Application Scenario Analysis of Styleshot Based Game Style Generation

School of Computer Science and Technology, East China Normal University, Shanghai, China

10245102426@stu.ecnu.sdu.cn

Abstract

The continuous improvement of hardware and technology has made a huge leap in the development of games, and the beauty and variety of game graphics have also gone to the next level. However, as the volume of games increases, there is a need for a project that can migrate different game styles to the characters or scenes that we want to generate, so as to ensure that the generated images fit the original styles and increase the efficiency of the work. The aim of this work is to select the most suitable project for migrating styles in game development by comparing the advantage zones of various style migration projects. With the help of comparative experiments, literature reading, and dataset analysis, it is concluded that the style migration project named StyleShot can meet the requirements well with its style-aware encoder, content fusion encoder, and a well-organized style dataset named StyleGallery, which improves the expressiveness and generalization of the style representation.