为了解决传统的基于光辐射传输预计算(PRT)的全局光照技术在大规模场景下渲染效率低下的问题,在对PRT光照算法的本质进行分析的基础上,提出了一种基于四叉树的自适应网格细分算法。根据光照方程中的可见性函数,只在需要产生阴影的地方进行网格细分,并利用四叉树来重组模型数据,避免了传统的网格均匀细分方法导致的模型复杂度大幅增加,从而会影响绘制效率的问题。通过在虚拟机场仿真系统中的实际应用,对两种算法进行了比较,表明了该算法在效果和效率上均优于原有算法。
To solve the rendering inefficiencies problem of the traditional global illumination techniques based on precomputed radiance transfer(PRT) in large-scale scene,with analysis the nature of PRT illumination algorithm,a new algorithm is presented which used quadtree algorithm for adaptive mesh subdivision.According to the visibility function in the light equation,only place grid subdivision where need to cast a shadow on,and use quadtree to reorganize the model data,avoiding the rendering efficiency problem of substantial increase in scene complexity which results from the traditional uniform model grid subdivision.Through the application in virtual airport simulation system,indicated that this algorithm is better than the traditional algorithms both in effect and efficiency.