One-for-All: Towards Universal Domain Translation With a Single StyleGAN

In this paper, we propose a novel translation model, UniTranslator, for transforming representations between visually distinct domains under conditions of limited training data and significant visual differences. The main idea behind our approach is leveraging the domain-neutral capabilities of CLIP as a bridging mechanism, while utilizing a separate module to extract abstract, domain-agnostic semantics from the embeddings of both the source and target realms. Fusing these abstract semantics with target-specific semantics results in a transformed embedding within the CLIP space. To bridge the gap between the disparate worlds of CLIP and StyleGAN, we introduce a new non-linear mapper, the CLIP2P mapper. Utilizing CLIP embeddings, this module is tailored to approximate the latent distribution in the StyleGAN's latent space, effectively acting as a connector between these two spaces. The proposed UniTranslator is versatile and capable of performing various tasks, including style mixing, stylization, and translations, even in visually challenging scenarios across different visual domains. Notably, UniTranslator generates high-quality translations that showcase domain relevance, diversity, and improved image quality. UniTranslator surpasses the performance of existing general-purpose models and performs well against specialized models in representative tasks.

I. Introduction

Recent generative models are developed with a growing emphasis on universality, aiming to enhance the real-world applicability in solving complex challenges [1], [2], [3]. Significant advances have been made in visual domain translation [4], [5], [6], [7], [8], which harnesses the transformation of images by exploiting inherent content correlations across disparate realms. The defining feature of universality for domain translators is their ability to seamlessly convert images from any real-world source domain to a chosen target domain. This pursuit of universality in visual domain translation erodes the barriers segregating different domains and provides invaluable technological support for a wide range of applications. These span from artistic creations such as anthropomorphic or skeuomorphic designs to the entertainment industry, including customized effect generation on various platforms.