@latenlazy: not true. If you have a single picture and the right software, you can actually calculate the distance between the aircraft and the camera due to perspective distortion between the closest and furthest elements of the aircraft, when the size of these elements are known. Then you can determine the relative distance of the aircraft based on its movement relative to an object, etc etc etc. Your issue is more that I'm suggesting your pet plane isn't maneuverable.
@Deino: data-analysis of poor-quality is doable, but it does present issues regarding accuracy of data. Think about it another way, Say, you have a measuring tool that can give you the weight of a canteloupe within 1 kg, when a canteloupe usually weighs around 2 kg, +-1 kg. So, when you weigh the canteloupe, you'll get figures between 0 kg and 4 kg. The solution is to repeatedly weigh the canteloupe over and over again. As you get into really high numbers of measurements, the average of all your measurements begins to approach the average actual weight of the canteloupe. The same is possible with poor quality data; i.e, get the resulting figures out of poor quality data, then collate the vast sum of measurements together.
What latenlazy is saying is, don't bother. But if all you have is your wonky scale, and you desperately need to get the exact weight of the canteloupe, you're going to start collating bad data together to establish an estimate. This is what we're doing with these videos.
@AFB: here's the same video, slowed down 20x. You can definitely see the vapor clouds form in the video. I finally timed the video, and it is roughly a 60 degree/sec instantaneous turn if you ignore the fact that it's been sped up. So we get figures like 30 degrees, 20 degrees, 15 degrees, and 10 degrees depending on the degree to which it is sped up.