I may have mis-heard, but I think Stephen said that when he was doing pano's, his rotation was around the scanning sensor - this isn't the nodal point as you would normally expect.  Is it different for a scanning back?