search

edward1997104 / Wavelet-Generation

Neural Wavelet-domain Diffusion for 3D Shape Generation [SIGGRAPH Asia 2022]
MIT License
86 stars 12 forks source link

Elaboration requested on the config's 'data_files' #2

Closed MarkTension closed 1 year ago

MarkTension commented 1 year ago

Thanks for your interesting work! I have a question on the config file. What is the exact difference in config settings when generating the 2 different data files? I assumed simply reducing resolution_index in convert_wavelet.py by 1 would generate the second wavelet file from the SDFs, but is that correct?

data_files = [('wavelet_data/03001627_0.1_bior6.8_3_zero.npy', 3), ('wavelet_data/03001627_0.1_bior6.8_2_zero.npy', 2)]

Thanks a lot!

edward1997104 commented 1 year ago

Yes, you are correct. If you reduce the resolution_index by 1, you can generate the second file, which is the detail coefficient volume for the shape. Generally, the resolution_index of "3" represents the coarse coefficient volume, while "2" represents the detail coefficient volume.