Real Time Dense Smoke Simulation Based Particle System

This thesis presents a new algorithm for simulating smoke animation based on the particle system and physical dynamics principle. To accelerate the collision detection between particles and scene objects, we propose to build the binary space partitioning tree for scene objects as a pretreatment. During the animation of particles, they are clustered on-the-fly as particle cluster trees. The binary space partitioning tree and particle cluster are used to detect the collision of between particles and scene objects, through which better performance is achieved. We show how to accomplish the rendering of particles by quads based rendering technique. Furthermore, the fading effect is simulated by means of a gradual color thinning and syncretizing technique. Experimental results show that our algorithm achieves real-time simulation to both the animation and appearance of dense smoke in consumer PC platform.