基于GPU的实时毛皮渲染及优化

GPU-based real-time fur rendering and optimization

传统毛皮材质模拟需要使用大量细长面片,在实时渲染中会占用过多的绘制资源,造成性能瓶颈。为了解决场景中大量毛皮材质模型的实时渲染性能问题,文中提出一种基于多层纹理渲染实现的实时毛皮渲染方法,通过逐层不透明度衰减的方式模拟多层毛发切片叠加,从而快速、高效地模拟出大量短小毛发。根据毛发的光照特性,同时结合线性兰伯特光照算法、环境光遮蔽算法、菲涅尔近似轮廓光算法、各向异性高光算法等多种光照算法,经过伽马校正后模拟出具有真实感的毛皮光照效果,并提出通过修正不透明度衰减公式减少渲染层数的优化方案,减少绘制开销,使性能与渲染效果达到相对平衡。优化后,文中方法能实时处理一定规模的真实感毛皮渲染,对于毛皮材质在游戏、动画、虚拟现实等领域的应用具有重要意义。

Traditional fur material simulation uses plenty of slender mesh to simulate fur,which occupies too much rendering resource in real-time rendering and causes performance bottleneck. In order to solve the real-time rendering performance problem of a large number of fur material models in the scene,a real-time fur rendering method based on multilayer texture rendering is proposed,which can simulate multi-layer fur slice superposition by means of layer-by-layer opacity attenuation,so as to quickly and efficiently simulate a large number of short fur. after the gamma correction,the realistic fur lighting effect is simulated according to the lighting characteristics of fur,and combined with linear Lambert lighting algorithm,ambient light masking algorithm,Fresnel approximate contour lighting algorithm,anisotropic highlight algorithm and other lighting algorithms. An optimization scheme to reduce the number of rendering layers by modifying the opacity attenuation formula is proposed to reduce the rendering overhead and achieve the relative balance between performance and rendering effect.After optimization,the method in this paper can process a certain scale of realistic fur rendering in real time,which is of great significance for the application of fur materials in games,animation,virtual reality and other fields.