World Labs launches an AI system that generates a 3D world from a single image, supporting real-time interaction

PANews December 3rd news, World Labs, a space intelligence company founded by Stanford University professor Fei-Fei Li, announced the development of a trapAI system, which can generate a complete 3D world from a single input image and solve the problem of lack of controllability and consistency in existing generative AI models. The system can not only estimate 3D geometric shapes, but also fill in unseen parts in the scene, generate new content, allowing users to freely explore 3D scenes in a game-like manner. The generated 3D scenes support real-time rendering in the browser, allowing the simulation of 3D camera effects (such as shallow depth of field and push-pull focus). Compared with pixel prediction generation models, the system follows 3D geometric rules, and the scenes are consistent and support dynamic interaction, such as modifying light, appearance, and inserting new objects.