Abstract

This paper presents an innovative solution for 3D character model texture mapping based on artificial intelligence (AI) image generation models. This study explored the potential of AI-generated textures to enhance the efficiency and effectiveness of the traditional manual workflow. By leveraging state-of-the-art AI image generation techniques, we proposed a new workflow that could address existing challenges in the current texture mapping process. This research began with an introduction to AI image generation models with a review of the current status and applications of these models in texture map generation. We then identified and analyzed issues associated with the existing character model texture workflow, highlighting inefficiencies and areas for improvement. Our proposed AI-assisted workflow aimed to streamline the texture creation process, significantly reducing the time required while maintaining high-quality outputs. To validate the new workflow, we conducted a series of experiments with professional 3D artists, yielding clear and significant results. The new AI-assisted workflow significantly reduced texture creation time by an average of 76% to 89% compared to traditional manual methods, demonstrating its potential to drastically improve efficiency. Qualitative feedback from artists highlighted that the new workflow not only streamlined the process, but also made it more user-friendly and accessible. However, some limitations were identified, including the need for further refinement of AI-generated textures to achieve a quality level comparable to those created by highly skilled artists.