Skip to content

Latest commit

 

History

History
55 lines (45 loc) · 2.52 KB

models.md

File metadata and controls

55 lines (45 loc) · 2.52 KB

reference: https://github.com/sxyu/smplxpp/blob/master/data/models/README.md

SMPL-X

  1. Please go to http://smpl-x.is.tue.mpg.de and register
  2. In downloads, click "download SMPL-X v...".
  3. Open the downloaded zip and extract models/smplx/SMPLX_*.npz to ./smplx, so that we have: PROJ_ROOT/data/models/smplx/SMPLX_NEUTRAL.npz

You should now have

  • data/models/smplx/SMPLX_NEUTRAL.npz
  • data/models/smplx/SMPLX_MALE.npz
  • data/models/smplx/SMPLX_FEMALE.npz

SMPL

  1. Please go to http://smpl.is.tue.mpg.de and register.
  2. In Downloads, download the 'for Python users' zip.
  3. Unzip to project root, so that there is a folder smpl on the same level as CMakeLists.txt
  4. cd to the project root directory and run (for SMPL 1.0) python tools/smpl2npz.py smpl/models/basicModel_f_lbs_10_207_0_v1.0.0.pkl smpl/models/basicmodel_m_lbs_10_207_0_v1.0.0.pkl or (for SMPL 1.1) python tools/smpl2npz.py smpl/models/basicModel_f_lbs_10_207_0_v1.1.0.pkl smpl/models/basicmodel_m_lbs_10_207_0_v1.1.0.pkl smpl/models/basicmodel_neutral_lbs_10_207_0_v1.1.0.pkl

You should have

  • data/models/smpl/SMPL_MALE.npz
  • data/models/smpl/SMPL_FEMALE.npz

Neutral

NOTE: as of SMPL 1.1, the neutral model is included in the SMPL release, and this is no longer needed

  1. Please go to http://smplify.is.tue.mpg.de and register.
  2. Download the zip under 'Code and model'.
  3. Unzip to project root, so that there is a folder smplify_public on the same level as CMakeLists.txt
  4. cd to the project root directory and run python tools/smpl2npz.py smplify_public/code/models/basicModel_neutral_lbs_10_207_0_v1.0.0.pkl

You should have

  • data/models/smpl/SMPL_NEUTRAL.npz

SMPL+H

  1. Please go to http://mano.is.tue.mpg.de and register
  2. Download 'Extended SMPLH model for AMASS' in Downloads
  3. tar xf /path/to/smplh.tar.xz
  4. Manually copy gender/model.npz to data/models/smplh/SMPLH_GENDER.npz

You should have

  • data/models/smplh/SMPLH_NEUTRAL.npz
  • data/models/smplh/SMPLH_MALE.npz
  • data/models/smplh/SMPLH_FEMALE.npz

Notes

  • SMPL/SMPL-H UV maps are provided by [github.com/radvani] in this issue Lotayou/densebody_pytorch#4
  • SMPL-X UV map is custom-made and likely not very good
  • To use your own UV map, modify data/models/<model>/uv.txt UV map format:
    • First line: n = number of UV vertices
    • next n lines: u v = float coordinates of a vertex
    • next num_faces lines (num_faces is the number of rows in 'f' of the model): a b c = int, 1-based indices of UV vertices (from above) in each triangle