On Fri, 11 May 2001, Rob Geraghty wrote:
> OK, it sounds like some sort of aberration in the scanner lens system.
> Is there anyone near you with another film scanner you could send
> a sample slide to in order to test it? Maybe with a Polaroid scanner?
> Out of interest, does it make any difference if you insert the slide into
> the scanner the other way up?
Yes, it makes a difference... I did some further testing last night.
I turned the slide 90 degrees, and sure enough the ghost rotated 90
degrees in repect to the stars (that is if you keep the orientation of
the stars fixed) => so clearly due to the scanner and not the slide.
Furthermore, I measured the effect on about 10 bright stars in the field,
and noticed the following behavior. Along the line that is parallel with
the long edge and goes through the center of the slide you seem to have no
ghosts, but as you approach the longer edges of the slide you see these
the ghosts emerging from the stellar image, the closer you are to one
of the long edges of the slide the more pronounced the effect is. You can
understand this as internal reflections (as was suggested by Art in
the messages that arrived last night) if you consider that the scanner
scans perpendicular to the long edge of the slide. Clearly such a process
is optimized to the centerline. Actually the ghost images are also there
on the center line, but they are superimposed on the stellar images
making a nearly unnoticeable halo.
Yes, replying to several of you, I have a neighbour just down the road who
has a scanner too.. It's not a polaroid, but a Canon FS 2710. So I ran the
same picture on through his scanner - similar resolution. And boy that
must have been one of those moments, when I was happy to see an optical
distortion. A similar phenomenon was visible in the images scanned with
the Canon scanner, but here the image looked like a small comet (similar
to a coma distortion), but with two separate tails, a red one and a
green one. So instead of one blue-green ghost spot, there were two more
noticeable tails of different colors pointing in the same general
direction as the spot is in the picture scanned with the Nikon. The size
of the distortion was more or less similar to what I had in the Nikon
scanner. It appeared that the tails had a significantly higher level
than the ghost spot I saw with the Nikon. It appears that I hit the (
not so bad afterall) limit of the Coolscan IV scanner.
It seems also that this phenomenon may be a common problem to desktop
scanners. I think you should see it at any bright source (e.g streetlight)
against a (nearly) pitch black background and here only on edge of the
light that is closer to the edge of the slide. It should not effect
significantly ordinary day time images.
The web reference had one typo in it... So here they are again.
Since it appears that I'll have to live with it, are there any remedies
for removing this effect from the images? It appears that if I could scale
the image by a few % in y direction only, skysubrtact and multiply
the new image by a suitably small number, I would have a "mask" that I
could subtract from the orginal image to get rid of the effect. Can
this be done easily?