search

c-he / NeMF

[NeurIPS 2022] Official implementation of "NeMF: Neural Motion Fields for Kinematic Animation"
MIT License
156 stars 9 forks source link

Regarding preprocessing script for AMASS Dataset #13

Closed KosukeFukazawa closed 1 year ago

KosukeFukazawa commented 1 year ago

Thank you for sharing your fantastic work! I have a question about src/datasets/amass.py. AMASS dataset uses SMPL+H skeleton topology, but your script seems to use SMPL skeleton topology. According to this repository, the 24th joint of SMPL is right hand, but the 24th joint of SMPL+H is left_index2. Are you referring to the wrong rotation value on this joint?

c-he commented 1 year ago

Hi, thanks for your interest in our work! For data preprocessing, we first use the scripts provided in HuMoR. As you can see in L355-L357, they separately load and store root, body and hand joints, and there are 21 body joints in total. Therefore, we actually ignore the last 2 joints which, as you said, are different from SMPL/SMPL-H/SMPL-X. Instead, we pad 2 identity rotations at last, making our processed data still having 24 joints and suitable for visualization in Blender. https://github.com/c-he/NeMF/blob/79918430970fd138ae730510459c8f34893a3f86/src/datasets/amass.py#L81

KosukeFukazawa commented 1 year ago

Thank you for your kind reply! I did not check HuMoR script.