Image-to-Video Generation on VBench White-box quantitative results
97.7Subject FidelityLTX-Video
Evaluation Results
| Method | Links | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| LTX-VideoMode=Clean, Base Model=LTX-Video2026.07 | 97.7 | 98.15 | 95.09 | 97.87 | 60.87 | 68.22 | 99.04 | 99.36 | |
| CogVideoXMode=Clean, Base Model=CogVideoX2026.07 | 97.15 | 98.87 | 95.06 | 97.38 | 61.48 | 70.23 | 96.99 | 98.15 | |
| I2VGuardBase Model=CogVideoX2026.07 | 94.58 | 97.14 | 93.32 | 96.17 | 58.61 | 65.3 | 96.47 | 97.61 | |
| I2VGuardBase Model=LTX-Video2026.07 | 94.36 | 95.56 | 90.25 | 95 | 56.28 | 62.63 | 98.23 | 98.9 | |
| Anti-PromptBase Model=LTX-Video2026.07 | 93.64 | 95.07 | 89.36 | 94.99 | 54.99 | 61.44 | 97.95 | 98.67 | |
| Anti-PromptBase Model=CogVideoX2026.07 | 91.2 | 94.28 | 87.54 | 93.16 | 54.72 | 63.71 | 94.47 | 95.66 |