Automatic generation of large scale 3D cloud based on weather forecast data

Abstract

3D cloud scenes generation is widely used in computer graphics and virtual reality. Most of the existing methods for 3D cloud visualization first model the cloud based on the physical mechanism of cloud, and then solve the illumination model of the cloud to generate 3D scenes. However, this kind of methods cannot show the real weather condition. Moreover, the existing cloud visualization methods based on the weather forecast data cannot be applied to the large scale 3D cloud scenes due to the complicated solution of the illumination model. Borrowing the idea of particle system, this paper proposes an algorithm for automatic generation of large scale 3D cloud based on weather forecast data. The algorithm considers each grid point in the data as a particle, whose optical parameters can be determined by the input data. Multiple forward scattering is used to calculate the incident color of each particle, and the first order scattering is utilized to determine the incident color to the observer. Experimental results demonstrate that our algorithm could not only generate realistic 3D cloud scenes from the weather forecast data, but also obtain an interactive frame rates for the data that contains millions of grids.