search

MrTornado24 / Next3D

[CVPR 2023 Highlight] Next3D: Generative Neural Texture Rasterization for 3D-Aware Head Avatars
https://mrtornado24.github.io/Next3D/
470 stars 29 forks source link

About re-render the template mesh for Discriminator #6

Closed Talegqz closed 1 year ago

Talegqz commented 1 year ago

Thanks for your excellent work.

In the paper, you said that "we re-render the template mesh under the rendered pose to get the synthetic rendering Isynthetic and feed it into Ddual along with image pairs".

However, I am having difficulty locating this specific process within the repository. Could you please provide more information about how this process is implemented or direct me to a resource that explains it more clearly?

Thanks for your help!

MrTornado24 commented 1 year ago

Hi, as we mentioned in README.md, re-rendering will slow down the training speed so we currently don't release the part. We are trying to re-implement it faster.

Talegqz commented 1 year ago

I see. Thanks for your reply.