NPR And Its Practice:A Psychological Perspective

From the trend in computer science,we can get the impression that harmony between human and machine is its future objective. Computer graphics field, playing an important role in Human-Computer Interaction, can learn many things from such an objective. Among them one is that to synthesize realistic images practitioners in CG should pay more and more attention to the human side, it's vision and perception ,and not just appeal to impersonal physical or mathematical models. The emergence and development of Nonphotorealistic Rendering(NPR) is the result of such quest.In the past years, many research work have been done to explain how human perceive shapes and images. This paper means to introduce some perceptual work into the process of producing line drawings of 3D models.External contours have significance in distinguishing a object from its background.We also believe some inner geometrical properties of the object can help observer master its shape.Concave is such a property and in a change detection task, We can find a dramatic advantage for concave compared to convex changes of equal magnitude. Based on aforemetioned knowledge, An algorithm was designed to generate line-style drawings of 3D triangle meshes. At preprocessing stage, mean curvatures of vertices were estimated. At runtime, through interaction with user, concave regions under consideration were detected and skeletons of them were extracted using a heuristic searching method. Combining skeletons with contours and adding illumination effects shapes could be illustrated with a hand-made appearance. Experimental results validate the algorithm.Chapter one reviews the history of traditional computer graphics and also the challenge it faces. Our focus is to justify the appearance of Nonphotorealistic Rendering, its unique adavantages over photorealistic rendering in some applications. Finally we reemphasize the thesis' contribution.Chapter two surveys the progress about artistic rendering techniques of 3D models. Contour extraction and the visibility problem are in more detail.Chapter three proposes the necessity of developing a theory of psychology within non-photorealistic rendering.Chapter four describes our alogrithm for line drawings of 3D models in detail. In the end, we make comments on the alogrithm.Chapter five exposes the implementation of our algorithm. Lines of core codes are discussed.At last, we summarize our work and discuss future perspectives.