- Create "DFaust" folder under
<workspace_folder>
.
cd <workspace_folder>
mkdir DFaust
- Download SMPL+H parameters, gets "DFaust.tar.bz2".
- Move "DFaust.tar.bz2" to
<workspace_folder>/DFaust
, and unzip to get folderDFaust_67/...
- Move "DFaust.tar.bz2" to
- Download DFaust Scan Data, gets "50002.tar.xz".
- Here use 50002 as example for the following steps.
- Unzip "50002.tar.xz" to
<workspace_folder>/DFaust
, and gets folderscans/50002/...
- Download SMPL model, gets "basicmodel_m_lbs_10_207_0_v1.0.0.pkl", "basicModel_f_lbs_10_207_0_v1.0.0.pkl".
- Move "basicmodel_m_lbs_10_207_0_v1.0.0.pkl", "basicModel_f_lbs_10_207_0_v1.0.0.pkl" to
<workspace_folder>/SMPL
- Move "basicmodel_m_lbs_10_207_0_v1.0.0.pkl", "basicModel_f_lbs_10_207_0_v1.0.0.pkl" to
- Download SMPL meta data, gets "uv_info.npz", "smpl_resample_idxs.npz".
- Move "uv_info.npz", "smpl_resample_idxs.npz" to
<workspace_folder>/SMPL
- Move "uv_info.npz", "smpl_resample_idxs.npz" to
- Download SMPL+H, gets "smplh.tar.xz".
- Move "smplh.tar.xz" to
<workspace_folder>/SMPL
, and unzip to get foldersmplh/...
- Move "smplh.tar.xz" to
- Download DMPLs, gets "dmpls.tar.xz".
- Move "dmpls.tar.xz" to
<workspace_folder>/SMPL
, and unzip to get folderdmpls/...
- Move "dmpls.tar.xz" to
- Clone AutoAvatar to
<workspace_folder>
cd <workspace_folder>
git clone https://github.com/facebookresearch/AutoAvatar.git
- Create
external
folder under<workspace_folder>
cd <workspace_folder>
mkdir external
- Clone human_body_prior under
external
folder.
cd <workspace_folder>/external
git clone https://github.com/nghorbani/human_body_prior.git
- Now we should have the folder structure as link.
- Install Anaconda or Miniconda. Then run the setup script.
cd <workspace_folder>/AutoAvatar
conda create -n AutoAvatar python=3.8
conda activate AutoAvatar
bash setup.sh
- Install human_body_prior for DFaust data preprocess.
cd <workspace_folder>/external
cd human_body_prior
python setup.py develop
- Run
DFaust_generate.py
to preprocess data. - Note that this may take a long time due to the mesh simplification.
- Mesh simplification is to speed up data loading during training.
cd <workspace_folder>/AutoAvatar
export PYTHONPATH=<workspace_folder>/AutoAvatar
python data/DFaust_generate.py --ws_dir <workspace_folder>