Integrated graphics is CPU that does this rasterization right? So what is different about a GPU that makes it better?
Is there any reason behind describing primitives as triangles instead of squares? Since we need to make it a bitmap image at the very end, it might be easier to color an image made up of squares.
It seems that we actually take one D (the array) to three-D (object created by triangle) and then get the 2D (on the screen) that we want. Although the scene is 3D, for calculation purposes, is there any way we could calculate the 2d picture directly from the input we get? (by some supermatrix or formula)
Why do we choose triangles among all the polygons? I understand that the triangles are the most basic polygons which can construct others. However, considering the physical implementation, the common pixels are rectangles, so why don't we just use rectangles as 3D "primitives"?
I know that modern GPUs often allow you to be very specific in defining the features of any given triangles, to the point where the descriptions could be Turing complete (which was the basis for CUDA, as I understand it). Will we be using CUDA at all in this class?
What are advantages of using triangles as 3D primitive? If we want a rough rasterization with faster computation, is changing primitive type (let's say from triangle to rectangle) be a possible approach?
How did people come up with the idea of using triangles as primitives or the idea of using such primitives in the first place?
what are the special properties of triangles that make it the best primitive type?