World Labs launches RTFM world model generating explorable 3D worlds from one image
drfeifei · x · 2026-09-09
World Labs (Fei-Fei Li's startup) released RTFM, a generative interactive world model that turns a single image into real-time video for exploring 3D worlds. Trained on large-scale video data to predict next frames, it achieves unbounded persistence via spatial memory; a live demo is available.
A team member also demoed Atlas: starting from a photo of a kids' playroom, Atlas-turbo prompted an entirely new outdoor area into existence — breaking past the small, enclosed spaces of the original RTFM demo.
More from Multimodal
- World Labs demos real-time next-view prediction with accurate 3D consistency — drfeifei · 2026-09-09
- I tested the top rated LoRAs and checkpoints on CivitAI for August — cointalkz · 2026-09-09
- Filmmaker releases 'HAUL' prologue built with the latest AI filmmaking tools — Eric520CC · 2026-09-09
- A custom Astra skill turns photos into Monet-style impressionist travel logs — floguo · 2026-09-09
- GPT-6 recreates the Neuromancer poster font in 13 minutes with one prompt — bennash · 2026-09-09
- Painterly samples surface from rumored OpenAI model GPT-Image-2.5 Sunburst — DavidmComfort · 2026-09-09