In the previous chapter, I introduced the volume rendering integral, which has provided the foundation of volume rendering for scientific visualization since the late 1980s. In this chapter, I will review the current techniques used to apply the volume rendering integral to volumetric models.
We can break the process of volume rendering into two tasks. The first task is that of determining cell-ray intersections. This task is the process of determining which cells each viewing ray intersects. The result is a list (or stream) of samples (or segments) along each viewing ray enumerating the properties of the volume.
The second task is that of performing color computations. This task is the process of applying the volume rendering integral to the properties of the volume already sampled along the ray.
Although I describe volume rendering as a two-step process, in reality, volume renderers most often perform these two tasks together in a pipeline configuration. Even if we do not perform the process in a true hardware pipeline, pipelining the tasks prevents the system from having to store the intermediate samples along viewing rays, which are numerous for high quality renderings.