Not Quite Sure What To Call This

[DaveWM]

would be interesting to see if you could see it on a scope. Need a fixed image of course and then look at each line of data (horz scan rate) compare the composite video to the output of the detector,then trace thru to the CRT grids (or cathode if that is how it goes).

Did you try a composite video injection at the video amp grid (sorry if you mentioned it already)?

Maybe just a weak CRT that the cutoff is not that sharp.