Video Frame Interpolation on UCF101 (PSNR, LPIPS, FID, FloLPIPS, PS)
34.7PSNRLDMVFI
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| LDMVFIEvaluation protocol=VFI task-specific2026.03 | 34.7 | 0.024 | 12.93 | 0.032 | 126.43 | |
| LoRA-r128-N256LoRA rank=128, Training samples=256, Multi-dataset training=false2026.03 | 31.73 | 0.034 | 15.48 | 0.044 | 90.27 | |
| LoRA-r128-N256-MultiLoRA rank=128, Training samples=256, Multi-dataset training=true2026.03 | 31.71 | 0.034 | 16.36 | 0.045 | 95.65 | |
| LoRA-r64-N128LoRA rank=64, Training samples=128, Multi-dataset training=false2026.03 | 30.99 | 0.036 | 17.29 | 0.048 | 83.39 | |
| LoRA-r128-N128LoRA rank=128, Training samples=128, Multi-dataset training=false2026.03 | 30.96 | 0.037 | 16.95 | 0.048 | 84.86 | |
| LoRA-r128-N64LoRA rank=128, Training samples=64, Multi-dataset training=false2026.03 | 30.52 | 0.039 | 16.94 | 0.052 | 80.22 | |
| LoRA-r64-N64LoRA rank=64, Training samples=64, Multi-dataset training=false2026.03 | 30.02 | 0.043 | 20.27 | 0.059 | 78.31 | |
| Baseline (image edit)Status=Frozen, Backbone=Qwen-Image-Edit2026.03 | 19.08 | 0.295 | 74.98 | 0.318 | 33.77 |