Towards Photographic Image Manipulation with Balanced Growing of Generative Autoencoders

TL;DR

This paper introduces an improved generative autoencoder that achieves fast encoding, high-quality high-resolution images, and a well-structured latent space for semantic manipulation, advancing face image editing technology.

Contribution

It significantly enhances the PIONEER autoencoder model by altering training dynamics, leading to better image quality, identity preservation, and disentangled latent representations.

Findings

Improved face identity conservation on CelebAHQ.

State-of-the-art latent space disentanglement.

Enhanced performance on LSUN Bedrooms dataset.

Abstract

We present a generative autoencoder that provides fast encoding, faithful reconstructions (eg. retaining the identity of a face), sharp generated/reconstructed samples in high resolutions, and a well-structured latent space that supports semantic manipulation of the inputs. There are no current autoencoder or GAN models that satisfactorily achieve all of these. We build on the progressively growing autoencoder model PIONEER, for which we completely alter the training dynamics based on a careful analysis of recently introduced normalization schemes. We show significantly improved visual and quantitative results for face identity conservation in CelebAHQ. Our model achieves state-of-the-art disentanglement of latent space, both quantitatively and via realistic image attribute manipulations. On the LSUN Bedrooms dataset, we improve the disentanglement performance of the vanilla PIONEER, despite having a simpler model. Overall, our results indicate that the PIONEER networks provide a way towards photorealistic face manipulation.