search

williamyang1991 / StyleGANEX

[ICCV 2023] StyleGANEX: StyleGAN-Based Manipulation Beyond Cropped Aligned Faces
Other
510 stars 36 forks source link

Question about video editing training. #26

Closed Dratlan closed 4 months ago

Dratlan commented 4 months ago

Thanks for your great work! I'm now trying to train a new edit direction: slender. I'm a little confused about the data preparing. In data.config it shows the realign320 for training. but in the paper says training with generated data. besides, is it all the 70000 in-the-wild image are used for training and testing?

williamyang1991 commented 4 months ago 
We process FFHQ [17] to obtain 70,000 aligned training images of 1280 × 1280 resolution for all tasks except two video-related tasks that use StyleGAN generated data.

Only video editing and video stylization use generated data.

Other tasks use realign320 for training.

Yes, we use all 70000 images for training.

Dratlan commented 4 months ago

Got it, thanks for your quick relpy