Counting Pixels in the Zapruder film.

[Drew Phipps]

After some back and forth about the validity of the Zapruder film on the forums recently, I thought I might use my pixel counting software to see if I could detect anything unusual, by comparing the apparent motion of the limousine across the background of bystanders. I only analyzed Z-frames 277 through 344, reasoning that this was a) the time of greatest interest, and b) the time when the limo was moving nearly perpendicular to the camera. I reasoned that if the limo was moving partially towards the camera, the apparent motion of the bystanders compared to the limo would be too slow.

The limo is being more or less tracked by the camera, but Zapruder's tracking isn't perfect. By comparing the relative position of the limo to the bystanders, I hoped to minimize Zapruder's naturally imperfect tracking. I believe the same technique is used to "stabilize" the image of the limousine on the film. If applied to the film, the limousine would not change its position in the image, but the bystanders would pan to the left.

There are some caveats: Counting pixels always enters an error factor of at least 2 pixels, one on each end of the measurement, as you cannot tell whether a pixel is partially or totally filled with a bit of the image. The Zapruder film is low resolution, even the best images are a bit blurred, and there are many frames in which the blur seems to be the defining characteristic of the image. I tried to minimize the impact of this fact by only reporting here differences in relative movement that exceed 50% of the normal relative speed. Last, I cannot say for sure whether the anomalies I notice below were caused by a Bell & Howard camera malfunction, or by the process of digitalization of the film, or represent actual "cut and paste" editing of the celluloid (or the digitized) film. I used the "Costella" images available on the net at http://www.assassinationresearch.com/zfilm/

I selected a fixed point on the limousine, the upper passenger side corner of the windshield area (there are some sort of grey attachments to the windshield, they come to a sharp point and make an easy reference point). For bystanders, I selected a reference point on each bystander which was visible in all frames in which they appeared. In each frame in which the bystander appeared, I compared its point of reference's apparent movement to the limo's point of reference.

(continued)

[Drew Phipps]

Here is a quick summary of the bystanders and their apparent speed in pixels per frame:

Brehm's kid 14 frames* avg speed 39 pixels/frame
Brehm 18 frames* avg speed 37 pixels/frame
Jean Hill 22 frames avg speed 37 pixels/frame
Moorman 22 frames avg speed 35 pixels/frame
Toni Foster 27 frames avg speed 32 pixels/frame (walking a bit to the right)
White blur 23 frames avg speed 30 pixels/frame
Running Man 17 frames** avg speed 33 pixels/frame (running a bit to the right)
Bothun 13 frames** avg speed 31 pixels/frame
Altgens 6 frames** avg speed 31 pixels/frame

*Brehm, and his kid, start in the middle of the image at Frame 277.
**Running Man, Bothun, and Altgens are still in image when the limo reference point moves offscreen at Frame 344

Individual frame anomalies:

In the following frames I noticed anomalous relative movement of bystanders.

Frame 282: Brehm has moved left 25% more than his average speed. Brehm's kid does not display this unusually fast motion.

Frame 283: Brehm has only moved left at 50% his average speed. Brehm's kid has also only moved left at 50% his ordinary speed.

Frame 296: Hill has moved 45% faster to the left than her average speed. Moorman has moved 51% faster to the left than her normal speed.

Frame 304: Toni Foster has moved "up" 15 pixels. Hill and Moorman do not exhibit this unusual movement.

Frame 308: a) Hill moves "up" (relative to the limo) 19 pixels. Mary Moorman does not move up. Toni Foster does not move up.
(Hill also moved up 10 pixels in Frame 307. She could be rising up on her toes, if that can be accomplished in (2 / 18.3) sec. If so, she passes out of frame without settling back down.)

b) Toni Foster has moved to the left only half her average speed. Hill and Moorman do not exhibit this unusual movement

Frame 312: Toni Foster moves left 65% faster than average. Nothing to compare Foster motion with.

Frame 313: Toni Foster moves left 45% faster than average. White blur appears. Nothing to compare Foster motion with.

Frame 324: a) White blur moves left only 20% of its normal speed.

b) Toni Foster appears to move RIGHT with respect to the limousine.
(This could be a result of some sort of image displacement in the "sprocket area," which contains other displaced images as well)

Frame 325: White blur moves left 50% faster than its normal speed. No background objects to compare.
(Could be a darting dragonfly or some other large bug with iridescent wings, explaining shape anomalies, but dragonflies are rare in Texas in November)

Frame 331: White blur moves left only 50% of its usual speed. Running Man and Bothun do not exhibit unusual motion.

Frame 333: All three background objects (white blur, Running Man, and Bothun) drop down more than 30 pixels with respect to the limo.
(Zapruder could have lowered the camera to produce this effect, but it seems unlikely that he could do this without any effect showing up in Frame 332 or Frame 334, which means he began and completed a significant physical movement of the camera in 1 / 18.3 sec., and without a lot of extra blurring)

Frame 342: Running Man drops down 21 pixels with respect to the limousine. Bothun does not drop down anywhere near that amount. Altgens does not drop down anywhere near that amount.

