We only render the visible parts in a picture to avoid the redundant computation of invisible objects, which usually are dominant parts in outdoor scenes, at a fixed view field.