Book Image

Hands-On C++ Game Animation Programming

By : Gabor Szauer
Book Image

Hands-On C++ Game Animation Programming

By: Gabor Szauer

Overview of this book

Animation is one of the most important parts of any game. Modern animation systems work directly with track-driven animation and provide support for advanced techniques such as inverse kinematics (IK), blend trees, and dual quaternion skinning. This book will walk you through everything you need to get an optimized, production-ready animation system up and running, and contains all the code required to build the animation system. You’ll start by learning the basic principles, and then delve into the core topics of animation programming by building a curve-based skinned animation system. You’ll implement different skinning techniques and explore advanced animation topics such as IK, animation blending, dual quaternion skinning, and crowd rendering. The animation system you will build following this book can be easily integrated into your next game development project. The book is intended to be read from start to finish, although each chapter is self-contained and can be read independently as well. By the end of this book, you’ll have implemented a modern animation system and got to grips with optimization concepts and advanced animation techniques.
Table of Contents (17 chapters)

Pre-generating the skin matrix

One of the bigger problems with Vertex Shader Skinning is the number of uniforms that the system takes up. One mat4 object takes up four uniform slots and the skinned vertex shader currently has two matrix arrays that have 120 elements each. That comes to a total of 960 uniform slots, which is excessive.

What happens with those two matrix arrays in the vertex shader? They get multiplied together, as follows:

mat4 skin=(pose[joints.x]*invBindPose[joints.x])*weights.x;
  skin += (pose[joints.y]*invBindPose[joints.y])*weights.y;
  skin += (pose[joints.z]*invBindPose[joints.z])*weights.z;
  skin += (pose[joints.w]*invBindPose[joints.w])*weights.w;

One easy optimization here is to combine the pose * invBindPose multiplication so that the shader only needs one array. This does mean that some of the skinning process is moved back to the CPU, but this change clears up 480 uniform slots.

Generating the skin matrix

Generating...