Blink Comparator Views of
the Plane at the Pentagon

By David Chandler, based on prior work by Ken Jenkins

In March, 2002 five frames from one video camera were released (leaked?) as small bmp images that purportedly showed the Pentagon plane. It had an erroneous time stamp (added later it turns out; not in the original), lack of context, small image size, and it did more to sow confusion than shed light on what hit the Pentagon. Here is the first frame of that set just before impact, together with a zoomed detail view.

Frame 1 of 5

Longer videos from two cameras, one with and one without the foreground obstruction, were released in 2006 in response to a FOIA request for surveillance camera videos at or near the Pentagon on 9/11. One of the videos contained the original five frames. These images are archived by the project through the website, which also documents the source FOIA.

Many people viewing these as still frames over the years (ourselves included) failed to see the plane. The lack of a clear visual image of the plane led to speculation that there was instead a small plane or a missile, and it opened the door to speculation that there was no plane at all. By the time the two longer videos were released in 2006, "no-plane at the Pentagon" speculation was well established to the point that this belief was virtually synonymous with the 9/11 Truth movement, in many people's minds.

The images shown below "blink" the frames that contain the plane with the frames just prior to them. These png images, blinked with a custom Javascript routine, use the images provided in the 2006 FOIA release. Animated GIF images were tried earlier, but the GIF compression reduces the color resolution. We have done no image enhancement: only magnification of the zoomed image pairs.



This is a zoomed-in detail of the view from what we shall call Camera 1.



The view from Camera 2 shows the plane a fraction of a second earlier. The wide angle lenses introduce barrel distortion, so the earlier image of the plane, which is closer to the edge of the frame, appears shortened.



This view does a blink comparison between the frames from Camera 1 and Camera 2 that contain the plane. The motion of the plane between these two views is apparent. This establishes that the two cameras were out of synch by a fraction of a second.



This view blinks the full image from Camera 1.



This view blinks the full image from Camera 2.


Except for the white smoke trail, the presence of a plane was not recognized by most people due to several factors:

  1. The low contrast between the backlit plane and the complex background.
  2. The extreme wide angle lens and low resolution of the video camera/recorder.
  3. Some blurring due to the motion of the fast moving plane.
  4. The small scale, low quality images that have circulated on the internet.

Blink comparators are commonly used to detect subtle differences between pairs of astronomical photographs. Variable stars show up as blinking lights. Asteroids move from night to night, even from hour to hour, so they appear to jump back and forth when blinked. Blinking two frames can also make a low contrast object stand out against a noisy background. This technique, applied to the security camera frames, allows the eye to more easily discern the presence of a plane, allowing it to pop out from the background each time it flashes into view.



A number of image processing programs, such as GIMP (which is what I used), have functions to compensate for lens distortion. Note that in the full views shown above, straight lines, such as the curbs, lines in the side walk, the roof of the Pentagon, etc., are bowed outward by the lens, an effect called barrel distortion. The parameters of the lens correction function can be varied until the curves become straight. This results in a "pincushion" distortion (inward bowing) that cancels the barrel distortion. In GIMP this is a manual process, using a slider to adjust the amount of distortion.



This view cancels the barrel distortion of Camera 1.



This view cancels the barrel distortion of Camera 2. Note that the process that cancels the barrel distortion of the image introduces a pincushion distortion of the rectangular frame.



This is a zoomed and cropped view from Camera 1.



This is a zoomed and cropped view from Camera 2.



This blinked image pair shows the motion of the plane between the two camera views using the corrected images. The images have been positioned so the background features on the horizon align. Based on the dimensions of the plane and motion relative to background markers, the plane moved about 115 ft between the two exposures. Using the FDR speed measurement of 556 mi/hr, we can compute that the two cameras were out of sync by about 0.14 sec. (I don't believe this time offset has been measured before.)


Note that when barrel distortion is eliminated, the image compression near the edge of the field is eliminated, so the plane appears longer, with proportions resembling a 757. (Remember, the plane is also moving toward us at about a 45 degree angle.)

One feature of the plane image helps us identify it as an American Airlines plane. Note the purple stripe along the side of the plane. American Airlines planes have parallel red and blue stripes. At the small scale of the original image it is quite likely that this purple stripe is a merger of the color information from a red and a blue stripe.