Preface
This book is based on modern OpenGL v3.3 and above. It covers a myriad of topics of interest ranging from basic camera models and view frustum culling to advanced topics, such as dual quaternion skinning and GPU based simulation techniques. The book follows the cookbook format whereby a number of steps are detailed showing how to accomplish a specific task and are later dissected to show how the whole technique works.
The book starts with a gentle introduction to modern OpenGL. It then elaborates how to set up a basic shader application. Following this discussion, all shader stages are introduced using practical examples so that readers may understand how the different stages of the modern GPU pipeline work. Following the introductory chapter, a vector-based camera viewing model is presented with two camera types: target and free camera. In addition, we also detail how to carry out picking in modern OpenGL using depth buffer, color buffer, and scene intersection queries.
In simulation applications and games in particular, skybox is a very useful object. We will detail its implementation in a simple manner. For reflective objects, such as mirrors and dynamic reflections, render-to-texture functionality using FBO and dynamic cube mapping are detailed. In addition to graphics, image processing techniques are also presented to implement digital convolution filters using the fragment shader, and basic transformation, such as twirl is also detailed. Moreover, effects such as glow are also covered to enable rendering of glowing geometry.
Seldom do we find a graphics application without light. Lights play an important role in portraying the mood of a scene. We will cover point, directional, and spot lights with attenuation and both per-vertex and per-fragment approaches. In addition, shadow mapping techniques are also covered including support of percentage closer filtering (PCF) and variance shadow mapping.
In typical applications, more complex mesh models are used which are stored in external model files modeled in a 3D modeling package. We elaborate two techniques for loading such models by using separate and interleaved buffer objects. Concrete examples are given by parsing 3DS and OBJ model formats. These model loaders provide support for most attributes, including materials. Skeletal characters are introduced by a new skeletal animation format (the EZMesh format). We will see how to load such models with animation using both matrix palette skinning and dual quaternion skinning. Wherever possible, the recipes also detail pointers to external libraries and web addresses for more information. Fuzzy objects, such as smoke are often used to add special effects. Such objects are typically handled using a particle system. We introduce a stateless and a state-preserving particle system in detail.
When a scene with a high depth complexity is presented, normal alpha blending techniques fail miserably. Hence, approaches such as depth peeling are used to render the geometry in the correct depth order with correct blending. We will take a look at the implementation of both the conventional front-to-back depth peeling as well as the more recent dual depth peeling approach. All steps needed in the process are detailed.
With computer graphics, we are always pushing the limits of hardware to get a true life-like rendering. Lighting is one thing that can convincingly represent such a depiction. Unfortunately however, normal everyday lighting is impossible to simulate in real-time. The computer graphics community has developed various approximation methods for modeling of such lighting. These are grouped under global illumination techniques. The recipes elaborate two common approaches, spherical harmonics and screen space ambient occlusion, on the modern GPU. Finally, we present two additional methods for rendering scenes, namely, ray tracing and path tracing. Both of these methods have been detailed and implemented on the modern GPU.
Computer graphics have influenced several different fields ranging from visual effects in movies to biomedical and engineering simulations. In the latter domain in particular, computer graphics and visualization methods have been widely adopted. Modern GPUs have tremendous horsepower, which can be utilized for advanced visualization methods, and volume rendering is one of them. We will take a look at several algorithms for volume rendering, namely view-aligned 3D texture slicing, single-pass GPU ray casting, pseudo-isosurface rendering, splatting, polygonal isosurface extraction using the Marching Tetrahedra algorithm, and half-angle slicing method for volumetric lighting.
Physically-based simulations are an important class of algorithms that enable us to predict the motion of objects through approximations of the physical models. We harness the new transform feedback mechanism to carry out two physically-based simulations entirely on the GPU. We first present a model for cloth simulation (with collision detection and response) and then a model for particle system simulation on the modern GPU.
In summary, this book contains a wealth of information from a wide array of topics. I had a lot of fun writing this book and I learned a lot of techniques on the way. I do hope that this book serves as a useful resource for others in the years to come.
