Subject-driven video generation on Human-Object Interaction (HOI) scenes
71.7DINO (v -> r)MV-S2V
Evaluation Results
| Method | Links | ||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| MV-S2VReference View Configuration=Multi-View (MV)2026.01 | 71.7 | 71.3 | 86.4 | 87.2 | 15.4 | 16.1 | 4.3 | 2.2 | 60.1 | 75.9 | 99.5 | 21.6 | |
| PhantomReference View Configuration=Single-View (SV)2026.01 | 68.2 | 68.7 | 85.9 | 86.8 | 15.3 | 16.2 | 33.1 | 27.6 | 59.8 | 75.1 | 99.3 | 19.9 | |
| PhantomReference View Configuration=Multi-View (MV)2026.01 | 65.9 | 67.5 | 82.8 | 85 | 17.7 | 17.3 | 30.6 | 29.5 | 57.6 | 72.3 | 99.2 | 19.7 | |
| MAGREFReference View Configuration=Single-View (SV)2026.01 | 58.5 | 60.6 | 80.2 | 82.5 | 20.3 | 18.6 | 33.9 | 29.7 | 57.9 | 74.4 | 98.6 | 19.6 | |
| MAGREFReference View Configuration=Multi-View (MV)2026.01 | 58.5 | 61.9 | 80.4 | 83.1 | 19.8 | 18 | 41.5 | 33 | 55.9 | 74 | 98.6 | 19.5 |