In-N-Out: Faithful 3D GAN Inversion with Volumetric Decomposition for Face Editing

arXiv:2302.04871v4 Announce Type: replace
Abstract: 3D-aware GANs offer new capabilities for view synthesis while preserving the editing functionalities of their 2D counterparts. GAN inversion is a crucial step that seeks the latent code to reconstruct input images or videos, subsequently enabling diverse editing tasks through manipulation of this latent code. However, a model pre-trained on a particular dataset (e.g., FFHQ) often has difficulty reconstructing images with out-of-distribution (OOD) objects such as faces with heavy make-up or occluding objects. We address …

abstract arxiv capabilities code cs.cv diverse editing enabling face gan gans images manipulation synthesis tasks through type videos view