Body Model Visualizer

Introduction

This is a simple Open3D-based GUI for SMPL-family body models. This GUI lets you play with the shape, expression, and pose parameters of SMPL, SMPL-X, MANO, FLAME body models. Features include:

• Interactive editing of shape, expression, pose parameters

01_model_editing.mp4

• Visualize body model joints and joint names

02_visualize_joints.mp4

• Simple IK solver to match an input pose

03_simple_ik.mp4

• Save edited model parameters

04_save_params.mp4

• View controls

05_viewing.mp4

• Lighting controls

06_lighting.mp4

• Material settings

07_material.mp4

• Web visualization support

Even though there are existing Blender/Unity plugins for these models, our main audience here is researchers who would like to quickly edit/visualize body models without the need to install a graphics software.

Installation

Clone the repo and install the requirements (use python3.9).

pip install -r requirements.txt

Download the SMPL, SMPL-X, MANO, FLAME body models:

• SMPL: https://smpl.is.tue.mpg.de/ (v1.1.0)
• SMPL-X: https://smpl-x.is.tue.mpg.de/ (v1.1)
• MANO: https://mano.is.tue.mpg.de/
• FLAME: https://flame.is.tue.mpg.de/
  • For landmarks: https://github.com/soubhiksanyal/RingNet/blob/master/flame_model/

Copy downloaded files under data/body_models, this folder should look like:

data
└── body_models
    ├── flame
    │   ├── FLAME_FEMALE.pkl
    │   ├── FLAME_MALE.pkl
    │   ├── FLAME_NEUTRAL.pkl
    │   ├── flame_dynamic_embedding.npy
    │   └── flame_static_embedding.pkl
    ├── mano
    │   ├── MANO_LEFT.pkl
    │   └── MANO_RIGHT.pkl
    ├── smpl
    │   ├── SMPL_FEMALE.pkl
    │   ├── SMPL_MALE.pkl
    │   └── SMPL_NEUTRAL.pkl
    └── smplx
        ├── SMPLX_FEMALE.npz
        ├── SMPLX_MALE.npz
        └── SMPLX_NEUTRAL.npz

Finally, run:

python main.py

If you want to enable web visualization, run:

python main.py --web

Guidelines

Saved model parameters

File > Save Model Params lets you save the edited body model parameters. Output is a pickled python dictionary with below keys:

dict_keys(['betas', 'expression', 'gender', 'body_model', 
           'joints', 'body_pose', 'global_orient'])