Cross-Domain Style Mixing for Face Cartoonization

TL;DR

This paper introduces Cross-domain Style mixing, a novel face cartoonization method that combines latent codes from different domains, enabling stylization with minimal training data and supporting various abstraction levels.

Contribution

It proposes a new style mixing approach that stylizes faces into cartoons using a single generator and limited training images, overcoming previous constraints.

Findings

Effective stylization across multiple cartoon styles

Requires only a limited number of training images

Supports various levels of face abstraction

Abstract

Cartoon domain has recently gained increasing popularity. Previous studies have attempted quality portrait stylization into the cartoon domain; however, this poses a great challenge since they have not properly addressed the critical constraints, such as requiring a large number of training images or the lack of support for abstract cartoon faces. Recently, a layer swapping method has been used for stylization requiring only a limited number of training images; however, its use cases are still narrow as it inherits the remaining issues. In this paper, we propose a novel method called Cross-domain Style mixing, which combines two latent codes from two different domains. Our method effectively stylizes faces into multiple cartoon characters at various face abstraction levels using only a single generator without even using a large number of training images.