StorySoftware
Atlas: World Labs releases omni world model handling text, images, video and 3D natively
World Labs released Atlas, an advanced world model pretrained from scratch to natively operate on text, images, video and 3D data. Atlas enables camera-controlled video generation up to 60 seconds at 1440p, reconstructs 3D from sparse images and simulates space-time dynamics for robotics.
- Omni model pretrained on text, images, video and 3D as native modalities
- Generates 1440p video up to 60 seconds with precise camera control
- Reconstructs 3D geometry and texture from 1-2 reference images
- Handles space-time simulation for robotics and physics prediction
- Released in early access with select partners; general availability pending