Research on simulation and optimization algorithm of watercolor paint mixing effect based on graph neural network

Abstract

With the rapid development of computer graphics and artificial intelligence, the simulation and optimization problems in digital art and design have gradually attracted widespread attention, especially the simulation of pigment mixing effect is of great significance in digital painting and artistic creation. As a unique painting medium, the mixing effect of watercolor pigments presents complex physical and chemical characteristics, which makes the accurate simulation face great challenges. In this paper, a simulation and optimization algorithm of watercolor paint mixing effect based on graph neural network is proposed. By constructing a graph model of pigment mixing, the algorithm uses graph neural network to capture the local and global information in the process of pigment mixing, and realizes the accurate simulation of different pigment combinations. In this paper, several watercolor paint data of different brands, colors and transparencies are used for experiments. The experimental results show that compared with traditional physical simulation methods, the proposed algorithm has significant advantages in visual effect and computational efficiency. During the simulation, the color difference ΔE value of the pigment mixing was reduced by an average of 15.3%, and the calculation time was reduced by about 40%. Furthermore, a user-interactive pigment recommendation system based on reverse mapping is integrated, enabling efficient suggestions for pigment combinations that match a target color.