Reply #68: No, look again.

[eomer]

It's not true that everything in 313 looks horizontally longer. Light objects look horizontally longer, while dark objects look horizontally shorter. This is true because light objects win out over dark objects in the process of light exposing the film.

So if the blurring motion were, say, three inches from left to right then the left edge of a light object will appear as if it is in the position at the start of the blur and the right edge of that same light object will appear as if it is in the position at the end of the blur, making the light object appear three inches wider than it really is.

The left edge of a dark object, on the other hand, will appear as if it is in the position at the end of the blur and the right edge of that same dark object will appear as if it is in the position at the beginning of the blur, making it appear three inches narrower than it really is.

You can clearly see this effect by looking at Connally's head in your animation. The light area of his forehead appears to get wider in the transition from 312 to 313, while the dark area of his temple appears to get narrower.

The area of controversy in these frames is the left side of Kennedy's head. We have to look at the left side because the right side is obscured. So we are looking at the left edge of a dark object with a light object to its left. If you want to try to realign things to correct for the blur then the accurate way to do it would be to take a slice from 313 and lay it over 312 while aligning all the areas that have an edge between a dark object on the right and a light object on the left. When you do this correctly you will notice that the areas of the frames that have the inverse relationship (an edge between a light object on the right and a dark object on the left) will not line up between the two frames but rather will be offset by the length of the blur. This is the right way. What you've done, apparently, is to split the difference.

The proof that your method is inaccurate is the observation that the right side of Connally's forehead appears to jump to the right in your animation. Connally's forehead is a light object that is to the left of a dark object (his wife's face, which is in shadow) and therefore is analogous to our area of controversy. You will know that you've lined things up correctly when the right edge of Connally's forehead does not move at all in the transition from 312 to 313.

As I said, it looks to me like what you've done is to split the difference. Instead of lining up the right edge of light objects, you've taken light objects and lined them up where the left edge is offset about the same amount as the right side. This splitting the difference results in you being off by half the length of the blur.

If you try it again the way I'm proposing then you will see that most, but not all, of the head movement gets eliminated by the correction, when done correctly.

In place of the two inch movement that your animation appears to show, I believe there is actually more like a one inch movement.

I don't have time this morning but when I get a break I will try to produce a new version of your animation based on lining up the right side of light objects that have a dark object next to them on their right. Or perhaps you have the time and would be willing to put it together. If we get it right then the right edge of Connally's forehead will appear to remain stationary. The left edge of Kennedy's head will, I believe, appear to move to the right about half of what your current animation implies, or about one inch.