| Paper | Code | ATE | RPE-T | RPE-R | ModelName | ReleaseDate |
|---|---|---|---|---|---|---|
| Video Generation Models are General-Purpose Vision Learners | 0.050 | 0.018 | 0.464 | GenCeption Specialist-L | 2026-07-10 | |
| Efficiently Reconstructing Dynamic Scenes One D4RT at a Time | 0.065 | 0.024 | 0.126 | D4RT | 2025-12-09 | |
| 4D-VGGT: A SpatioTemporal Foundation Model for Dynamic Scene Geometry Estimation | 0.065 | 0.026 | 0.249 | 4D-VGGT | 2026-09-08 | |
| TrackingWorld: World-centric Monocular 3D Tracking of Almost All Pixels | ✓ Link | 0.088 | TrackingWorld (DELTA) | 2025-12-09 | ||
| LEAP-VO: Long-term Effective Any Point Tracking for Visual Odometry | 0.089 | 0.066 | 1.25 | LEAP-VO (extra data) | 2024-01-03 | |
| Gen4P: Video Generators are 4D Perceivers | ✓ Link | 0.104 | 0.023 | 0.306 | Gen4P (Ours-II) | 2026-10-15 |
| V-DPM: 4D Video Reconstruction with Dynamic Point Maps | ✓ Link | 0.105 | V-DPM | 2026-01-14 | ||
| DynamicVerse: A Physically-Aware Multimodal Framework for 4D World Modeling | ✓ Link | 0.108 | DynamicGen | 2025-12-02 | ||
| RayMap3R: Inference-Time RayMap for Dynamic 3D Reconstruction | ✓ Link | 0.166 | 0.056 | 0.72 | RayMap3R | 2026-03-21 |