Fluid Engine Development 1st Edition

This text is written to be developer friendly. The code used is C++ with occasional use of some features of its latest C++11 revision so you may have occasion to refer to Bjarne Stroustrup's webpage. Explanation of the vector analysis needed and the physics is informal as the author admits. For example the author views viscosity as producing a force which tends to reduce velocity difference between neighboring points. The velocity field is therefore blurred and this blur can be produced by adding the Laplacian of the velocity field to the velocity field. Though the Laplacian may measure bumps or curvature, there's no proof here. He knew the result he had to get and essentially used that to define viscosity. That's okay though as he just wants to bring about some feel for it. He's more interested in implementing these effects in the code. Eberly's Game Physics 2nd edition is of help here especially chapter 5 on fluid dynamics. If you'd like to see an adequate rigorous development and proof of the full Navier-Stokes equations as well as a complete explanation of viscosity see Victor Streeter's Fluid Dynamics (cheap on Amazon).The engine is to simulate an incompressible, viscous fluid so that density is constant and the divergence of the velocity field is zero. As he is concerned with questions of convergence and stability in the code his explanation of numerical algorithms is thorough. For grid-based solution (Eulerian) he uses the finite difference method and there is no finite elements. He also uses a Lagrangian approach (particle trajectory not just within a fixed region) called smoothed particle hydrodynamics (SPH) This alone is worth the price of the book. This is fluid simulation by pointilism, Basically you take a dust of points. Regions around each point are filled with the aid of a kernel function which is used to distribute some property. Key objects are density and pressure. Pressure gradients tend to increase density and density will have to the be adjusted to maintain incompressibility. You'll find these kernels act as weighting functions for the masses of the points in determining density. In fact they're a discrete analog of green's functions which act as weights in some E&M integrals. This probably is no surprise as the method originated to simulate compressible flow in astrophysics (magnetohydrodynamics and plasmas) through the insight of Australian physicist, J.J.Monaghan. It seems kind of crazy to use this for incompressible flow but relatively few points can be used which saves computer time and expense. In short he shows and explains the code of how this method can be implemented in this case. Here you run into stuff like the Pressure Poisson equation and how to avoid as well as the predictive corrective method. He later explains hybrid methods. You'll have a pretty good feel for the material when you get through it I'm sure-maybe even enough to try compressible flow. The author maintains a site on gitHub where you can copy source code.

I honestly didn't find this useful about 1/3rd of the book is basic material.I returned it...

This book starts at the fundamentals and builds from there. It goes into a good bit of depth and provides all the necessary information in a logical order. But the book's biggest flaw is that the code snippets in the book aren't actually enough to build a working fluid simulation. To remedy this, the full source code is placed online on the author's github, but the code on github doesn't match the code in the book very well. The github code is quite hard to follow due to it's heavy use of inheritance which makes different classes (such as which method for solving systems of linear equations) interchangeable, but it also make it very hard to understand where the actual work gets done.After reading the chapters and typing up the code snippets myself, I still had to spend significantly more time to get a working simulator.

This is the best book I have ever read regarding fluid sim development, if you're familiar with C/C++ and the basic vector computation, and interested in writing you're own fluid solver, this book would be the best option for you to start with.The author introduces the concepts from scratch to the full solver structures accompanying with a lot of examples and figures, it's very easy to get understand the ideas. Other than that, the author also explains every part of the C/C++ code in details, that does help the readers to build their own program and improve their programming skills.Personally, I think this book is not only for programmers but also very suitable for FX TDs/Artists who want to understand the algorithms of the fluid solvers in the 3D software such as Maya or Houdini.

Easy to follow and touches many topics.