What is "the secondary image"? I see a single image, and the flare is
most prominent along (his) left sleeve.
Look more closely. The jacket doesn't just fade into the background, as
it would if it was simply flare. There is a distinct second image to
the upper right of the original. This is most obvious around the collar
of the jacket where the dark portion of the shirt can be seen between
the main image and the secondary one. You can see the same effect in
other areas of the image as well, such as around the shoulder, where the
crease is quite prominent and visible in the secondary image. Simple
flaring would not look like this.
Can you clarify what you meant by "the CCD window"?
The CCD is encapsulated in a package just like a standard semiconductor
chip. Few CCDs are supplied unpackaged. In some cases this package is
ceramic, but in most low cost CCDs it is plastic, just like the plastic
packaged chips on your computer motherboard. For a CCD (and some other
devices such as UV erasable chips) the package has a transparent window.
For visible CCDs that window is usually glass, for some specialist
devices it may be silicon or germanium or some other exotic material
which transmits in the region of the spectrum that the device is
intended to operate.
For example, take a look at this linear CCD data sheet:
http://www.fairchildimaging.com/main/library/pdf/ccds/ccd191.pdf
This is a fairly expensive device - too expensive to be found in a
consumer grade scanner. It is in a ceramic package with an AR coated
front window.
A more typical, low cost plastic packaged CCD found in a consumer
scanner would be something like:
http://products.sel.sony.com/semi/PDF/ILX558K.pdf or
http://products.sel.sony.com/semi/PDF/ILX585K.pdf
but the Sony data sheets don't include a pretty picture, so unless you
know what you are looking for, it is hard to see.
Most people in the industry refer to this as the device "window"
because, as mentioned previously, it need not be glass. Indeed the term
glass can be misleading since it is often interpreted as a powered
element as in "spectacle glasses". I notice that you used the term in
the plural, so perhaps you were thinking this was a little more complex
than it actually is.
Is it the same as
the frame window?
No - it is intrinsic to the CCD package, the frame window is the scanner
design.
I have seen similar flares around the edges of a scan
when the film does not completely cover the frame window. Where the
frame has a sliver of opening is where the flare will occur on the scan.
This kind of flares can easily be diagnosed by scanning a small crop
that does not include the opening but includes the high shadow/highlight
contrast.
But you will get a similar effect due to veiling glare at any of the
optical surfaces in the scanner, not just from the CCD window.
If "the CCD window" means something different, can flares be diagnosed
in a similar manner, i.e. by scanning a crop, or by scanning the film
rotated 90 degrees?
Well, you could image a point source that is about 2/3rds of the way
from the centre towards the corner of the frame and see if you get
multiple point sources. An analogy to this is if you look through a
window at night, particularly a double glazed window, at an angle
towards a light source. You will see the main light source and
reflected secondary, tertiary and so on copies of it. Now, there is
another issue to deal with in the scanner, because the reflections will
be defocused. However, if it is just the presence of the CCD window
that is causing the effect then you will see multiple images in your
scan. If it is scattering from any optical surface in the scanner,
including the CCD window, then you will see flared images.
Even more so when you have had your scanner for a couple of years (or
less) and discover that it needs to be cleaned! There are several sites
showing how to do this with Nikon scanners, which usually only require
the mirror and lens surfaces cleaned. In the past you could be pretty
certain a newer, better scanner would be available to replace it, since
cleaning by Nikon (or any of the main manufacturers) can be quite
expensive. However, with the general transition to digital imaging I
don't expect that will continue to be the case for much longer. I
suspect the market for 35mm film scanners must be pretty limited already
and the release of new 35mm scanner will become a rare event fairly
soon.
Any comments on the "CCD protective glasses" causing the flares, as
suggested by one poster?
The thing to realise is that *every* optical surface will degrade the
image, since every optical surface is a source of scatter and diffusion
as well as its primary function. That is why no professional
photographer permanently fits a skylight filter over their prime camera
lens but many amateurs do. The filter protects the lens, but degrades
the image too - so the owner makes a trade-off between risk of damage
and image quality. So the fact that removing a protective window from a
CCD package improves the flare is not really surprising. Doing it
without getting dirt on the CCD and *keeping* it that way is another
issue.
If you want to go down this road, and I *strongly* advise against it,
you will also have to adjust the scanner optic to accommodate the change
in focus, since the window will have a refractive index which is
different from the free air you replace it with. Then, if you are
really lucky, the spherical aberration introduced by the change in
refractive index between the lens and the image plane might be less than
the deficiency that you started out with. ;-)
As I said before, I have never seen this effect in a Nikon scanner
personally, although I have seen veiling glare from dirty optics
(generally the mirror) on many. So I suspect that the example you gave
which demonstrated the problem was actually a faulty unit.
