WorldGen is an AI model and library that can generate full 3D scenes in a matter of seconds from either text prompts or reference images. It is designed to create interactive environments suitable for games, simulations, robotics research, and virtual reality, rather than just static 3D assets. The core idea is that you describe a world in natural language and WorldGen produces a navigable 3D scene that you can freely explore in 360 degrees, with loop closure so that the space remains consistent as you move around. It supports a wide variety of scenes, including both indoor and outdoor settings, and can handle realistic as well as stylized or fantastical environments. Rendering is decoupled from generation, so you can render at arbitrary resolutions and camera trajectories in real time, which makes it easier to integrate into custom pipelines.
Features
- Text and image based 3D scene generation in seconds
- 360° free exploration with loop closure for spatially consistent navigation
- Support for diverse indoor, outdoor, realistic and stylized environments
- Flexible rendering at arbitrary resolution and camera paths in real time
- Simple Python API with a minimal demo script for quick prototyping
- Open source project under the Apache 2.0 license for research and integration