prior planning, the system creates video segments; each new frame is generated based on past events and real-time user prompts. Through a chat box, users may guide the video's development with natural language suggestions while it plays, and the AI continuously adjusts to each input. The model can replicate realistic behaviors like waves moving over water or light shifting on surfaces because it learns physics and dynamics from video data. Compared to other AI video platforms, Odyssey-2 seems very distinct and is more akin to a hybrid world generator than Veo or Sora. The real-time, open-ended exploration seems like a notion with a ton of promise for new content experiences, even though the quality might not be as remarkable as other models.