Mipmapping creates a sequence of
textures, each time downsampling the image by a factor of two in width and
height. The process continues until the
final image is only 1x1 pixel.
The mipmapped texture then contains
multiple levels of detail.
OpenGL will automatically choose
the level of detail. For example, points
far from the camera will be texture mapped with the smaller textures in the
mipmap, while points nearer the camera are textured with larger textures.
Mipmaps are memory efficient, only
requiring 1/3rd
more storage.