๐ŸŒŠ Wat3R

Underwater 3D Geometry Learning without Annotations

Upload a video or a set of images. Wat3R predicts camera poses, depth maps, and 3D point maps, then exports an interactive GLB reconstruction.

๐Ÿ“„ Paper ๐Ÿ’ป Code

Please upload a video or images, then click Reconstruct.

Prediction Mode
0 100
0 0

Example Scenes