Can I train the model with unaligned faces? #173

Open
opened 2026-01-29 21:44:55 +00:00 by claunia · 1 comment
Owner

Originally created by @YichengWu on GitHub (Mar 15, 2022).

Based on the instruction, it says

You can try a simple version ( options/train_gfpgan_v1_simple.yml) that does not require face component landmarks.

Does it mean that I can train a model using unaligned faces without any face cropping? I guess it won't work since the styleGAN decoder requires aligned faces?

Thanks!

Originally created by @YichengWu on GitHub (Mar 15, 2022). Based on the instruction, it says > You can try a simple version ( options/train_gfpgan_v1_simple.yml) that does not require face component landmarks. Does it mean that I can train a model using unaligned faces without any face cropping? I guess it won't work since the styleGAN decoder requires aligned faces? Thanks!
Author
Owner

@vlordier commented on GitHub (Sep 4, 2022):

Or, you might want to look into replacing it with StyleGANv3-R decoder as it is equivariant

StyleGAN3 transformation visualizations video

@vlordier commented on GitHub (Sep 4, 2022): Or, you might want to look into replacing it with StyleGANv3-R decoder as it is equivariant [StyleGAN3 transformation visualizations video](https://www.youtube.com/watch?v=SzVf1WHHrR4)
Sign in to join this conversation.
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: TencentARC/GFPGAN#173