Physical Plausibility Verification on PhysicsIQ Wan2.1 14B I2V (N=16)
29.7Spatial IoUOracle (Upper bound)
Evaluation Results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| Oracle (Upper bound)N=16, Generator=Wan2.1 14B (I2V)2026.06 | 29.7 | 23.6 | 15.8 | 0.01 | 39.7 | |
| WMRewardN=16, Generator=Wan2.1 14B (I2V)2026.06 | 23.5 | 21 | 13 | 0.008 | 33.1 | |
| Qwen3-VLN=16, Generator=Wan2.1 14B (I2V)2026.06 | 22.7 | 19.3 | 11.7 | 0.009 | 30.7 | |
| GEOPHYSN=16, Generator=Wan2.1 14B (I2V)2026.06 | 22.5 | 23.2 | 12.8 | 0.007 | 34 | |
| Qwen2.5-VLN=16, Generator=Wan2.1 14B (I2V)2026.06 | 16.8 | 13.2 | 7.2 | 0.011 | 21 | |
| BaselineN=16, Generator=Wan2.1 14B (I2V)2026.06 | 14.3 | 13.3 | 7 | 0.018 | 18.9 | |
| VideoMAEN=16, Generator=Wan2.1 14B (I2V)2026.06 | 12.2 | 4.4 | 4.2 | 0.013 | 11 |