Robust GAN inversion

TL;DR

This paper introduces a new GAN inversion method working in the native latent space $W$, using a learnable regularization strategy to improve reconstruction quality and efficiency, outperforming existing methods.

Contribution

We propose a novel GAN inversion approach in the native latent space $W$ with a learnable regularization, reducing distortion and computational cost while enhancing attribute hyperplane construction.

Findings

Achieves lowest distortion with 4x fewer parameters

Outperforms traditional methods in reconstruction quality

Shows slight improvement in attribute hyperplane quality

Abstract

Recent advancements in real image editing have been attributed to the exploration of Generative Adversarial Networks (GANs) latent space. However, the main challenge of this procedure is GAN inversion, which aims to map the image to the latent space accurately. Existing methods that work on extended latent space $W+$ are unable to achieve low distortion and high editability simultaneously. To address this issue, we propose an approach which works in native latent space $W$ and tunes the generator network to restore missing image details. We introduce a novel regularization strategy with learnable coefficients obtained by training randomized StyleGAN 2 model - WRanGAN. This method outperforms traditional approaches in terms of reconstruction quality and computational efficiency, achieving the lowest distortion with 4 times fewer parameters. Furthermore, we observe a slight improvement in the quality of constructing hyperplanes corresponding to binary image attributes. We demonstrate the effectiveness of our approach on two complex datasets: Flickr-Faces-HQ and LSUN Church.