I have a project where I need to be able to rotate, stretch, and skew images fairly quickly. The images will almost always be completely in memory (IE no file to deal with). So I figured "Hey, Graphics cards are good at the sort of stuff, why not use them?" The other benefit to this is that it would make the cross-platformness of the algorithm much easier to handle.
So here is what I would like my pipeline to look like. Get image data and put it in some OpenGL readable texture, Tell OpenGL to transform the texture, retrieve the transformed data from opengl and place it into some new image container (IE an HBITMAP or XImage).
So can you do something like that? or do I just have to rely on the cpu to process this stuff?