The frames were transformed to be as if they were made with the camera was stationary and pointed with the azimuth the same as frame 7 but with 0 elevation. The car was assumed to be traveling in a straight line that is perpendicular to the camera (in frame 7).  Several parameters had to be determined simultaneously. It would help to have reference photos from the spot, but the parameters are probably close enough that there is not much error. When the sequence restarts, it is apparent that the transformation is not quite right as there is a slight shift in alignment. Here is the sequence made from the frames.

Transformed sequence

The limo positions were determined by measuring positions around the front wheel and then fitting a circle to them. The limo position is the center of the circle. The outer edge of the whitewall on the tire is probably what was measured. It is hard to distinguish the whitewall from the rim but there is a lighter colored circle there. The radius was determined by fitting measurements of all frames with circles with individual centers and a common radius.  This is the result of a single measurement.

The direction of the camera for each frame was determined by aligning background features.  Corresponding positions were found by comparing the ramps at contrast boundaries. . 

This is a graph of limo positions.

This is a rough velocity graph made by subtracting the position in one frame from the next. The head shot frame is 22.  The vertical scale is feet/second.

This is the same data after smoothing with a radius of 3 frames. The graph shows a decrease of about 1.2 fps between frame 18 and 22 (the head shot frame) but also another dip after that. This rest of this initial write-up looking for a cause of the dip in velocity starting in frame 29.  Perhaps Nix started leaning to the left when he started panning.  He would have had to move about 2 inches in 0.5 seconds to produce what is seen in the graph.  The camera rotated horizontally by about 0.6 degrees between 29 and 38.  That would translate the camera but even if Nix was rotating the camera so that the focal point was 24 inches from the center of rotation, something that would not be natural at all, then he would only translate the camera by about 0.25 inches.  So he would have to be leaning.  If so, then the rest is moot.

The dip in velocity starting at frame 29 must be an artifact since there is a corresponding increase a few frames later, but I can't pinpoint the cause. This corresponds fairly well with an increase in panning rate. In the following graph, moving up is moving the camera to the right and down is to the left.

The panning accelerates to the left starting at 26 but the dip in velocity does not begin until 29. The following graph is a measure of the panning rate vs. the movement of the limo across the frame. When the graph is moving down, the panning is not keeping up with the limo. When it is moving up, the camera is panning faster than the limo.

Starting about 27, the panning is faster than the limo and the limo is moving backwards in the frame. This is probably the underlying reason for the anomalous dip in velocity, but I don't understand why it is happening the way it does.