There seems to be a lot of transforming vertices, first to camera space, then coordinate space, then screen space. Are there any methods in modern hardware to do this all in one step or make it more efficient?