Human cinemagraph generation on Synthetic dataset (test)
6.52MAEToy
Evaluation Results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| Toydescription=baseline generating output V = {I, I, ..., I} from input image I2023.03 | 6.52 | 197.73 | 12.36 | 61 | 40.76 | 218.39 | |
| CycleNetData modality=RGB2023.03 | 7.64 | 175.24 | 12.21 | 56 | 37.64 | 230.4 | |
| Controllable Anim.2023.03 | 9.37 | 302.26 | 15.89 | 48 | 34.49 | 220.86 | |
| CycleNetData modality=Normals, Note=Ground truth normal maps, Re-shading=True2023.03 | 15.75 | 563.76 | 22.63 | 41 | 32.11 | 231.29 | |
| CycleNetData modality=Estimated normals, Note=Normal maps estimated from synthetic RGB data, Re-shading=True2023.03 | 17.32 | 674.62 | 24.31 | 34 | 31.81 | 231.74 | |
| LIAmode=single image-based controllable video generation2023.03 | 23.77 | 1,131.62 | 32.45 | 11 | 25.5 | 400.26 |