OpenS2V-Eval
Benchmarks
| Task Name | Dataset Name | SOTA Result | Trend | |
|---|---|---|---|---|
| Subject-preserving video generation | OpenS2V-Eval Human-Domain | Total Score64.38 | 17 | |
| Subject-to-Video Generation | OpenS2V-Eval zero-shot (test) | Total Score60.42 | 16 | |
| Video Personalization | OpenS2V-Eval & Self-Constructed Cross-Domain (test) | NANO-CLIP Score0.636 | 11 | |
| Reference-to-video generation | OpenS2V-Eval 2025a | Total Score57.91 | 9 | |
| Multi-Subject Video Generation | OpenS2V-Eval | Total Score61.01 | 6 | |
| Video Generation | OpenS2V-Eval | VBench Avg.0.806 | 6 |
Showing 6 of 6 rows