(continued)

[Drew Phipps]

Overall conclusions:

a) The average speed of the limousine is decreasing from frame 277 onward, because the fastest moving background bystanders are Brehm and his kid. Brehm is stationary, but appears to be shifting his weight from his right leg to his left leg, which should slow down his relative speed, but his movement is still faster than subsequent bystanders.

I would expect the apparent speed of a limo (at constant road speed) to increase as its movement becomes perpendicular to the camera. But that isn't what happens. With respect to the bystanders, the limo is moving no faster than 81% of its original velocity (at frame 277) when it reaches the middle of my test segment (Frame 310) and does not regain that original velocity until after frame 344 (where my reference point passes out of frame)

b) In all likelihood, the "white blur" is not moving with respect to the background individuals. At most it is "drifting" against the wind (which I understand was blowing mostly from right to left in the Zapruder film) just a tiny bit (2 pixels per frame at most), which is well under the margin of error for such measuring. This suggests that it is a fixed point on the ground (which somehow changes shape) or possibly, a hovering insect. (See Frame 324-325 above) It never displays any rising or falling motion relative to the limo, which you might expect if it was some freefalling remnant of a gunshot.

c) If you think that the bystanders were part of a background that was added in as a whole piece to the images of the limo, you can find the bystanders anomalously moving as a group (with respect to the limousine) in Frame 283, 296, and 333.

d) If you think that individual bystanders were cut and pasted into the background you will find evidence of that at Frames 282, 304, 308, 324, and 342

e) There are other anomalies in the background movement with respect to the limo at frames 312, 313, 324, and 325; but there are no other background images whose movement can be compared to the anomalies I noted above.

Final Conclusion: There are certainly anomalous visual images on the Zapruder film. I personally did not expect to find quite so much anomalous movement. Although "proof" in my mind would require some sort of camera or film expert to eliminate other camera causes, and image distortions, I am prepared to conceed the possibility that the Zapruder film has been altered substantially, in some way other than mere deletions of unwanted frames.

[Drew Phipps]

Here are the most pertinent frames from the Costella site of the Zapruder film:

Frame 282

Frame 283

Frame 296

Frame 304

Frame 308

Frame 324

Frame 333

Frame 342

[Drew Phipps]

There is another anomaly involving Joe Brehm (the boy) and the Babushka Lady's coat. Examine these cropped versions of Zap Frames 277, 283, 289:







Each of these frames is taken less than 1 / 3 of a second apart, so the whole sequence takes just under a second. Joe Brehm has moved out from behind his father, Charles, WITHOUT moving his feet. While clapping. Some of the apparent movement is Charles shifting his weight from his right leg to his left, but that doesn't explain the placement of Joe's and Charles' feet. You might expect to see something like this is the camera was physically moving left with respect to the placed objects, but not standing still, panning from left to right.

Nor does it explain how Joe manages to drag a corner of the Babushka Lady's blueish coat with him, even though she is standing a considerable distance to the rear. The Babushka Lady's coat has done something strange, gone from nearly vertical in ZF 277 to an angle of 28 degrees off vertical in that same time frame.

[Drew Phipps]

In this sequence from Z279 to Z281 (1 / 6 of a sec) a shadow to the right and behind Charles Brehm and the Babushka Lady appears and disappears. Cropped photos:







The next bystander to come into view is Jean Hill, but she isn't standing in a place where she might cast that shadow. Is this a film error, a "cut and paste" mistake, or has a bystander deliberately been eliminated from the background?

Tjere are other strange black lines extending from the right edge of the photo in F285 and F286. The one in 286 looks most like some sort of film error. There are other similar black lines in F291 that eventually become the shadows of Toni Foster and a group of folks walking behind Toni Foster.

[Herbert Blenner]

After some back and forth about the validity of the Zapruder film on the forums recently, I thought I might use my pixel counting software to see if I could detect anything unusual, by comparing the apparent motion of the limousine across the background of bystanders. I only analyzed Z-frames 277 through 344, reasoning that this was a) the time of greatest interest, and b) the time when the limo was moving nearly perpendicular to the camera. I reasoned that if the limo was moving partially towards the camera, the apparent motion of the bystanders compared to the limo would be too slow.

The limo is being more or less tracked by the camera, but Zapruder's tracking isn't perfect. By comparing the relative position of the limo to the bystanders, I hoped to minimize Zapruder's naturally imperfect tracking. I believe the same technique is used to "stabilize" the image of the limousine on the film. If applied to the film, the limousine would not change its position in the image, but the bystanders would pan to the left.

