An Effective Method For The Generation Of Non-photorealistic Animation

Non-Photorealistic Rendering and Animation(NPRA) is a newly developed research area in computer graphics and, in some cases, NPRA is more attractive than photorealistic rendering and animation. Thus, it has a wide range of applications in the fields ranging from education, entertainment to art.As a unique animation style, more attention is paid to non-photorealistic animation because of its unique expression style, but a great deal of time and skills are required. Generally, accurate 3D models and illumination models are not requested in non-photorealistic animation, so, some researchers hope to help the ordinary users to create simple NPA conveniently. Based on this requirement, steerable filters are introduced: A pair of even and odd filtered images is generated by convoluting a quadrature pair of steerable filters with a source image, and then through continuously varying their phases over time, it gives the vision of an object in motion without changing its position. Animation is conveniently created in this way. Because of the discrepancy of background color between the even and odd filtered images, the animation has the problem of background flicker. To eliminate this problem, we separate the foreground and background through threshold segmentation, and apply this method to the foreground image only.In addition, we expand this method to colorful image. The channel L of Lαβcolor space denotes neutral channel, compared to the traditional color channels, Lαβreduces relativity between color channels, i.e. color channels have the least relativity from each other. Finally, we get the colorful NPA effect through color conversion between the color space of Lαβand RGB.Finally, we expand this algorithm to 3-D space. 3-D steerable filters generated in Matlab are called in VC, and then steerable filters are applied to 3-D models, 3-D non-photorealistic animation is effectively generated in this way.