World Labs Launches AI System Capable of Generating 3D Interactive Scenes from Single Images

Founded by AI expert Fei-Fei Li, the startup World Labs has recently introduced its first project: an artificial intelligence system capable of generating video game-like 3D environments from a single image. This technology allows users to navigate these 3D settings using a keyboard and mouse, extending beyond the original input image with all additional content created by AI.

Unlike other technologies that convert photos into 3D models, World Labs' 3D environments are interactive and customizable. The scenes are rendered in real-time within a browser, featuring controllable cameras and adjustable depth of field effects, which blur background objects as depth increases. Although the scenes have a slightly cartoonish style, they can already be experienced through demonstrations on the company's website.

As part of the emerging category of "world modeling" in artificial intelligence, World Labs' approach ensures that once a scene is generated, it remains consistent and adheres to fundamental physical laws, providing the scene with a sense of solidity and depth. Additionally, the system can add interactive effects and animations to the scenes, such as changing object colors or dynamically adjusting backlighting.

However, the current version has some limitations, including a restricted exploration range and occasional rendering errors. World Labs has stated that this is merely an early preview, and the team is actively working to enhance the scale and fidelity of the generated worlds while experimenting with new user interaction methods.

Launched earlier this year, World Labs has raised $230 million in venture capital from investors including Andreessen Horowitz, Ashton Kutcher, Intel Capital, AMD Ventures, and Eric Schmidt, achieving a valuation of over $1 billion. The company is expected to launch its first product in 2025.

In addition to creating interactive environments, World Labs plans to develop professional tools for artists, designers, developers, filmmakers, and engineers, targeting a wide range of clients from video game developers to film studios. Currently, building virtual interactive worlds requires significant funding and development time, but world modeling is expected to substantially lower these barriers.