Physics modeling fluid image software#
Physics modeling fluid image code#

Furthermore, the isotropic kernel function is defined as where is the radius of the kernel and is the distance between 2 points. is the desired density of the fluid and is the estimated local density around particle. This constraint aims to keep the incompressible property of the fluid (water in particular). This allows us to change parameters of the simulation without re-compiling the code.

Physics modeling fluid image code#

We found that the hyperparameters of the equations affect the effect of simulation significantly so these parameters are now loaded in through the scene JSON file using a modified version of the parsing code that was already available in Cloth Simulator. These adjustments are applied to the particles (along with collision detection) to simulate the fluid. Once we have computed the neighbors, we use this information to compute various quantities like incompressibility correction, vorticity confinement adjustment etc. Since the method heavily relies on efficient neighbor finding data structure, we find that using the naive grid hashing approach unsatisifiable so we used an external library FLANN that implements KD tree, whcih allows us to find neighbors more efficient neighbor look up. We implemented all functions used for evaluating the parameters needed and also the simulation steps as faithfully as the paper. Each Particle has with properties like position, velocity, viscosity etc. Class Fluid is a class that serves as the abstraction for the fluid simulated and it owns all the Particle struct in the scene.

Physics modeling fluid image software#

Our software architecture for this part uses a similar structure as project 4 Cloth Simulator. The core component of the project is position-based simulation of water particles. Technical Approach Particle-based Fluid Simulation Lastly, we attempted to use a deep learning techniques to accelerate the simulation step. Then, we reconstructed the mesh representing the surfaces of the particle-based fluid using isotropic kernels and render the mesh with path tracing. Over the course of the past month, we have implemented the Position Based Fluids paper written by Nvidia by Macklin et al. We are interested in producing photo realistic simulation of water and understanding the underlying physics of water. Gauthier Dieppedalle (24362705), Utkarsh Singhal (25674685), YiDing Jiang (25551317)įluid simulation is an interesting topic that has many important applications in computer graphics. CS 184: Computer Graphics and Imaging, Spring 2018 Final Project: 3D Position Based Fluid Simulation and Surfacing