WINE: Wavelet-Guided GAN Inversion and Editing for High-Fidelity Refinement

2022-10-18

AI Summary

High-frequency detail loss in GAN inversion leads to reconstruction artifacts, and existing methods struggle to balance fidelity and editability. This paper pioneers frequency-domain modeling for GAN inversion, proposing a wavelet-transform-based framework: (i) a wavelet coefficient loss explicitly constrains high-frequency reconstruction error; and (ii) a feature-level wavelet fusion mechanism enables cross-scale detail transfer. Evaluated on Face and CelebA-HQ datasets, our method achieves significant PSNR/SSIM improvements over state-of-the-art approaches. In editing tasks, it simultaneously enhances reconstruction quality and semantic controllability. The core contribution is the first differentiable wavelet modeling paradigm tailored for GAN inversion, establishing a novel direction for frequency-guided inverse problems in generative modeling.

Abstract

Recent advanced GAN inversion models aim to convey high-fidelity information from original images to generators through methods using generator tuning or high-dimensional feature learning. Despite these efforts, accurately reconstructing image-specific details remains as a challenge due to the inherent limitations both in terms of training and structural aspects, leading to a bias towards low-frequency information. In this paper, we look into the widely used pixel loss in GAN inversion, revealing its predominant focus on the reconstruction of low-frequency features. We then propose WINE, a Wavelet-guided GAN Inversion aNd Editing model, which transfers the high-frequency information through wavelet coefficients via newly proposed wavelet loss and wavelet fusion scheme. Notably, WINE is the first attempt to interpret GAN inversion in the frequency domain. Our experimental results showcase the precision of WINE in preserving high-frequency details and enhancing image quality. Even in editing scenarios, WINE outperforms existing state-of-the-art GAN inversion models with a fine balance between editability and reconstruction quality.

Problem

• GAN models
• image processing
• detail preservation

Innovation

• WINE model
• Frequency perspective
• Image quality enhancement