Text-to-Video Generation on PhyGenBench
28Mec ScoreMMPhysVideo
Evaluation Results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| MMPhysVideoBackbone=CogVideoX-5B, Distillation strategy=true2026.04 | 28 | 32 | 36 | 25 | 30 | 30.6 | |
| CogVideoX1.5-5BBackbone=CogVideoX1.5-5B, Distillation strategy=false2026.04 | 27.5 | 34 | 40 | 32.5 | 33.1 | 29.4 | |
| MMPhysVideoBackbone=Wan2.1-1.3B, Distillation strategy=true2026.04 | 27.5 | 56 | 30.7 | 35 | 39.4 | 38.1 | |
| MMPhysVideoBackbone=CogVideoX-2B, Distillation strategy=true2026.04 | 25.5 | 25 | 14 | 23.5 | 23.1 | 24.4 | |
| CogVideoX-2BBackbone=CogVideoX-2B, Distillation strategy=false2026.04 | 25 | 14 | 13.3 | 15 | 16.9 | 18.1 | |
| VideoREPABackbone=CogVideoX-5B, Distillation strategy=false2026.04 | 25 | 30 | 28.3 | 24.5 | 27.5 | 25.6 | |
| Wan2.1-1.3BBackbone=Wan2.1-1.3B, Distillation strategy=false2026.04 | 25 | 54 | 30 | 30 | 36.2 | 37.5 | |
| CogVideoX-5BBackbone=CogVideoX-5B, Distillation strategy=false2026.04 | 20.5 | 35 | 20 | 15 | 23.8 | 21.2 | |
| VideoREPABackbone=CogVideoX-2B, Distillation strategy=false2026.04 | 20 | 24 | 16.7 | 20 | 21.2 | 22.5 | |
| PhyT2VBackbone=CogVideoX-5B, Distillation strategy=false2026.04 | 20 | 32 | 23.3 | 22.5 | 25 | 22.5 | |
| PhyT2VBackbone=CogVideoX-2B, Distillation strategy=false2026.04 | 18.5 | 20 | 16.7 | 17.5 | 18.8 | 20.6 | |
| OmniVDiffBackbone=OmniVDiff, Distillation strategy=false2026.04 | 5 | 20 | 16.7 | 20 | 15.6 | 19.4 |