Graphics Reference
In-Depth Information
CHAPTER 10
Rendering Techniques
10.1
Introduction
As computers get faster and faster the goal of producing very realistic images in real
time is becoming attainable even on a PC. This chapter describes ray tracing and the
radiosity method, methods that generate the most impressive images. We also briefly
discuss volume rendering.
We already touched on ray tracing in Chapters 7 and 9. What we called an “image
precision algorithm” in Chapter 7 is basically a very simple generic ray-tracing algo-
rithm. A more complete ray-tracing algorithm is the Whitted ([Whit80]) algorithm
mentioned in Chapter 9. In this chapter, we describe ray tracing in more detail. The
basic idea is really very simple. The problem is that to carry it out involves finding a
large number of intersections of rays with objects making it very slow. In the early
days of ray tracing it was common to have individual ray traced pictures take many
hours of computer time. We start off in Section 10.2 with ray tracing in a continuous
world. Ray tracing in the context of volume rendering is considered in Section 10.4.
Section 10.2.1 presents code for a basic ray tracer. Section 10.2.2 presents the math-
ematical solution to some more ray-object intersection problems. The special problem
of ray tracing CSG objects is discussed in Section 10.2.3. For more about implicit
objects see [Bloo97].
Section 10.3 is on the radiosity method. Because this approach is much more
involved we only give an overview of the main points and refer the reader to other
more comprehensive treatments of the subject. Section 10.4 gives a brief overview of
some topics in volume rendering.
Finally, although computer graphics has progressed to a stage where the pictures
it produces are really impressive, Greenberg ([Gree99]) questions their “realism.” He
raises the point that, by in large, the pictures are not yet accurate from a visual point
of view (as opposed to geometric accuracy, which is being handled adequately). They
may look good but do not represent a real physical scene most of the time. We still
have a ways to go if we want to use graphics to simulate the world and make pre-
dictions with regard to visual properties (in the same way that we can accurately
predict geometric properties). Although we have accurate light reflectance and trans-
port (how the light propagates through the environment) models, one is often forced
