Hair combing dynamics modeling on 7 unseen hairstyles (test)
0.0775CD Point Error (mean)DYMO-Hair
Evaluation results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| DYMO-HairGrid size=64 × 64 × 128 grids, Voxel size=~5 mm, State-action fusion mechanism=ControlNet-style, Pre-training=pre-trained latent space2025.10 | 0.0775 | 0.124 | 12.03 | 26.12 | 0.1005 | 0.1878 | |
| V-UNetGrid size=64 × 64 × 128 grids, Voxel size=~5 mm2025.10 | 0.0792 | 0.1345 | 14.25 | 31 | 0.1047 | 0.1966 | |
| V-FiLMGrid size=64 × 64 × 128 grids, Voxel size=~5 mm, State-action fusion mechanism=FiLM-style, Pre-training=trained on the same pre-trained latent space as ours2025.10 | 0.0807 | 0.1334 | 13.6 | 29.59 | 0.1065 | 0.2011 | |
| PC-GNNResolution=2K-point, Implementation=message-passing2025.10 | 0.0814 | 0.1359 | 13.34 | 30.16 | 0.1052 | 0.1983 |