David Megginson wrote: > First, the intensity is far too low. At 0.5, I wasn't sure if > turbulence was working, and at 1.0 (maximum), I was still able to > control the aircraft easily.
Hrm, so it is. :) I must have changed something right at checkin. My memory of the final code is that the Cub would be flipped on its back by max turbulence while on the ground, and that the jets would bounce around noticeably on their gear. If you want to try playing around with the internals, you'll find a MAX_TURBULENCE value in Turbulence.cpp. This defines the upper bound for the velocity field. You can also try tweaking MEANINGFUL_GENS by one or two. This is harder to explain: if you set this low, then all of the energy goes into the high frequency components, and the velocity can swing betwenn +/- MAX_TURBULENCE faster. If you set it high, then more of the "bandwidth" goes into the long-wavelength variation, and there isn't as much spacial "bump" to the turbulence field. There's another secret constant in the rate parameter to the Turbulence::update() function. Right now, it presumes that the "Hz" parameter refers to the highest frequency component of the Perlin field, so there's a hardcoded factor of two. But that doesn't have any fundamental meaning. You could try tweaking this to get faster or slower progression of the field through time. > My other suggestion is to fade out turbulence a bit near the ground, > say, in the last 50 feet or so (at the point, the main problem is > wind gusts). Indeed. Certainly the vertical component of the turbulence field needs to go to zero at the ground. This isn't done yet. I think this is consistent with your description; "wind gusts" aren't quantitatively different from turbulence along the ground plane (someone correct me if I'm wrong). I'm not quite certain of the values to use. I found a reference to the "Dryden" turbulence model, which attenuates *all* turbulence to zero at the ground (clearly wrong; take a walk on a gusty day to see the proof), but drops the horizontal component along a steeper curve than the vertical. I might try something along these lines, but fudge the intercept so that the horizontal component is non-zero at ground level. Andy _______________________________________________ Flightgear-devel mailing list [EMAIL PROTECTED] http://mail.flightgear.org/mailman/listinfo/flightgear-devel
