FLAME retopology

A pipeline to align FLAME onto various shapes on your template mesh

Step 0: Manually align your neutral/canonical mesh to FLAME's one
Step 1: Compute correspondences
Step 2: Retopology your meshes with different shapes and expressions into FLAME by optimizing FLAME parameters with PyTorch

iPhone mesh sample

git submodule update --init
- FLAME_PyTorch is registered as git submodule. Slightly modified version is used to avoid crash.
Download FLAME model and extract to ./model/
- 2020 version is recommended
Download addtional landmark info used for RingNet from here and put to ./model/FLAME2019/
Generate neutral FLAME as obj
- python save_FLAME_neutral_obj.py
Donwload sample iphone data from here and extract to ./data/
- It includes manually aligned iPhone mesh onto FLAME neutral mesh, which was done by Wrap3. Other sample inputs are also included.
Compute correspondences between neutral meshes of FLAME and mannually aligned iPhone
- python correspondence.py
Perform retopogy from iPhone to FLAME
- python retopology.py

Name		Name	Last commit message	Last commit date
Latest commit History 5 Commits
FLAME_PyTorch @ d4f1d3d		FLAME_PyTorch @ d4f1d3d
data		data
.gitmodules		.gitmodules
FLAME_util.py		FLAME_util.py
README.md		README.md
correspondence.py		correspondence.py
obj_io.py		obj_io.py
retopology.py		retopology.py
save_FLAME_neutral_obj.py		save_FLAME_neutral_obj.py