If only I could get access to the pixels in a video in a uniform manor. Have not learned yet, but not prepared to.
Why not learn? There are not too many examples out there per se, but the bigflake collection
provides plenty of initial hints; there's a bunch more including multithreaded GL optimizations to really get there. You can directly covert to an OpenGL texture a camera frame or decoding an existing file and when rendering a GL frame / flip buffers you can have that encoded. So you'll be using GL 2+ shaders for real time or offline manipulation.
A great way to learn would be to create your brightness / white balance video conversion app and you could get pretty far modifying certain bigflake examples.