search

eladrich / pixel2style2pixel

Official Implementation for "Encoding in Style: a StyleGAN Encoder for Image-to-Image Translation" (CVPR 2021) presenting the pixel2style2pixel (pSp) framework
https://eladrich.github.io/pixel2style2pixel/
MIT License
3.2k stars 568 forks source link

Model parameters at other resolutions #288

Closed li-car-fei closed 2 years ago

li-car-fei commented 2 years ago

Can you provide the model parameters trained on the dataset FFHQ with a resolution of 256?

yuval-alaluf commented 2 years ago

We don't have such a model available. SG2 generators trained on FFHQ at a resolution of 256 are available on the official SG2 repo from nvidia.

li-car-fei commented 2 years ago

We don't have such a model available. SG2 generators trained on FFHQ at a resolution of 256 are available on the official SG2 repo from nvidia.

Can an encoder with a resolution of 10241024 be used for stylegan2 with a resolution of 256256?

yuval-alaluf commented 2 years ago

No since the latent spaces are different. You will need to retrained your encoder, although you could try starting from the pretrained encoder rather than training from scratch.