.3 D Model Of Non-photorealistic Rendering Effect

Non-photorealistic rendering (Non-Photorealistic Rendering, NPR) is a research field combined pictorial arts with computer technology. It focuses on the abstraction and process of real scene aiming at highlighting the important information and making a more intuitive, more profound impression. So it has a wide range of applications in education, entertainment, arts and other fields. Although most researches of non-photorealistic rendering focus on the image, the ones focus on the 3D objects are relative scarce. In 3D NPR, extraction and stylization of the silhouettes and the models' rendering are key points. The paper carries out the research based on these key points which can be summarized as:(1)Silhouettes'highlight and renderingSilhouettes detection of the 3D models is the fundamental problem of non-photorealistic rendering. Silhouettes represent the basic shape of the models and are necessary in perception and aesthetic. Its representation directly relates to the performance of the non-photorealistic rendering. This paper describes an algorithm that is based on a sharpening filter to extract the feature lines of 3D models. We use a pre-processing which makes the models'features stands out. The use of a simple contour detection algorithm can get more feature lines based on the idea of sharpening. In addition, we extract the suggestive contour which shows the models'shape more vivid.(2) Silhouettes stylizationSilhouette stylization is an important form of artistic expression in NPR. First, the paper stylized the silhouettes by taking advantage of mesh sharpening which generate more feature lines to enhanced expression. Then a new technique of silhouette stylization which is less computational was described. We extrude both feature lines along their normal vectors with the goal of giving breadth to the stroke. Therefore, an edge is extended to be a quad. After constructing the quad, we can build a trivial one-to-one correspondence between the quad and a 2D texture and then map the 2D texture to the quad directly. It applies a style to silhouettes successfully. Experiment finds that this algorithm maintains continuity between the feature lines.(3)3D models'rendering First, this paper describes an improvement on the cartoon rendering. It improves the traditional cartoon rendering by introduced the curvature information. Depict the models'shape by shading instead of drawing lines. It renders the 3D models by using the mean curvature as the 1D texture coordinate. This method can depict the outline and interior features of the models.Second, the paper describes an improved rendering method based the radical curvature. Existing algorithms just have limited rendering effects and can not display the internal character of the 3D models. The view-dependent radical curvature is introduced to improve the X-Toon shading. It introduces a shading method based on the radical curvature. With the appropriate choice of 2D texture, it can mimic more material effects, e.g. gouache, mental and wax. Besides, it can depict more details of the models too.These improved methods have important use for reference in film and television animation, three-dimensional animation modeling, and publications.