These might help
You could also write your own collision detection based on dimensional arrays....
what I mean is...a sprite object is just an array of pixels...you could make an array for each sprite the requires pixel perfect collision and just program some code that constructs the array as a sprite...
I mean C2 is doing that already...but you might find a better faster way if you look into it..
each array entry point could contain as much information as you wish including positional data etc.
I mean its a double up but you might learn something new in the process which is what this is all about