John Costella's theory: No pincushion distortion present in the Stemmons sign

[Jerry Organ]

Please bear with me chaps.
I've had a slightly wrong-headed idea about pincushion distortion and that it was a very localised effect as the images passed from the 'central display' to the area with the sprocket holes. I'm now getting the impression it's something that affects almost the entire image.
Just to clarify an earlier point Chris - you were quite adamant that Costella had not distorted any images then went on to explain about the 'pincushion correction program' which, I assume, distorts the original film to compensate for the pincushion effect. So, the flopping pole is created by distorting the images using the 'pincushion correction program', it's just that other aspects of the original which were flopping around in the original due to pincushion distortion are no longer flopping around because the pincushion effect has been corrected (I'm sure I could've put that better)

The Costella frames are very compromised and degraded in quality by modern standards. Yet many critics trust the frames because a critic "fixed" them.

Turning now to Jerry's excellent graphic (once again).
As I understand it, these are 'non-corrected' images from the Z-film, so the pole should be staying straight as the corner of the wall 'flops around'. The first thing that stands out to me is the slice from z228. It is clear, when compared to the other frames that the top of the 'wall corner' is leaning towards the left in the way indicated by the arrow. What is equally clear to me is that the bottom of the pole is doing exactly the same thing. At about the halfway point the pole bends noticeably in the direction of the arrow, this 'bending' gets more pronounced as it reaches the sprocket hole.
Should this be happening?

The pincushion effect distorts along curved lines.



The center of the pincushion is in the normal projected frame area of a Zapruder frame, not the full width type frame that includes the interprocket area.

[Dan O'meara]

The pole definitely looks like its bent but my tired old eyes might be misleading me.

[Chris Bristow]

 If you take the grid Jerry printed and enlarge it, then look at the vertical line closest to the corner. As it raises up the length of the first two boxes there is no measurable bend. At the 3rd box the bend suddenly shows up. The nature of pincushion is weird when viewed on a rectangle because pincushion increases in a radial way. The pixel in the very corner is farther from the center than any other pixel so it get pushed out the farthest. That is why the corners look like pincushions. Everything in the corners is farther than pixels on the side. So it just seems to work out that the group of pixels near the corners stretch and bend a pole in a way that does not perceptibly curve it.

   

[Chris Bristow]

I wanted to confirm just how much pincushion there should be in the Stemmons pole so I used other objects sitting in the  upper half of the images that correspond to the same place in the lower quadrant were the pole sits. Matching locations in the upper and lower quadrants have the same pin distortion because they sit in equal but opposite positions relative to the center of the frame.
Taking the lightpole in frame 48 that sits in the upper right quadrant equals the location of the pole in fr193. comparing the corrected frame 48 with the uncorrected fr 48 allows for a very accurate measurement because I don't have to worry about rotating a frame. The result is 1 1/2 degrees of distortion.
  Doing the same thing for frame 228 I used frame 226 because the corner of the wall in the upper left matches very close to the pole in fr 228. Comparing corrected vs non corrected showed a distortion of 2 degrees.
 Based on this the total distortion should be 3 1/2 degrees.
 About half that is visible in the uncorrected frames when you compare the pole in 193 and 228. The other half is found in the corrected frames when you measure Costella's separation of the pole in 193 vs 228.
 The non corrected Groden frames showing only half the amount of pincushion they should have resulted in the correction program over correcting. That seems to be why the correction program causes the mismatch Costella found.
 knowing that Costella's program worked on most every object in the film is a verification that the program was sound. The real fundamental question is why the uncorrected film does not show enough pincushion because that is what causes Costella's anomaly. In fact Costella's program being sound is proof there is a problem with the uncorrected Groden version. So what happened to the uncorrected version that caused only the stemmons sign to go haywire when corrected for pincushion.

[Dan O'meara]

Hi Chris,
I'm a bit out of my depth here technically but would just like to ask this as I'm not sure whether you answered this or not.
In reply #9 I posted a picture which, I believe, shows the pole bending in a non-corrected frame. Is this the case or have I missed something? (if you could keep it in layman terms it would be much appreciated)
The only reason I want to follow this up is because I am of the firm belief the Zapruder film hasn't been altered, not because of technical reasons, my reasons are purely subjective.

[Chris Bristow]

Hi Chris,
I'm a bit out of my depth here technically but would just like to ask this as I'm not sure whether you answered this or not.
In reply #9 I posted a picture which, I believe, shows the pole bending in a non-corrected frame. Is this the case or have I missed something? (if you could keep it in layman terms it would be much appreciated)
The only reason I want to follow this up is because I am of the firm belief the Zapruder film hasn't been altered, not because of technical reasons, my reasons are purely subjective.

Yes things do Bend but consider the graphic that Jerry posted was an exaggerated version. You can see a bending Arc in the wall behind the Stemmons sign in the uncorrected version. But if you look at it Jerry's image of pincushion distortion the vertical lines in the very corner raise up from the bottom to the level of the second Square without showing any Bend.
 When you try and look at the Stemmons Pole it is very hard to judge subtlety because those Shadows on it obscure straight lines. If you look at the edge of the sign it might be a more accurate measure.

[John Tonkovich]

Deleted