Optimization Methods for Real-time Volumetric Cloud Simulation

Addressing the issue of insufficient realism and real-time property of the existing volume cloud simulation, a multi-noise rendering method for simulation optimization is proposed. First of all, in terms of cloud modeling, the Perlin/Worley noise modeling is used to increase cloud diversity; then, Curl noise is also introduced to achieve cloud irregularity. Secondly, with respect to illumination of volume cloud, the dual Henyey-Greenstein phase function is selected for the approximate simulation of Mie scattering, thus overcoming such a shortcoming of the single Henyey-Greenstein phase function as later phase scattering while enhancing the realism and real-time efficiency of illumination. In the end, the improved Raymarching is adopted for rendering, with variable step size and early jump-out to enhance rendering efficiency. According to analysis of the experimental results, the method proposed herein can effectively simulate the effect of volume clouds and guarantee the real-time performance of the system.