World's First Multimodal World Model: Fei-Fei Li's World Labs Unveils Atlas, Enabling Pixel-Perfect Camera Control and 'Bullet Time' Cinematics
Summary
IT Home reported on September 2 that Fei-Fei Li's startup World Labs released 'the world's first multimodal world model' — Atlas, which can generate images and video frames with pixel-perfect camera control and reconstruct them in 3D. Atlas supports text-to-image generation, 360-degree panoramas, and can reconstruct real scenes, outputting up to 1 minute of 1440p video. Some partners have already received early access, with broader early access opening in the coming weeks.
Bubble analysis
This news is merely about the release of an AI model, with no investment amounts or valuations involved, so it is far removed from the AI bubble discussion. It may affect expectations about AI capabilities but does not directly change investment logic.
Related signals
linked by shared tags
- Fei-Fei Li's World Labs Brings Cyberpunk's 'Braindance' to Life#world-model#worldlabs
- Just Now, Fei-Fei Li Upsets the Table! World's First Multimodal World Model Released, a Few Photos Save Hundreds of Camera Positions#world-model#atlas
- Hard Kr First | 1999-born Tongji Associate Professor and former DJI core member starts world model venture, raises tens of millions in funding within a month of founding#world-model
- Unitree: First Fully Autonomous Humanoid Robot Fighting#world-model
- Unitree: World's First Real-Time World Model-Driven Fully Autonomous Humanoid Robot Fighting#world-model