透過您的圖書館登入
IP:3.15.25.32

並列摘要


NPR (Non Photo-Realistic Rendering) is an important issue in Computer Graphics. It presents a totally different style from those using realistic methods. In this paper we take advantage of a 3D NPR system in color presentation, and apply stroke merging and sketch-based modeling to synthesize NPR images with volumetric effects, such as oil painting, bas-relief sculpture, and clay relief. To represent these volumetric visual effects on canvas we adopt a mesh generation method which generates the thickness of strokes. By controlling the ratio of height and width to the base area, the surface curvature is adjusted to reflect the shape differences between impasto oil painting and clay relief. Then a mesh smoothing method is applied to each group of strokes. In addition, users can personalize their own color palette by using keywords such as the name of the artist, or by choosing color sets through our GUI which categorizes color space into an emotional map. Users choose by using keywords, then related images are gathered from online resources using web-mining technology, and the appropriate colors are extracted to construct a color chart by analyzing the images. Then, the color chart is used to control the distribution of colors on the canvas. As shown in the experiment results, our 3D NPR system creates interesting and vivid images of oil painting and clay relief works. Finally, we integrate our volumetric NPR image with a relief mapping technique to provide real-time interaction.

延伸閱讀