ScholarWorks@중앙대학교

초록

Webtoon-style portrait stylization has gained prominence across various fields, including social media, virtual characters, and the Metaverse, with significant applications in demo booths for real-time audience interaction. However, achieving high-quality, real-time stylizations presents challenges, such as preserving facial identity while maintaining the webtoon aesthetics and ensuring robustness to diverse inputs. This study explores diffusion-based generative models for webtoon-style portrait stylization, focusing on real-world scenarios such as demo booths. We address challenges such as artifact reduction and adaptability to varying inputs, showcasing the potential of diffusion models to enhance interactive and artistic applications, particularly in demo booth settings. © 2013 IEEE.

키워드