Customized Video Generation on Subject-video-prompt pairs (test)
36.3CLIP-TSMRABooth
Evaluation Results
| Method | Links | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| SMRABoothArchitecture=DiT, Backbone=WAN2.12025.12 | 36.3 | 70 | 51.9 | 62.89 | 95.31 | 98.8 | 21.14 | 62.18 | 67.46 | |
| DualRealArchitecture=DiT, Backbone=WAN2.12025.12 | 35.1 | 69.2 | 50.9 | 45.75 | 94.07 | 98.2 | 20.58 | 61.03 | 65.27 | |
| WAN2.1Architecture=DiT, Backbone=WAN2.12025.12 | 33.9 | 58.6 | 16.5 | 35.18 | 95.62 | 98.8 | 19.96 | 61.09 | 65.12 | |
| WAN2.1 + LoRAsArchitecture=DiT, Backbone=WAN2.1, Tuning=LoRAs2025.12 | 31.4 | 68.1 | 46.4 | 60.08 | 94.33 | 98 | 19.85 | 56.17 | 55.9 |