Keywords: face generation, generative networks
TL;DR: Reproduction study of LiftedGAN, which disentangles and lifts a pre-trained StyleGAN2 for 3D-aware face generation.
Abstract: In this study, we present our results and experience during replicating the paper titled "Lifting 2D StyleGAN for 3D-Aware Face Generation". This work proposes a model, called LiftedGAN, that disentangles the latent space of StyleGAN2 into texture, shape, viewpoint, lighting components and utilizes those components to render novel synthetic images. This approach claims to enable the ability of manipulating viewpoint and lighting components separately without altering other features of the image. We have trained the proposed model in PyTorch, and have conducted all experiments presented in the original work. Thereafter, we have written the evaluation code from scratch. Our re-implementation enables us to better compare different models inferring on the same latent vector input. We were able to reproduce most of the results presented in the original paper both qualitatively and quantitatively.
Paper Url: https://openaccess.thecvf.com/content/CVPR2021/papers/Shi_Lifting_2D_StyleGAN_for_3D-Aware_Face_Generation_CVPR_2021_paper.pdf
Paper Venue: CVPR 2021
Supplementary Material: zip