VGGT-Ω is introduced as a 4D foundation model for spatial intelligence that generates point clouds, meshes, and colored grids from video across static and dynamic scenes

— 48-second demo video shows reconstructions from drone flights and extreme conditions.

Original post

GG#665@GEORGIAGKIOXARIOPJIJianyuan|@JIANYUAN_WANG

Introducing VGGT-Ω: scaling feed-forward reconstruction across static and dynamic scenes, and studying whether the learned geometric representations transfer beyond reconstruction.

6:04 PM · May 18, 2026 View on X

Reposted by

KD#908|@CSPROFKGD

QUOTE POSTBS#1909Bilawal Sidhu|@BILAWALSIDHU

This thing just rips at 3d reconstruction of otherwise impossible scenarios — fpv flying through windows, people flying through the sky; I mean damn!

JIJianyuan@jianyuan_wang

Introducing VGGT-Ω: scaling feed-forward reconstruction across static and dynamic scenes, and studying whether the learned geometric representations transfer beyond reconstruction.

1:04 AM · May 19, 2026 · 719.2K Views

12:34 PM · May 19, 2026 · 8.5K Views