BREAKING
Robbyant Open-Sources LingBot-Map
0
FPS
at 518x378
0
+
frames stable
0
x
consumer GPU
Phone Video to 3D in One Pass
1
MP4 phone video
↓
2
Feed-forward pass
↓
3
Point cloud + trajectory
LingBot-Map vs Classic SLAM
Classic methods
ORB-SLAM2
●
Iterative optimization
●
Heavy compute
●
Drift on long clips
LingBot-Map
GCT
●
Single feed-forward pass
●
Streaming input
●
Stable over long footage
Apache 2.0, No LiDAR Needed
Video-to-3D Goes Local
AI NEWS BLITZ
Ant Group's Robbyant has open-sourced a model that turns phone videos into 3D scenes.