基于预计算核函数的PBF流体模拟

为加速基于物理模型的流体模拟过程,提升PBF模型解算的实时性,引入一种预计算核函数方法。方法在模型初始化阶段构造并存储函数值表,在解算过程中,通过直接查表得到近似的核函数值及梯度函数值,避免了复杂的高次运算,在误差允许范围内,模型解算速度提升显著。此外还提出了一种PBF框架下的多相流模拟方法,能够模拟密度差别较大的两种不相溶流体之间的交互过程。为图形学中基于物理的多相流场景的模拟提供了一种稳定且高效的解决方案。

Position Based Fluids Simulation Based on Precomputed Kernel Functions

In order to accelerate the fluid simulation process based on physical models and improve the real-time performance of PBF model,a precomputed kernel function method is introduced in this paper.The method constructs and stores the function value tables in the initial stage of model.In the process of solving,the approximate kernel function value and the gradient function value are obtained by looking up tables directly,which avoids complicated high-order operations.Within the error tolerance range,the model solution speed is significantly improved.In addition,this paper proposes a multiphase flow simulation method under the PBF framework,which can simulate the interaction process between two incompatible fluids with large density differences.It provides a stable and efficient solution for the simulation of multiphase flow based on physical models in computer graphics.