There are some caveats: Counting pixels always enters an error factor of at least 2 pixels, one on each end of the measurement, as you cannot tell whether a pixel is partially or totally filled with a bit of the image. The Zapruder film is low resolution, even the best images are a bit blurred, and there are many frames in which the blur seems to be the defining characteristic of the image. I tried to minimize the impact of this fact by only reporting here differences in relative movement that exceed 50% of the normal relative speed. Last, I cannot say for sure whether the anomalies I notice below were caused by a Bell & Howard camera malfunction, or by the process of digitalization of the film, or represent actual "cut and paste" editing of the celluloid (or the digitized) film. I used the "Costella" images available on the net at http://www.assassinationresearch.com/zfilm/

I selected a fixed point on the limousine, the upper passenger side corner of the windshield area (there are some sort of grey attachments to the windshield, they come to a sharp point and make an easy reference point). For bystanders, I selected a reference point on each bystander which was visible in all frames in which they appeared. In each frame in which the bystander appeared, I compared its point of reference's apparent movement to the limo's point of reference.

(continued)

I suggest that you use a program such as Paint to read the coordinates of a point then calculate distance between two points. This method is far easier on the eyes.

[Paul Rigby]

The next bystander to come into view is Jean Hill, but she isn't standing in a place where she might cast that shadow. Is this a film error, a "cut and paste" mistake, or has a bystander deliberately been eliminated from the background?

As matters stand, it's impossible to be sure who was eliminated from the real scene by the compilers of the fake film, but three candidates suggest themselves, all female book depository employees who said they stood together on the south curb of Elm at roughly the mid-point between the TSBD and the overpass (Holt, Simmons and Jacob in 22WCH652-3; presence on Elm St in shooting's immediate aftermath confirmed by policeman Lewis in 19WCH526).

An attempt to preserve two of the three vanishing TSBD employees from the south curb would have necessitated the removal of Hill and Moorman. Interestingly, the ground appears to have been prepared for just such a switch.

All three TSBD women's statements were seemingly taken on the same day, 18 March 1964.The statements of Holt and Jacob were structured by the same pair, A. Raymond Switzer and Eugene F. Petrakis. The job of fashioning Simmons', by contrast, was undertaken by E.J. Robertson and Thomas T. Trettis. The latter duo had very interesting recent "form."

Five days earlier Robertson and Trettis had interviewed Jean Hill, another, albeit rather better, publicised south curb witness. This was to produce startling results. Let Harold Weisberg take up the story: "Reporting their interview with Mrs. Hill, the agents write things they must have known to be wrong. The two women [Mary Moorman, being the other PR] were not opposite the main entrance of the Texas School Book Depository Building' but considerably west of there, opposite the location of the President's car at the time of the fatal shot" (Photographic Whitewash, 1976, p.36). Hill and Moorman, ridiculous as it may now seem, were to be shifted and "lost" in the crowd.

How to make sense of this? On the surface of it, simple enough: Five days before Robertson and Trettis separated Simmons from Holt and Jacob, the same pair removed Hill and Moorman from the south curb, too. Why? Did the latter pair make one pair too many for the Z-film? The entire problem arose, I hasten to point out, because the film was refashioned only after the appearance of the stills in Life magazine in its first post-assassination edition.

[Chris Davidson]

Drew,

If you will revisit some of the info I have posted about the limo slowing down before the extant 313 headshot, you will find you are right where you need to be with your pixel analysis.

Remember, Paul Mandel in his Life Magazine article (Dec 6,1963) stated there was 74 frames between the 1st and 2nd shots.

Robert West who did the corrected survey for Life Magazine on Nov25,1963 plotted the first shot at what is extant frame 207.

Those two previous actions get us to frame 281.

He also states there is 48 frames between shot 2 and 3. This is where the slowdown occurs with the limo traveling at 3.74 mph, what would cause the limo to slow down abruptly, a shot.

Simmons tells us they are using a 240 slant distance as one of their targets. What's not on film?

If you are not aware of CE884, the last two slant entries are for frame 255-313. Frame 255 is 218 ft and 313 is 265.3ft away.

The slant distance between these two points is 47.3ft. Total frames 58.

Just averaging this out equals 1.22frames per 1 slant ft.

So, 218-240 slant ft = 22 x 1.22frames = 27 frames = 255+27 = Frame282

This is another reason why the last two entries are 58 frames apart. It was much easier to hide the fuzzy math and film alterations with a bigger span of frames to work with and then average things out.

chris

P.S.

Proof of Concept + Film Alteration Proof coming soon.

[Drew Phipps]

The Bronson photo here:



Is said to have been taken at around z225 or z226. You can see at this time the boy Joe Brehm is well to the side of his father (at least from this angle). This is 53 frames (allegedly ~2.9 seconds before frame 277). The boy Joe Brehm has moved behind his father and then back out from behind him in 3.9 seconds or less without visibly moving his feet.