search

bryandlee / animegan2-pytorch

PyTorch implementation of AnimeGANv2
MIT License
4.37k stars 641 forks source link

ResizeRight and Antialiased CNNs recommendation #6

Open torridgristle opened 2 years ago

torridgristle commented 2 years ago

Rather than F.interpolate with optional alignment of corners you could try ResizeRight and in places where convolution strides are above 1 you can add Antialiased CNNs's BlurPool after them and their activation functions to make it more shift-invariant, hopefully increasing quality and making things more consistent. It's thankfully painless to implement.

bryandlee commented 2 years ago

Hi, thanks for the suggestions! I've been actually trying out some translation-equivariance techniques for video applications.