AI startup creates walkable 3D worlds from single photos, with limitations
World Labs has developed an AI system that turns individual photos into three-dimensional environments you can explore in a web browser. While innovative, the technology faces significant technical constraints in its current form.
The company says its system stands apart from other AI models in the field because it generates consistent 3D environments rather than just pixels. These virtual spaces maintain stability and follow basic physics rules.

Limited movement, visible glitches
The technology's current limitations are substantial. Users can only move a few virtual meters before hitting invisible barriers. The system also shows regular rendering problems, with objects sometimes merging incorrectly.
Video: Eric Solorio/World Labs
The platform includes various camera features like adjustable depth of field and dynamic lighting. World Labs describes this release as an "early preview." The company generated the initial 2D images using several AI tools: Flux 1.1 Pro, Ideogram, Midjourney, and DALL-E.
So far, World Labs has only shown demo environments. The company hasn't released preview software or published technical documentation with benchmarks.
Video: Brittani Natali/World Labs
High investment despite early stage
Despite being in early development, World Labs has secured $230 million in venture funding. Reuters reports that investors include venture firms Andreessen Horowitz and New Enterprise Associates, along with tech companies AMD, Intel, and Nvidia.
The 20-person company was founded by AI researcher Fei-Fei Li, who created the ImageNet project and previously led Google Cloud's AI division. Through World Labs, Li aims to develop "large world models" capable of understanding three-dimensional physical spaces.
While World Labs plans to launch its first commercial product in 2025, it remains unclear whether the company can overcome its technical challenges by then.
Major advances in world models
AI-powered 3D generation has seen major progress lately. WonderWorld, introduced last fall, cut generation time for new 3D scenes from hours to just ten seconds.
Another development came from startups Etched and Decart with their "Oasis" project. This system creates playable 3D worlds in real time at 20 frames per second.
While the Minecraft-style game world allows basic interactions like moving, jumping, and picking up objects, it shows consistency issues. Players turning around might find themselves in an entirely different environment than before.
AI News Without the Hype – Curated by Humans
As a THE DECODER subscriber, you get ad-free reading, our weekly AI newsletter, the exclusive "AI Radar" Frontier Report 6× per year, access to comments, and our complete archive.
Subscribe nowAI news without the hype
Curated by humans.
- Over 20 percent launch discount.
- Read without distractions – no Google ads.
- Access to comments and community discussions.
- Weekly AI newsletter.
- 6 times a year: “AI Radar” – deep dives on key AI topics.
- Up to 25 % off on KI Pro online events.
- Access to our full ten-year archive.
- Get the latest AI news from The Decoder.