I am directing a short film that employs iPi motion capture, and my animator suggests that a feature that can "automatically reanalyze the actor every set number of frames" would "compensate subtle changes in lighting across the screen." He says the animation tracks much better if he reanalyzes the actor every 100 frames or so. Thanks!

You are right, that's a possible way to improve tracking robustness in future versions. It's a bit more complicated then what you described since a naive implementation of this algorithm does not work because of misconvergence. There exists a more advanced approach called "Active Appearance Models" in computer vision science. At this point I cannot tell how much time it will take to implement this feature into software, but we'll be working in this direction.