基于可控域扭曲和有序抖动降噪的高性能云层湍流渲染

High-performance Cloud Rendering with Turbulence Based on Controllable Domain Warping and Denoising by Order Dithering

针对传统动态云层渲染算法存在湍流效果缺失与其效率不高的问题,提出了一种高性能且具有丰富湍流细节的动态云层渲染算法。该算法的核心思想是通过可控域扭曲的反Worley噪声的布朗分形运动的采样结果,应用到云层建模时的密度采样中,以模拟云层内部复杂的流体相互作用,使得渲染出的云层在流动时效果更富有真实感。此外,根据图像处理中常常使用到的基于贝尔矩阵的有序抖动技术,提出了一种特殊的降噪方式,该降噪方法在较少采样数下就可以得到采用其他方法需高采样数的效果。利用贝尔矩阵的特殊视觉效果大幅度减少了采样数,实现了更高性能的动态云层渲染。实验结果表明,云层渲染的真实感效果以及效率均有提升。

Referring to the problem of the lack of turbulence effects and the insufficient efficiency of traditional dynamic cloud rendering algorithms,this paper proposes a high-performance dynamic cloud rendering algorithm with rich turbulence details.The core idea of this algorithm is to apply the sampling results of the fractal Brownian Motion of anti-Worley noise through the controllable domain warping to the density sampling during cloud layer modeling to simulate the complex fluid interaction inside the cloud layer.As a result,the effect is more realistic when rendering the dynamic clouds.In addition,we propose a special noise reduction method based on the ordered dithering technique using Bayer matrix which is also often used in image processing.This image noise reduction method allows the effect of high samples to be obtained with fewer samples.The special visual effect of the Bayer matrix is used to greatly reduce the number of samples to achieve higher performance.The experimental results show that the realistic effect and efficiency of cloud rendering are both improved.