PANews
PANews|Oct 16, 2025 15:21
**[World Labs Releases Real-Time Generative World Model RTFM: Real-Time 3D World Generation with a Single H100 GPU]** WorldLabs, the spatial intelligence company founded by Stanford University professor Fei-Fei Li, has officially released the research preview of its next-generation real-time generative model, RTFM (Real-Time Frame Model). This model enables the real-time generation of videos and 3D scenes during interactions, creating a continuously existing virtual world. RTFM is designed around three core principles: efficiency, scalability, and persistence. The model can run at interactive frame rates on a single NVIDIA H100 GPU, generating scenes with complex effects such as reflections, shadows, and gloss from a single image without explicit 3D modeling. Its "spatial memory" mechanism maintains the persistent existence of the world through frame pose modeling and "context juggling," supporting long-term interactive exploration by users.
+6
Mentioned
Share To

Timeline

HotFlash

APP

X

Telegram

Facebook

Reddit

CopyLink

Hot Reads