to emulate 'real play restrictions', pmb is using some more code to couple e.g. pitch and yaw to some degree, I think this isnt necessary ... anyway, the pictures above are all done using discretizations, but I used linear interpolation for plotting. the numbers below show the number of the frame, only with the topmost picture, there is a missing 10^4 or 10^5 on the screenshot, therefore it is maybe looking nondiscrete
the improvement I was testing in the 2 pictures below was 4th order runge kutta if that tells you anything. main reason to implement such an algorithm would be better stability of aiming against low framerates, @50fps you see hardly a difference. I guess for our purposes an adaptive step size control like "if we are below 20fps, run as any iterations in one frame that it looks like 20fps for the algorithm" would be sufficient
