Game Development Reference
In-Depth Information
Because of the near unanimous adoption of this model, it is often used
without giving it a name, and in fact there is still some confusion as to
exactly what to call it. You might call it the Phong lighting model, be-
cause Phong introduced the basic idea of modeling reflection as the sum of
diffuse and specular contributions, and also provided a useful empirically
based calculation for specular reflection. (The Lambert model for diffuse
reflection was already known.) We saw that Blinn's computation for spec-
ular reflection is similar but sometimes faster. Because this is the specific
calculation most often used, perhaps we should call it the Blinn model?
But Blinn's name is also attached to a different microfacet model in which
diffuse and specular are at different ends of a continuous spectrum, rather
than independent “orthogonal” components being mixed together. Since
most implementations use Blinn's optimization for Phong's basic idea, the
name Blinn-Phong is the one most often used for this model, and that's
the name we use.
A huge part of realistic lighting is, of course, realistic shadows. Although
the techniques for producing shadows are interesting and important, alas
we will not have time to discuss them here. In the theory of the render-
ing equation, shadows are accounted for when we determine the radiance
incident in a given direction. If a light (more accurately, an emissive sur-
face) exists in a particular direction, and the point can “see” that surface,
then its light will be incident upon the point. If, however, there is some
other surface that obscures the light source when looking in that direction,
then the point is in shadow with respect to that light source. More gener-
ally, shadows can be cast not just due to the light from emissive surfaces;
the light bouncing off reflective surfaces can cause shadows. In all cases,
shadows are an issue of light visibility, not reflectance model.
Finally, we would like to mention several important physical phenomena
not properly captured by the Blinn-Phong model. The first is Fresnel 16
reflectance, which predicts that the reflectance of nonmetals is strongest
when the light is incident at a glancing angle, and least when incident from
the normal angle. Some surfaces, such as velvet, exhibit retroreflection; you
might guess this means that the surface looks like Madonna's earrings, but
it actually means that the primary direction of reflection is not the “mirror
bounce” as predicted by Blinn-Phong, but rather back towards the light
source. Finally, Blinn-Phong is isotropic, which means that if we rotate the
surface while keeping the viewer and light source stationary, the reflectance
will not change. Some surfaces have anisotropic reflection, due to grooves or
other patterns in the surface. This means that the strength of the reflection
varies, based on the direction of incidence relative to the direction of the
grooves, which is sometimes called the scratch direction. Classic examples
16 Pronounced “fre-NELL.”
Search WWH ::




Custom Search