Research On Real Time Fire Simulation Based On Particle System

The shape of most sceneries in nature, such as clouds, fire, ocean waves, rain and snow etc., changes with time past, and it is difficult to model and visualize these natural sceneries by using traditional methods. So the simulation of natural sceneries becomes a hot topic in the research field of computer graphics. Moreover, the simulation of fire is of a great challenge because the variability of the shape of fire in time and space.This thesis dedicate to the research on the simulation of fire. First, the definition and character of fire are introduced, and then a survey of some existing algorithms of fire simulation is presented. On the premise of visual realism of fire and real time implementation, an algorithm for simulating real time fire based on particle system is proposed, the fire generated by which can be applied to virtual scene in games, movies and TV programs.According to the appearance of fire, the positions of new particles are initialized with Gauss random numbers, thereby the density of fire particles satisfies Gauss distribution. In order to decrease the implementation time, Gauss random numbers are evaluated approximately by adding three random numbers. Based on the analysis of the motion of fire, those complex motion fields of fire are simplified to four simple motion fields. Thus the implementation velocity is enhanced, and real time simulation becomes possible. At the same time, dynamic motion fields are introduced. When motion fields move, they make the trajectories of fire particles time-variant. So the actions of fire particles become more active and the motion of fire becomes more realistic. Furthermore, we use surface particles instead of dot particles to render fire. Each surface particle can substitute many dot particles. So the implementation time is further decreased, and we also use texture map technique to enhance the render quality.Finally, we develop a demo system based on our algorithm and show the results generated by our simulation algorithm. Experimental results show that our algorithm can produce real time simulation of fire in consumer PC Platform and the effect of fire looks realistic.