Text-to-Video Generation on MovieGen VideoBench 250 Randomly Selected Prompts
61.45Text Alignment Win RateS2DiT-KD
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| S2DiT-KDVariant=Knowledge Distilled (KD)2026.01 | 61.45 | 54.3 | |
| S2DiT-ARVariant=Autoregressive (AR)2026.01 | 60.08 | 57.39 | |
| S2DiT-Pre.Variant=Pre-trained2026.01 | 49.8 | 46.99 | |
| Wan2.1-1.3BComparison Group=Group 1 (vs S2DiT-Pre.)2026.01 | 47.39 | 43.37 | |
| Wan2.1-1.3BComparison Group=Group 3 (vs S2DiT-AR)2026.01 | 37.48 | 35.62 | |
| Wan2.1-1.3BComparison Group=Group 2 (vs S2DiT-KD)2026.01 | 36.66 | 36.29 | |
| LTX-0.9.5Comparison Group=Group 1 (vs S2DiT-Pre.)2026.01 | 2.81 | 9.64 | |
| LTX-0.9.5Comparison Group=Group 3 (vs S2DiT-AR)2026.01 | 2.44 | 6.99 | |
| LTX-0.9.5Comparison Group=Group 2 (vs S2DiT-KD)2026.01 | 1.89 | 9.41 |