[Klaus]

Then it probably was the filter anyway... The light has to be bright enough and the exposure long enough for the ghost lights to become clear (and of course the ghost light spots need to fall on a dark part). What I find most interesting is that the point-mirror-ing seems to not have happened through the optical axis in this particular image. Maybe that has to do with the IS unit.

 

With the vertical orientated bridge image the mirroring did go more or less through the middle of the image, as one would expect. The other bridge image in the gallery has the line of lights go through the diagonal of the frame, this probably explains why the ghost lights are not visible to us.

 

This is not the whole image but just a crop :-)