![[CVPR2022] FENeRF: Face Editing in Neural Radiance Fields](/upload/2022/02/75961171_p4-566cd9b9c8bb4a57bc9044bd70de775e.jpg)
[CVPR2022] FENeRF: Face Editing in Neural Radiance Fields
Introduction关于肖像生成的方法,目前主要有两种,2D GANs和3D-aware GANs,两者各有擅长的方面。2D GAN可以生成高分辨率图像并具有很高的可编辑性,但是生成不同视图时的一致性很差,3D-aware GAN则相反,能生成一致性很高的视图,但是视图的分辨率和可编辑性不高该文

DF-GAN: A Simple and Effective Baseline for Text-to-Image Synthesis
|
A Simple and Effective Baseline for Text-to-Image Synthesis
![[ECCV2022] CCPL: Contrastive Coherence Preserving Loss for Versatile Style Transfer](/upload/2022/10/src=http___i0.hdslb.com_bfs_article_fdc883288f102748024337d8a33e8520b5d7fc64.jpg&refer=http___i0.hdslb.webp)
[ECCV2022] CCPL: Contrastive Coherence Preserving Loss for Versatile Style Transfer
|
一种通用的风格转移方法,可以实现艺术、逼真的风格迁移,以及实现视频的风格迁移,同时在训练过程中并没有视频的参与

Demystifying Neural Style Transfer
|
Style transfer is essentially a domain adaptation problem, which aligns the feature distributions.
![[arXiv] Pivotal Tuning for Latent-based Editing of Real Images](/upload/2022/06/src=http___pic1.win4000.com_wallpaper_2019-04-15_5cb43701cb8c3.jpg&refer=http___pic1.win4000.webp)
[arXiv] Pivotal Tuning for Latent-based Editing of Real Images
基于优化的,更新StyleGAN参数的反演方法