World Labs, the AI startup co-founded by Stanford AI pioneer Fei-Fei Li, has debuted a “spatial intelligence” system that can generate 3D worlds from a single image. Although the output is not photorealistic, the tech could be a breakthrough for animation companies and video game developers. Deploying what it calls Large World Models (LWMs), World Labs is focused on transforming 2D images into turnkey 3D environments with which users can interact. Observers say that reciprocity is what sets World Labs’ technology apart from offerings by other AI companies that transform 2D to 3D.

These AI-created environments “are already available to play with on the World Labs website with more to come in the future,” writes Tom’s Guide, noting that “companies like Netflix are exploring how to deploy this type of AI gaming, and Midjourney is actively looking to create a Holodeck system.”

The World Labs demos show how its app can simulate camera effects like depth of field and dolly shots. The system can also apply effects, like changing color and lighting. The company says the tech is in the early preview stage, with improvements to come.

While transforming 2D images into 3D using AI has been done before, “this is the first time we’ve seen AI which comes with the kind of motion-focused interactivity which is going to be vital for future AI applications,” suggests Tom’s Guide.

The WorldLab imagery is “rendered live in the browser,” TechCrunch explains, pointing out that the output is not only interactive but modifiable, with an emphasis on consistency. “World Labs’ approach ensures that scenes remain the same once they’re generated and that they obey the basic laws of physics, meaning they have a sense of solidity and depth.”

“World Labs will develop spatially intelligent LWMs that can understand and reason about the 3D world from images and other modalities,” per a blog post by investor Radical Ventures. “Over time, we expect to train increasingly powerful models with broader capabilities that can be applied in a variety of domains, working alongside people.”

“By modeling 3D worlds, using ‘spatial intelligence’ AI, the team hopes to become a key player in the future of robotics, AI video and the metaverse,” Tom’s adds, pointing to “a potential Holodeck future for us all.”