World Labs, the AI startup founded by renowned AI expert Fei-Fei Li, has launched its first groundbreaking project: an AI system capable of generating interactive 3D scenes from a single photo.
Unlike other AI tools that create static 3D models or environments, World Labs’ system creates explorable, dynamic 3D worlds that users can navigate in real time.
Available through a demo on World Labs’ website, the AI-generated scenes are rendered live in your browser and feature a controllable camera with an adjustable depth of field (DoF).
The effect blurs background objects, creating a more realistic sense of depth. Though the scenes appear a bit cartoonish, they are consistent and obey basic physics laws, offering a solid sense of space and dimension.
The system is part of an emerging field known as “world models,” which aims to create fully immersive, simulated environments. While other systems have struggled with consistency and artifacting—such as poor resolution or glitches in level designs—World Labs has ensured that their scenes remain stable and believable.
The AI can even apply interactive effects, such as changing the color of objects and dynamically lighting backdrops.
However, there are still limitations. The scenes are currently restricted to a small area, and users cannot explore beyond a set boundary. Additionally, occasional rendering errors can cause objects to blend unnaturally. But World Labs views this as an early preview and is actively working on improving the size and fidelity of the worlds.
With $230 million in venture capital backing from investors like Andreessen Horowitz (a16z) and Intel Capital, World Labs aims to refine this technology and release its first product in 2025.
The company is focusing on developing tools for professionals in fields like game development, filmmaking, and engineering. The ultimate goal is to democratize the creation of interactive, vibrant 3D worlds, drastically reducing the time and cost involved in creating such environments.
