- likes
- 1,967
- comments
- 89
Post
Today, we take our next major step in solving spatial intelligence. Introducing Atlas: the world’s first multimodal world model that generates image and video frames with pixel-perfect camera control and reconstructs them in 3D. We pretrained Atlas from scratch to take multimodal inputs, including camera movement, and turn it into 3D grounded views and explorable worlds. This means: - Architecture and construction teams can reconstruct a real site from just a handful of photos - Robotics teams can create endless environments to train and test robots, without hand modeling them - Filmmakers and designers can stage shots, instead of playing the prompt lottery - Anyone can design a world in 3D and step inside it, creating immersive and engaging experiences Atlas is a scalable foundation that enables humans and machines to collaborate in virtual and physical worlds. Blog link: https://lnkd.in/gUTxh4Xb