LingBot-Map overview
LingBot-Map is a feed-forward 3D foundation model for streaming reconstruction. It processes ordered image sequences and produces camera poses and 3D scene geometry.
Released models
| Checkpoint | Use |
|---|---|
| lingbot-map-long | Recommended for long sequences and large-scale scenes |
| lingbot-map | Balanced checkpoint for short and long sequences |
| lingbot-map-stage1 | Stage-one checkpoint that can be loaded into VGGT for bidirectional inference |
demo.py provides interactive reconstruction. Windowed inference supports sequences longer than 3,000 frames, and the offline rendering pipeline processes long videos or image folders.
Use Installation and Quick Start to open the interactive viewer. Use 3D reconstruction benchmark to prepare datasets and run evaluation.