liren2515/DrapeNet

Question about auto-encoder architecture

sunshineatnoon opened this issue · 2 comments

sunshineatnoon commented

Hi, Thanks for open-sourcing the repo. I have a question about the auto-encoder structure. I am wondering have you tried to learn a VAE instead of AE? This way the latent space might be more smooth.

liren2515 commented

Thanks for your interest.

Actually, we didn't experiment with VAE. This is an interesting point and worth trying.

sunshineatnoon commented

Thanks for the answer!