Tripo AI Releases EdenLive, an Open-Source Real-Time Generative Renderer
Tripo AI has launched EdenLive, a real-time generative renderer that turns rough visuals from coding agents or game engines into more detailed video. According to the company, it runs at 24 frames per second or higher on a single consumer GPU, and the model has been open-sourced.
- Tripo AI launched EdenLive, a real-time generative renderer within its Project Eden world model.
- According to the company, it renders 720p video at over 24 FPS on a single RTX 5090 consumer GPU.
- It takes a scene or reference video, a target first frame and a text prompt, and outputs 480p or 720p.
- Model weights, code, a text-to-game skill and a prompt-to-game frontend are open-sourced.
What EdenLive does
EdenLive is part of Project Eden, Tripo's world model for interactive environments. According to Tripo AI, it decouples what happens in a scene from how the scene looks. A code agent or a game or simulation engine decides how the scene evolves, while EdenLive generates the visuals in real time.
The company says the system converts rough visuals into richer, more detailed videos at 24 FPS or higher on a single consumer GPU. It also supports long-horizon synthesis, meaning it can generate video continuously in real time for several minutes at a time.
Users provide three inputs: a 2D or 3D scene or reference video, a target first frame, and a text prompt. EdenLive outputs at 480p and 720p. Experiments by Tripo AI's research team show 720p video in real time at over 24 FPS on a single RTX 5090, the company says.
Where it fits in Project Eden
Project Eden was released in May. Dr. Cao Yanpei, Chief Scientist of Tripo AI, describes it as separating the evolution of world state from visual rendering into three layers: continuously evolving world state, a state-to-visual interface, and a generative renderer. In his words, "EdenLive implements the second and third layers without restricting the first. It is a major step towards our vision of building interactive worlds."
Because the first layer is left open, the world state can come from different sources, such as a coding agent or an existing engine.
Intended uses
Tripo AI lists several applications:
- Game previews: coupling EdenLive with a coding agent for interactive game previews.
- Scene and animation previews: creators can explore different appearances while keeping the same layout, movements and camera path.
- Restyling and re-rendering: real-time replacement of characters, content or overall styles.
- Embodied AI: converting real human manipulation footage into a robot's viewpoint, which the company says bridges the gap between human demonstrations and robotic learning.
Open-source release and what it means
Tripo AI has open-sourced EdenLive. Developers get the model weights, the real-time video generation code, a skill that turns text prompts into simple playable games, and a frontend for the complete prompt-to-game pipeline.
For developers and creators, the release means the renderer can be tested on consumer hardware rather than only through a hosted service, at least according to the performance figures Tripo AI reports. The numbers come from the company's own experiments, and the announcement does not cite independent testing.
Tripo AI was founded in 2023 and develops 3D-native foundation models and world models. It also offers Tripo Studio and Tripo API. EdenLive extends that work toward interactive, real-time environments.
