Several settings, like for Auto Contrast, in Photoshop can be set to a
user preference. The location where that is done depends a bit on the
version, but with Version 8 (CS) it can be found e.g. in the
Levels/Curves dialog under Options. There the Clipping points can be
set, also to 0%, and they will apply to the Auto functions as well.
For others reading along, after opening the Levels or Curves dialog
box holding down Alt changes the Cancel button into Reset and Auto
into Options. At least that's how it works on my PS version 6
(Windows)
Even though changing the clipping may change the overall color balance
slightly, Auto still suffers from the same assumption that, for
example, highlights in the original scene were truly white. If in
reality they were not, then any Auto setting will actually introduce a
cast.
14-bits can span a linear density range of 4.21. Should be enough for
any slide film. Even if we assume 1.5-bits of noise, there is enough
room for a D=3.76. If the slide is underexposed, there is significant
scene luminance compression in the shadows. That may be hard to
recover, unless extreme tonescaling is applied to that part of the
response curve. Just for that, the additional bits may allow a less
posterized result (but still with very poor color accuracy).
That's exactly the case with Kodachromes. I bought a new scanner
precisely because (on paper) an LS-50 should give me enough dynamic
range. In practice, however - at least in case of my notorious
Kodachromes - it does not, and that's even with the 1.5 bit headroom
for noise.
I don't know if you've been following, but I now scan my Kodachromes
twice as a matter of course. Once at AE to get the highlights, and
then a second time at +2.0 to +3.0 ev to get the detail out of
shadows.
This works just perfectly, although it's very time consuming. It also
took me a lot of time to get the workflow right because I have to
sub-pixel align the images and, more importantly, "color calibrate"
the two images. This is something the conventional "contrast masking"
doesn't address at all which is why it doesn't work with gradients. I
needed a method that does, and does it automaticaly so I don't have to
agonize over each image.
Now that I got all that worked out I've essentially turned my 14-bit
scanner into a flexible 16-17 bit scanner. A "dynamic" dynamic range,
to coin a phrase... ;-)
I think the output of LEDs has improved enough over the years to allow
an other compromise; smaller aperture versus faster scan. I'm not sure
what Nikon did in the latest models.
In my case, the slides are extremely warped. I mean, some sag as much
as 3-4 mm in the middle. I don't know the units Nikon uses in its
focus setting but I get a significant difference between the middle
and the corners of the slide.
In the end, I just set the focus manually so the subject is
sharp(ish). This should improve once I move to unmounted slides
because they are not only flatter to start with but the film strip
holder flattens them even more.
Of course, I could rip up the Kodak cardboard mounts and use
anti-Newton frames but that's very "dangerous" not only risking
damaging the slides but also releasing "tornadoes" of dust particles -
so I chose the "lesser evil"...
Don.
