Dave Rogers - thank you for a reasonable reply.
Ace,
The two main issues, and my comments on them, are:
(1) The so-called "digital halo" around the plane. Other posters have suggested that these may be due to video compression artefacts. Have you studied the edges of the towers to see whether they show similar artefacts, or looked at compressed videos where you know no compositing has taken place, to compare effects and see whether what you observe is unique to digital compositing? Your analysis of this point suffers from the absence of a control sample, a basic principle of scientific analysis, and at present you have yet to prove that your explanation is superior to the alternatives that have been advanced.
Adding control samples is a great idea. I'm quite sure that planes don't leave trails of pink pixels behind them. I'm quite sure that noses of airplanes don't ever look like a swipe with a digital paintbrush.
(2) The variations in apparent aircraft speed. I accept that your normalisation technique is valid - measuring the difference in position between the aircraft and the edge of the tower is an effective way to eliminate camera movement. What you have failed to address is any kind of analysis of the random errors in your measurement of the airliner position. This is first year university course material. Your graph of speed against frame number looks very much like a horizontal line at 23 pixels/frame with a random error of +/- 4 pixels/frame, so your argument rests entirely on the unstated premise that you can measure the airliner position to an accuracy of significantly less than this. Given your own comments about the blurring of the edges of the airliner, and the clear grading of the edge in the stills you have posted, it appears to me that +/- 4 pixels is quite a reasonable measurement error, in which case your data would fall below the level of statistical significance.
I have certainly considered measurement accuracy. This is why I chose to track the outline of the entire plane. Shifting the outline even 1 or 2 pixels causes an obvious misplacement. To say that the placement of the outline could be off by 4 pixels in either direction, is simply not plausible to me. 1 pixel, quite possibly. 2 pixels, I doubt it but maybe. 3 pixels, no. 4, certainly not.
If you want to do some further work on this area, I would suggest you study the apparent movement of the tower side in the video and look for correlations between that and the velocity variation. If the video was truly composited as you suggest, there should be a very clear and strong anticorrelation between apparent movements of the tower (i.e. changes in camera angle) and velocity variations of the airliner, as in your scenario both arise from camera movement.
There would be strong anti-correlation between tower movement and velocity variations ONLY if the camera movement was significant, compared to the camera movement on the plane. It isn't. The chopper 5 camera movement is quite smooth. It was the camera movement on the original plane video that was unsteady.
As for the emergence of the nose, your suggestion as to how the video was composited is simply not plausible. Let us assume for the sake of argument that your hypothesis is correct, and a video of an airliner flying at constant velocity across a clear background was selected in advance of the events of 9/11 to produce a fake video.
Stop right there Dave. I clearly state that the airplane video is NOT at a constant velocity, rather that it has unstable velocity due to camera motion.
Given that the organisers of the attack had ample time to plan the events, and ample access to areas where airliners could be seen from the right angle and distance (I hope you are not trying to argue that either of these was not the case), what possible motivation could they have had for deliberately selecting a video in which the airliner appears to slow by a factor of four part way through the shot? To do so would then require that the camera angle from the helicopter be precisely set so that the airliner would enter the tower at exactly the right point on the screen, and any inadvertant camera movement would necessarily produce an unrealistic slowing of the airliner that would be obvious to a casual observer. It also makes no sense that the prepared video would have been taken from a helicopter, when a fixed ground-based camera would produce a video lacking any variations in camera angle or position, so that the manipulation of the video could start with a clip in which the airliner position was known precisely at all times.
It makes sense to have an unstable airplane video, knowing that the live shot will be from a helicopter. If the airplane flew at a steady apparent velocity, while the Chopper 5 camera moved, this would be a dead giveaway.
They chose a video where the plane appears to slow down in the middle of the frame, because they wanted it to disappear inside the tower. They did everything pretty much right, except, the chopper drifted to the left during the shot just enough to screw up their masking by that amount. That's why the darn nose appeared to come out the other side.
OOPS.