We have witnessed rapid progress on 3D-aware image synthesis, leveraging
recent advances in generative visual models and neural rendering. Existing
approaches however fall short in two ways: first, they may lack an underlying
3D representation or rely on view-inconsistent rendering, hence synthesizing
images that are not multi-view consistent; second, they often depend upon
representation network architectures that are not expressive enough, and their
results thus lack in image quality. We propose a novel generative model, named
Periodic Implicit Generative Adversarial Networks ($\pi$-GAN or pi-GAN), for
high-quality 3D-aware image synthesis. $\pi$-GAN leverages neural
representations with periodic activation functions and volumetric rendering to
represent scenes as view-consistent 3D representations with fine detail. The
proposed approach obtains state-of-the-art results for 3D-aware image synthesis
with multiple real and synthetic datasets.
%0 Generic
%1 chan2020pigan
%A Chan, Eric R.
%A Monteiro, Marco
%A Kellnhofer, Petr
%A Wu, Jiajun
%A Wetzstein, Gordon
%D 2020
%K GAN generation view_sythesis
%T pi-GAN: Periodic Implicit Generative Adversarial Networks for 3D-Aware
Image Synthesis
%U http://arxiv.org/abs/2012.00926
%X We have witnessed rapid progress on 3D-aware image synthesis, leveraging
recent advances in generative visual models and neural rendering. Existing
approaches however fall short in two ways: first, they may lack an underlying
3D representation or rely on view-inconsistent rendering, hence synthesizing
images that are not multi-view consistent; second, they often depend upon
representation network architectures that are not expressive enough, and their
results thus lack in image quality. We propose a novel generative model, named
Periodic Implicit Generative Adversarial Networks ($\pi$-GAN or pi-GAN), for
high-quality 3D-aware image synthesis. $\pi$-GAN leverages neural
representations with periodic activation functions and volumetric rendering to
represent scenes as view-consistent 3D representations with fine detail. The
proposed approach obtains state-of-the-art results for 3D-aware image synthesis
with multiple real and synthetic datasets.
@misc{chan2020pigan,
abstract = {We have witnessed rapid progress on 3D-aware image synthesis, leveraging
recent advances in generative visual models and neural rendering. Existing
approaches however fall short in two ways: first, they may lack an underlying
3D representation or rely on view-inconsistent rendering, hence synthesizing
images that are not multi-view consistent; second, they often depend upon
representation network architectures that are not expressive enough, and their
results thus lack in image quality. We propose a novel generative model, named
Periodic Implicit Generative Adversarial Networks ($\pi$-GAN or pi-GAN), for
high-quality 3D-aware image synthesis. $\pi$-GAN leverages neural
representations with periodic activation functions and volumetric rendering to
represent scenes as view-consistent 3D representations with fine detail. The
proposed approach obtains state-of-the-art results for 3D-aware image synthesis
with multiple real and synthetic datasets.},
added-at = {2021-08-13T13:57:21.000+0200},
author = {Chan, Eric R. and Monteiro, Marco and Kellnhofer, Petr and Wu, Jiajun and Wetzstein, Gordon},
biburl = {https://www.bibsonomy.org/bibtex/2e2ed972d9ba484037168709dd479a6ed/shuncheng.wu},
description = {https://arxiv.org/pdf/2012.00926.pdf
Project page:
https://marcoamonteiro.github.io/pi-GAN-website/
Code:
https://github.com/marcoamonteiro/pi-GAN},
interhash = {920abed8a15080074b1ba8c361d6ed07},
intrahash = {e2ed972d9ba484037168709dd479a6ed},
keywords = {GAN generation view_sythesis},
note = {cite arxiv:2012.00926},
timestamp = {2021-08-13T13:58:19.000+0200},
title = {pi-GAN: Periodic Implicit Generative Adversarial Networks for 3D-Aware
Image Synthesis},
url = {http://arxiv.org/abs/2012.00926},
year = 2020
}