Compositing, an interesting (if one is so inclined) and rather technical part of computer generated graphics.
It's essentially taking all the renders for one scene and combine them (or compositing, ha ha) to make one image (or in the case of a video project, series of images).
The renders for a scene, or
render passes, are different elements of a scene, for example, in the city scene, it was broken down as follows:
Hero Building (large building on the right)
Close City
Background City (the small buildings on the left)
Water Buildings
Water
And each of these are further broken down:
Shadows (only the background)
Diffuse (light hitting a surface)
Specular (shiny parts, such as metal, which reflect light)
Ambient Occlusion (how close things are together)
Beauty (a complete render)
And, well you get the idea.
Why...
Mainly, in my case, control, and to save time re-rendering. By breaking the scene down into sections, it gives an opportunity to adjust each part separately, and without re-rendering the entire scene, which is quite time consuming.
I may want to bring out the foreground building, or I notice that some buildings are being hidden by shadows. The subtle adjustments offered by rendering in layers, and compositing later make minor changes far easier.