Is albedo a term in rendering for the range of reflecticity, i.e. the wavelenths (color) of light reflected (and eventually absorbed by the eye)? e.g. An object with high albedo is going to reflect more wavelengths of light (like snow, hence it is white) and something like charcoal, which has a low albedo, meaning it reflects very few wavelengths and hence is dark.
Most 3D rendering doesn't work like that, but ignores physics completely.
The components basically look like this:
- self-illumination color
- ambient light * ambient color
- diffuse light * diffuse color
- specular
- reflection
Ambient light can either be constant and coming from all directions, or you can use a cube or spherical texture map pre-calculated from a radiosity solution and/or point/directional light sources. Ambient color is usually the same as diffuse color. Constant ambient is basically adding a bit of light to all pixels, ignoring surface normals and all, so it reduces contrast and depth and generally isn't preferred. Using an enviroment map will allow for changes in the ambient, so it brings out the shape of objects, and it works with normal maps as well.
I think Source can compile several point lights into an ambient cube map to light objects and characters; Halo3 looks like they're using some HDRI enviroment map for the indirect lighting (
see the Brutes video doc), and KZ does something like that too; other games like Gears and Lair seem to only use point lights and maybe a constant ambient.
For diffuse light you do a dot3 for surface normal and light vectors and multiply it with the light intensity and color. For speculars you can use various calculations depending on the lighting model: per-vertex and interpolate or per-pixel, phong or blinn or anisotropic etc. These have to be calculated separately for each light and the results are added together for diffuse; I dunno what blending mode is used for specular, adding is common but there are others too.
You have to do all these calculations for each light and combine the results together.
Reflections are also cube or spherical enviroment maps, either updated with render-to-texture or precalculated and placed into the scene as discrete samples. You can then spot the changes when the game changes between probes; one of the Cod4 movies had it, for example, on the scope of the sniper gun.
Half-Life 2 prefers to use reflection maps instead of speculars on level geometry and it looks quite good; it's probably also faster to calculate then a lot of speculars. Cars and glass obviously need reflections too.
So as you can see most of this is cheating, for two reasons: faster to calculate and easier to control for the artists and programmers. The dev guys here can surely add some more insight to the implementations if you're interested.
