You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Since there are no boundaries provided in the repo, I've tried my own to change facial attributes. Unfortunately, my boundaries work well if I use them only with pSp or e4e, but most of them do not work well with your encoder. They produce very ugly output images. I wonder why, since you also use StyleGAN2. So what is the reason and the sollution? And how do you look for directions using InterFaceGAN if its originally used for StyleGAN1?
Since there are no boundaries provided in the repo, I've tried my own to change facial attributes. Unfortunately, my boundaries work well if I use them only with pSp or e4e, but most of them do not work well with your encoder. They produce very ugly output images. I wonder why, since you also use StyleGAN2. So what is the reason and the sollution? And how do you look for directions using InterFaceGAN if its originally used for StyleGAN1?
Hi sorry, we had an issue with the git. The boundary files have been added. For the second question, InterFaceGAN can be trained in the latent space of StyleGAN2 as well.
Activity
panda-2014 commentedon Nov 18, 2022
U may use author's inversion to generate latents for interface training.
Xu-Yao commentedon Nov 18, 2022
Hi sorry, we had an issue with the git. The boundary files have been added. For the second question, InterFaceGAN can be trained in the latent space of StyleGAN2 as well.
panda-2014 commentedon Nov 20, 2022
wonderful works!