I reviewed pixel perfect collisions on the video card and while that would be great to learn, it doesn't fit what my designer wants to do for distribution. I'm concerned about the gpu performance on phones, so I'm going to try out Box2D instead. I won't get pixel perfect collisions, but the game design and image resources will have to compensate. While I'll only be using a small subset of Box2D, part of my reasons for doing jams is to poke on libraries and techniques that I don't normally get to use.
Next up: a test harness for Box2D spinning shapes and collisions along with a spinning camera.