Image-to-Video Generation on DAVIS 2017 (val)
316.91FVDTrack4Gen
Evaluation Results
| Method | Links | |
|---|---|---|
| Track4GenBackbone=SVD, Training Time=1.07x2026.03 | 316.91 | |
| SFTBackbone=SVD, Training Time=1x2026.03 | 322.81 | |
| BaseBackbone=SVD2026.03 | 395.87 | |
| SHIFTBackbone=SVD, Training Time=2.46x2026.03 | 396.45 | |
| FlowGRPOBackbone=SVD, Training Time=19.35x2026.03 | 396.56 | |
| DenseDPOBackbone=SVD, Training Time=2.40x2026.03 | 399.73 |