World Labs, co-founded by Fei-Fei Li, launched Atlas on September 1: a multimodal autoregressive diffusion transformer pretrained from scratch to natively handle text, images, video, and 3D in a shared spatial context. It generates camera-controlled video up to about 1 minute at 1440p, reconstructs scenes from sparse views with point-cloud and Gaussian-splat outputs, and supports Real-to-Sim robotics workflows. Atlas is in early access with select partners and will power future Marble products.

Key Takeaways

  • Native camera geometry for ~1-minute 1440p camera-controlled generation
  • Sparse-view 3D reconstruction with point clouds and Gaussian splats, plus Real-to-Sim for robotics
  • Early access for select partners; will power future Marble products
ADSponsored