Interpretable Screentone Representation

Interpretable screentone representation focuses on developing computational methods for efficiently and accurately manipulating the distinctive patterned shading (screentones) used in manga. Current research emphasizes creating models that can automatically generate, modify, and adapt screentones, often employing deep learning techniques like diffusion models and incorporating interpretable representations to disentangle intensity and type features for better control. This work aims to automate time-consuming tasks for manga artists, such as rescreening and retargeting, improving the efficiency and quality of manga creation and adaptation across different resolutions and styles.