World Labs' Atlas rebuilds The Matrix bullet time with just 3 iPhones, cutting 3D capture cost 50-100x
drfeifei · x · 2026-09-05
In an a16z conversation, World Labs co-founders Fei-Fei Li, Justin Johnson, Ben Mildenhall and Martin Casado detail Atlas, a world model for spatial intelligence.
Core idea: LLMs use next-token prediction, video models next-frame prediction, while Atlas is built on new-view prediction — the first model to unify pixel generation and pixel reconstruction, two problems computer vision has kept separate for half a century.
The practical result is a 50-100x reduction in the cost of capturing a 3D representation of a space: from 100-300 photos down to just three videos. Johnson illustrates with The Matrix's bullet-time shot, which needed hundreds of cameras and a green screen; with Atlas, three iPhones on tripods suffice to reframe, freeze time and fly the camera through the scene.
More from Models
- GPT-6 Astra and Claude Fable 5.1 write $10,000 research papers — experts reject all — sayashk · 2026-10-21
- GitHub Copilot's HydraFusion Orchestrates Multiple Models Per Task — WirelessLife · 2026-10-21
- OpenVLHarness turns any MLLM into a visual reasoner, +9.9 points across 22 datasets — yining_hong · 2026-10-21
- User claims Gemini analyzed his tennis match video and improved his game — zacharynado · 2026-10-21
- Opus 5.5 tops OpenRouter spending in 20 of 29 task categories — mmurph · 2026-10-21
- Meta team builds open-source 30B LLM from scratch in months, hails 'second era' — TimDarcet · 2026-10-21