Re:Atlas: A World Model for Spatial Intelligence
by exe34 on Tuesday September 01, 2026 @05:11PM
(#49528253)
Any ideas if it keeps continuity - when you turn around, is the invented chair still there, or does everything shift like in a dream?
[ Parent
| on HN ]
Re:Atlas: A World Model for Spatial Intelligence
by jcjohns on Tuesday September 01, 2026 @05:44PM
(#49528675)
Yes, one of the main goals of this model was to keep 3D consistency as the camera moves around; it can do this even without relying on an explicit 3D point cloud or Gaussian splat representation of the scene.
Re:Atlas: A World Model for Spatial Intelligence
by exe34 on Tuesday September 01, 2026 @05:44PM
(#49528678)
Amazing! Thanks for explaining!
Re:Atlas: A World Model for Spatial Intelligence
by dmarcos on Tuesday September 01, 2026 @05:19PM
(#49528377)
I believe so. This is not a model that generates pixels frame by frame from user input like genie 3. Instead, there’s an actual 3D scene / structure generated (point cloud, 3dgs) from the input images.
Re:Atlas: A World Model for Spatial Intelligence
by xyzsparetimexyz on Tuesday September 01, 2026 @08:10PM
(#49530092)
Wrong, it does go straight to generating images.
Re:Atlas: A World Model for Spatial Intelligence
by dmarcos on Tuesday September 01, 2026 @09:04PM
(#49530465)
From blog post:
“It generates both image frames from novel views and explicit 3D outputs”
Model can indeed generate novel views but I don’t think in real time (I could be wrong). If you want to navigate a space in real time as user above was asking gotta rely on the 3D output and the explicit representation will provide continuity. User likely was referring to the limitations we’ve seen on video-style world models like genie and precursors.