针对复杂3D烟花模拟方法存在计算量大,效率不高和逼真性不够等问题,提出了一种基于GPU粒子系统的烟花模拟方法。采用三维网格模拟烟花形状,通过扩展双面深度剥离算法在网格上快速均匀采样,提高顶点信息收集效率;提出并行迭代聚类算法,实现多层次的烟花爆炸,达到烟花的多重性效果;应用逆动力学原理控制烟花中粒子的运动,给出形状约束粒子的速度公式。实验结果表明,该算法对烟花模拟效果在真实性和速度上都有所提高。
Due to the problems such as timeconsuming calculation, lack of verisimilitude, etc. in existing fireworks simulation method, a new simulation technique of fireworks based on GPU particle system is proposed. The shape of fireworks is depicted through 3D meshes and fast point sampling is completed by extending the dual depth peeling algorithm to improve the collection efficiency of vertex information. The highly parallel iterative clustering algorithm is used to support multilevel fireworks explo sion, which makes fireworks simulation as natural as possible. Inverse dynamics is applied to control the motion of particles in fireworks. Experimental results show that the method enhances the verisimilitude and speed of the simulation.