I think comparing movies of the speedo the determine the power is only an indication. I think a more reliable way to compare the power is look at the time difference in a 4th gear datalog. I usually look at the time from 2000-4500 rpm, but also at the times for 500 rpm steps. So 4000-4500 rpm for power and 2500-3500 rpm for torque comparisons. This method proved to be very accurate when I tested different mappings on the road, and compared the predictions with the dyno figures.