← Back to live feed1 story
World Labs demonstrated its spatial intelligence by simulating a walkthrough of the Voyager headquarters for Nvidia. The model used 32 input images to generate a 3D reconstruction that allows for pixel perfect camera control and a flight like experience through the campus. The AI was pretrained from scratch on Blackwell GPUs.
Atlas incorporates text, images, video, and 3D assets into a shared spatial context to simulate environments and reconstruct scenes. The demonstration was built on Nvidia's open platform, allowing users to choose their path and view specific areas of the park in real time.
Sign in to suggest edits
Key sources
- SOURCE@nvidiaai“Atlas brings text, images, video, and 3D into a shared spatial context, enabling one model to generate new views, reconstruct scenes, and simulate worlds”x.com
- SUPPORT@nvidia“From just 32 images of our campus, World Labs’ Atlas model created a new way for you to move through our park in real time”x.com
- SUPPORT@drfeifei“From 32 input images to real-time flight through @nvidia's Voyager headquarters”x.com
- SOURCEmarketbrief.now