This is just computer graphics theory. The position and rotation matrix are used together with vertex data to determine the depth and normal at each pixel. You can also cause skews and shears in the object by setting the components in a non-orthonormal manner.
Try reading through this to get a better understanding of how transformations are performed to get to a rasterized object: Graphics pipeline - Wikipedia (couldn’t really find a simple article for novice level, hope this helps)