High performance video graphic architecture using hybrid logarithmic system

Abstract

It is known that the engines generate an image from a scene described as a list of vertices, a list of triangles and the position of the camera. The transformation stage transforms the vertices from the scene coordinates to the camera's viewing frustum coordinates, including perspective computations, Typically, this stage is very sensitive, as it determines the on-screen position of the triangles, and therefore requires some precision so as not to distort the objects, and will also involve repetitive arithmetic operations.