Image processing and ink diffusion simulation algorithms for digital painting on heterogeneous physical media

This article focuses on the research of image processing and ink diffusion simulation algorithms for digital painting on heterogeneous physical media. The purpose of the research is to enhance the ability of digital painting systems to reproduce the painting effects on surfaces of different materials. A digital painting simulation framework with material perception capability was constructed by introducing deep learning based methods for material recognition, color mapping, and texture synthesis, combined with physical modeling and generative adversarial network (GAN) driven ink diffusion simulation algorithm. The experimental part used a dataset containing multiple real media samples to evaluate the accuracy of material classification (up to 90.2%), color mapping error (average Δ E value less than 3), texture generation quality (SSIM of 0.91), and ink diffusion simulation accuracy (intersection to union ratio IoU of 87.5%). The results indicate that the proposed method outperforms existing simulation methods in terms of visual consistency and physical realism, and can effectively support high-quality digital painting applications on various media such as paper, fabric, and wood.