A Spatially Adaptive Morphological Filter for Dual-Resolution Interface Tracking of Fluids

Abstract

We present a novel surface-tracking technique for free-surface fluid animations. Unlike the semi-implicit Particle Level Set method (PLS) our interface-tracking approach is purely implicit and hence avoids some of the well-known issues like surface noise and inflated memory footprints. Where PLS augments the interface with Lagrangian tracker-particles, we instead employ a higher resolution level set represented as a DT-Grid. The synchronization of our dual-resolution level sets is facilitated by a novel Spatially Adaptive Morphological (SAM) filter that attempts to preserve fine details while still avoiding spurious topology changes and boundary violations. We demonstrate that our approach can achieve comparative results to the PLS, but with a fraction of the memory footprint. We also show how our technique can be used to effectively enhance thin interface sheets at the cost of volume gain.