LatentSwap3D: Semantic Edits on 3D Image GANs

TL;DR

LatentSwap3D introduces a versatile semantic editing method for 3D GANs that identifies and swaps latent code attributes, enabling consistent and disentangled edits across various models and datasets.

Contribution

The paper presents LatentSwap3D, a novel latent space discovery approach for semantic editing in 3D GANs applicable to any model and dataset, outperforming existing methods.

Findings

Provides consistent semantic edits in 3D GANs

Outperforms existing editing methods qualitatively and quantitatively

Works across multiple 3D GAN architectures and datasets

Abstract

3D GANs have the ability to generate latent codes for entire 3D volumes rather than only 2D images. These models offer desirable features like high-quality geometry and multi-view consistency, but, unlike their 2D counterparts, complex semantic image editing tasks for 3D GANs have only been partially explored. To address this problem, we propose LatentSwap3D, a semantic edit approach based on latent space discovery that can be used with any off-the-shelf 3D or 2D GAN model and on any dataset. LatentSwap3D relies on identifying the latent code dimensions corresponding to specific attributes by feature ranking using a random forest classifier. It then performs the edit by swapping the selected dimensions of the image being edited with the ones from an automatically selected reference image. Compared to other latent space control-based edit methods, which were mainly designed for 2D GANs, our method on 3D GANs provides remarkably consistent semantic edits in a disentangled manner and outperforms others both qualitatively and quantitatively. We show results on seven 3D GANs (pi-GAN, GIRAFFE, StyleSDF, MVCGAN, EG3D, StyleNeRF, and VolumeGAN) and on five datasets (FFHQ, AFHQ, Cats, MetFaces, and CompCars).