Fluid Simulation on Point Clouds

Abstract

With the effects of climate change becoming more and more apparent in the form of drought, heavy rainfall and the melting of the polar ice caps research into more efficient algorithms for fluid simulation become more important. For cases such as heavy rainfall and the melting of the polar ice caps it would be convenient to be able to use real life laser scans of landscapes to predict what damage they will do and where we need to take precautions.

The purpose of this thesis is to come up with a method to directly use fluid simulation on point clouds. We do this to skip the expensive reconstruction step which is normally done and simplify the collision detection between the marker particles and object.

Our algorithm combines the MLS algorithm with a Eulerian fluid simulation. For each grid point we compute the MLS values and use these to determine how solid a grid cell is and keep the marker particles from going into the point cloud.

Our experiments show that it is possible to directly perform fluid simulation on point clouds using our method, but that the performance of our algorithm is highly dependent on the type of scene and the used parameters.