search

QhelDIV / ShapeFormer

Official repository for the ShapeFormer Project
https://shapeformer.github.io/
190 stars 22 forks source link

data processing of a new mesh from shapenet #9

Closed linjing7 closed 1 year ago

linjing7 commented 1 year ago

Hi, thanks for your elegant and solid work. I have two questions about the code:

  1. I want to perform shape completion with some meshes (e.g., a subset from shapenet, which is not included in IMNet). I wonder whether the data processing of the mesh is the same as DFAUST? That is, can I run generate_dfaust_sdf_samples.py to generate the training pair? If not, could you please provide the data processing code? Thanks a lot!
  2. If I want to train the model on DFAUST, except changing the datamodule_opt in the yaml file, what else needs I to modify?
QhelDIV commented 1 year ago
  1. Yes, this generate_dfaust_sdf_samples.py file generate the training data from a mesh.
  2. You only need to modify the datamodule_opt configs since the model hyperparameters are the same.
linjing7 commented 1 year ago

Okay, thanks for your reply!