I've been about this curious myself.
One of the first things I want to try is running
my daughter's '3 Happy Cats' mocap, which was created in DMC 1.0 with two Kinects, through DMC 2.0. '3 Happy Cats' presented some interesting challenges to DMC 1.0 because the system was not designed to work with small children. Also, her movements in the motion are a bit 'quirky' to say the least. 1.0 managed to work through the data remarkably well but with a bit more supervised coaching than usual (for me anyway.) It should be very interesting to see how 2.0 can track this data any more accurately.
I'll post again after I've had a chance to run the tests and do a comparison.
G.