Point-track conditioned video generation on TapVid3D ADT
41FIDGo-with-the-Track
Evaluation Results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| Go-with-the-TrackBackbone=Wan2.2 14B2026.06 | 41 | 314.3 | 0.291 | 17.84 | 67.3 | 4.429 | |
| Go-with-the-TrackBackbone=Wan2.1 1.3B2026.06 | 45.88 | 366 | 0.307 | 17.91 | 65.9 | 5.342 | |
| Go-with-the-TrackBackbone=Wan2.1 14B2026.06 | 46.56 | 361.7 | 0.297 | 17.17 | 67.6 | 4.282 | |
| Wan-MoveBackbone=Wan2.1 14B2026.06 | 61.37 | 533.2 | 0.427 | 15.91 | 59.3 | 11.18 | |
| ATIBackbone=CogVideoX 5B2026.06 | 67.75 | 614.5 | 0.447 | 13.69 | 56.8 | 23.62 | |
| GWTFBackbone=CogVideoX 5B2026.06 | 75.28 | 632.3 | 0.43 | 14.61 | 60.1 | 15.6 | |
| DASBackbone=CogVideoX 5B2026.06 | 80.07 | 707.6 | 0.498 | 14.95 | 56.1 | 27.79 |