Fluid Animation in Chinese Painting

• Main Authors: Hsing-yu Fu, 傅星諭
• Other Authors: Chih-Yuan Yao
• Format: Others
• Language: zh-TW
• Published: 2012
• Online Access: http://ndltd.ncl.edu.tw/handle/09374200129625367569

碩士 === 國立臺灣科技大學 === 資訊工程系 === 101 === Rivers and falls are important elements in Chinese ink painting. Rivers and falls are painted by strokes to sketch the lines of the water flow and ink dispersion to express the spray of the flow surface in order to present the sense of smooth waterlines and the characteristics of coherent diffusion. This thesis uses the Navier-Stoke equation to simulate the water flow in a river and fall to create physically correct flow fields. At the same time, it also applies a tensor field smoothing method to the constructed flow fields in order to reduce the possible rendering artifacts when multiple flow patterns are specified. In the last, the constructed field would be animated as Chinese ink painting with smooth and time coherence. Our system provides an intuitive user interface to let the painter construct their virtual world with mountains, plains, rocks, rivers and falls. Then, the painter can specify the flow pattern in the rivers and falls, and the system can compute the physically correct flow field in the rivers and falls according to the boundaries, obstacles and user-specified fields. The physically correct flow field is used to construct the ink strokes with our smooth field stroke algorithm. Our system can easily generate view consistent and temporally coherent 3D animated rivers and falls in a Chinese ink painting style. The main contributions of our system lies in the integration of the Navier-Stoke equation and tensor field smoothing to generate a Chinese-ink-painting river and fall animation. Furthermore, we propose an interactive system that can quickly create a Chinese ink painting world containing the animated rivers and falls with the water flow drawn with spatial and temporal coherence. The overall animation is pleasant without the flickering artifacts commonly existing in a stroke-based non-photorealistic rendering (NPR) animation.