HappySqurriel on 23 July 2009
The can probably get "1 Million polygons per car" because of an efficient level of detail system which would only use that level of detail when part of the car was taking up most of the screen; and through the use of advanced datastructures for more efficient clipping and culling (consider that less than 1/3 of polygons from a car are probably displayed at most times, and if you're not displaying it you shouldn't render it).
