Parallel realistic visualization of particle‐based fluid

Abstract

In this article, an entirely parallel pipeline for creating realistic visualization of particle‐based fluid is presented, which significantly improves surface reconstruction and foam generation performance. Firstly, we develop a spatial hashing grid‐based strategy. Based on this strategy, a novel narrow band surface reconstruction method is presented, which could even discard the detection of surface particles methods to achieve the effect of accurate identification of surface vertices. And a novel narrow band like diffuse particle classification method is proposed, which ensures less memory access and less computation. Furthermore, in order to obtain a best performance, we put forward several parallel schemes to avoid potential race conditions and invalid threads. Experiments are conducted to verify that our developed method and code efficiently improve the particle‐based visualization.

Narrow‐Band Screen‐Space Fluid Rendering

“…These methods can use isotropic kernels [MCG03, ZB05, SSP07], adaptive size kernels [APKG07] or anisotropic SPH kernels [YT13]. Despite the existence of parallel implementations of these methods [AIAT12, YG20, CZZ21], if the liquid spreads more over the computational domain, the underlying MC grid and its resulting surface mesh become very large, causing excessive memory consumption. Recently, Sandim et al .…”