Giters

gwang-kim / DATID-3D

[CVPR 2023] Official implementation of "DATID-3D: Diversity-Preserved Domain Adaptation Using Text-to-Image Diffusion for 3D Generative Model"

Home Page:https://gwang-kim.github.io/datid_3d/

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Questions with 3D GAN inversion method

binarydaddy opened this issue · comments

commented

Hello,
First of all, congrats on CVPR 2023 and thanks for sharing such wonderful work!
I had some unclear parts with 3D GAN inversion method you have described on the paper. (How you generated results in Figure 8.)

Is this "instance-selected domain adaptation combined with 3D GAN inversion" method simply inverting the modified image to EG3D W space and reconstructing images from it?

Or is the statement closer to JoJoGAN, where EG3D fine-tuning is included?

Thank you!