Valburg (lkv1@psu.edu)
Sat, 10 Apr 1999 14:42:01 -0400
At 11:03 AM 4/10/99 -0600, Larry wrote:
(...)
>At the higher density's the curve gets pretty steep and quite a bit of
compression is needed to get to say a .4 to 1.6 type density, last night I
had to compress a gray scale tossing out 88 of 255 levels to get this
density range, so thats another concern. For a platinum neg, say .4 to 2.4
it wouldnt be quite as bad but would be close due to the steep slope. Where
the curve gets applied may make a difference here, I dont yet know where
would be best.
This is where 12 bits per color per pixel (or higher) scanning of the
negative comes in very handy. In the method in Dan Burkholder's book, you
apply the curve near the end of the process, but the image is always at 8
bits/color/pixel (256 step grayscale). So, a steep curve like above will,
as Larry notes, start losing steps.
If the curve is applied to a 12 or 16 bit/color/pixel file, and then
converted to 8bits, you maintain your 256 step grayscale (assuming this is
important to the image).
If you need to perform much in the way of image manipulation by eye, you
could run into trouble working on the post-curve image. I can envision, in
this case, doing one of two things. Do all the manipulations in 12-16
bits: ideal, if you have software with the capability, but Photoshop is
still somewhat limited in what it can do with a 16 bit file. Or do the
manipulations on an 8 bit non-curve version of the file with Photoshop's
Action function recording what I do, and then applying the Action to the 16
bit curve-file. The manipulations won't have EXACTLY the same effect on
the curve-file, but they'll be much closer than doing it by eye and
guessing, and you will have preserved your grayscale. The first approach
is much to be preferred.
Mitch Valburg
This archive was generated by hypermail 2.0b3 on Thu Oct 28 1999 - 21:39:30