Spatially-Adaptive Multilayer Selection for GAN Inversion and Editing

TL;DR

This paper introduces a spatially-adaptive multilayer selection method for GAN inversion and editing, improving results on complex images by predicting segment invertibility and assigning segments to appropriate latent layers.

Contribution

It proposes a novel approach that adaptively inverts different image regions into suitable GAN latent layers based on their difficulty, enhancing inversion quality for complex scenes.

Findings

Outperforms recent methods on complex categories

Maintains downstream editability

Improves inversion quality for occluded and scene-rich images

Abstract

Existing GAN inversion and editing methods work well for aligned objects with a clean background, such as portraits and animal faces, but often struggle for more difficult categories with complex scene layouts and object occlusions, such as cars, animals, and outdoor images. We propose a new method to invert and edit such complex images in the latent space of GANs, such as StyleGAN2. Our key idea is to explore inversion with a collection of layers, spatially adapting the inversion process to the difficulty of the image. We learn to predict the "invertibility" of different image segments and project each segment into a latent layer. Easier regions can be inverted into an earlier layer in the generator's latent space, while more challenging regions can be inverted into a later feature space. Experiments show that our method obtains better inversion results compared to the recent approaches on complex categories, while maintaining downstream editability. Please refer to our project page at https://www.cs.cmu.edu/~SAMInversion.