LSC-GAN: Latent Style Code Modeling for Continuous Image-to-image Translation

TL;DR

This paper introduces LSC-GAN, a novel approach for continuous image-to-image translation that models latent style codes for domain intervals and employs editing modules for smooth, diverse, and high-quality image transformations across continuous domains.

Contribution

The paper proposes a new method that explicitly models latent style codes for continuous I2I translation and introduces editing modules for smooth domain transitions and diverse results.

Findings

Achieves high-quality age and viewing angle translation.

Provides flexible and diverse image synthesis.

Outperforms existing methods in continuous domain translation.

Abstract

Image-to-image (I2I) translation is usually carried out among discrete domains. However, image domains, often corresponding to a physical value, are usually continuous. In other words, images gradually change with the value, and there exists no obvious gap between different domains. This paper intends to build the model for I2I translation among continuous varying domains. We first divide the whole domain coverage into discrete intervals, and explicitly model the latent style code for the center of each interval. To deal with continuous translation, we design the editing modules, changing the latent style code along two directions. These editing modules help to constrain the codes for domain centers during training, so that the model can better understand the relation among them. To have diverse results, the latent style code is further diversified with either the random noise or features from the reference image, giving the individual style code to the decoder for label-based or reference-based synthesis. Extensive experiments on age and viewing angle translation show that the proposed method can achieve high-quality results, and it is also flexible for users.