EnVision-Research/Lift3D

stylegan2 generates pseudo-multiviews

xuao812500279 opened this issue · 1 comments

Hello, how do you generate pseudo-multiviews with stylegan2?

Len-Li commented

Hi,

In our implementation, we use disentangled latents to generate pseudo-multiview images. But I do not include this code in the repo because of Iossing dependency.

From my perspective, I would recommend you to use another type of 2D generator to generate pseudo-multiview images. For example, I found that recent work on diffusion model can greatly surpass stylegan, check out zero123plus and take a look at its related work section. There are many of awesome work in this line.

Best,
Leheng Li