search

shrubb / latent-pose-reenactment

The authors' implementation of the "Neural Head Reenactment with Latent Pose Descriptors" (CVPR 2020) paper.
https://shrubb.github.io/research/latent-pose-reenactment/
Apache License 2.0
181 stars 34 forks source link

About Graph #26

Closed loveFaying closed 3 years ago

loveFaying commented 3 years ago

Thank you for your contribution，What drawing software do you use？，I strongly think your structure diagram is very beautiful。thank you :)

shrubb commented 3 years ago

:smiling_face_with_three_hearts: Thanks! I use MS PowerPoint for such diagrams.

loveFaying commented 3 years ago

Good，So beautiful！I have another question，， I want to achieve an effect：The lighting of the result video follows the lighting of the input video. The purpose is to make the model produce different lighting effects. But I failed. Could it be that the model does not allow this?

shrubb commented 3 years ago

Yes, the model intentionally doesn't allow this. That was one of the main points of our paper.

If you want to allow lighting from driving images to influence the result, try switching off some of the lighting augmentations during training. No guarantee that it will work though, your goal is very ambitious.

loveFaying commented 3 years ago

Thank you for your reply and wish you success in your work