FotoDave@aol.com
Date: 05/25/00-02:16:03 PM Z
In a message dated 5/25/00 2:11:23 PM Eastern Daylight Time,
jeffrey.d.mathias@worldnet.att.net writes:
> Have you or do you know of any study that has investigated the number of
> distinct tones necessary for discriminating a certain range of values?
Jeff, I have read it somewhere, but I forgot the figure and forgot where I
read it from. If I find it, I will let you know.
But for comparison purpose, you can consider this: if you are scanning a
reflective print and you directly map black and white to 0 and 255,
respectively, then the theoretical values (and the natural response of
sensor) will make it such that you have 128 values from paper white to
density of 0.3 and 4 values from density of 1.5 to 1.8. Coincidentally (and
fortunately), our eyes are also more sensitive to highlight difference than
shadow difference, thus we need more values in the highlight area.
But in printing practice, if we use the above mapping, then the shadow is too
sensitive. You can change the rgb value a little bit but the shadow can jump
to another step!
For continous-tone equipment like film recorders, I calibrate it to gamma of
1.8. I will omit the math here, but that will give 80 values from white to
0.30 du and about 12 values from density 1.5 to 1.8. We might need all 80
values for the highlight, but you can see it is certainly more than 10
"zones" that you have.
The scanner sensor responds similar to the theoretical desired value, thus it
is best to scan positive than negative (because in a negative, the dark value
is scanned to have small separation but the dark values are actually the
highlights where you need good separation).
That is why if you have to scan negative, increase the contrast, scan in
12-bit, and the adjust the curve (or the scanner software might support curve
too) before you change it to 8 bit. Scanning in 8 bit will cost too much lost
in shadow.
For the math and theoretical discussion, search the Photoshop's newsgroup
archive for names like Dave Martindale or Chris Fox (he is from Adobe).
For best result and to get every details out of a negative, you should scan a
positive; but if you do that, you might as well make a good interpositive and
make contone internegative. Your original question is about making highlight
masks with printer. You might want to try duotone / multitone as suggested by
Dan. The requirement for making 2 negatives is much easier because you print
twice and you can control the contrast in each pass.
I guess the bottom line is where you want to place your control: in computer
(through curve), in negative (through mask), or in the final print stage
(through duotone / tritone).
Another method which I haven't seen discussed (in technical details) before
is to use lith film for your interpositive with careful place of tone and
control of contrast. It has a short scale (and I now find this beautiful).
You process it to low contrast (using LC-1 developer, for example), the high
density will separate well although slightly compressed, that is, it has a
nice S curve.
Now in making your interpos, you can carefully place your zone 1 on Dmin +
0.1, zone 3 on the toe, zone 7 on the linear area, and zone 9 slightly
shoulder off, thus you have a reverse S curve, which is a desirable
reproduction curve (as opposed to completely linear curve, which is desirable
in scientific application) because it emphasizes the separation of zone 3 to
zone 7 while maintaining separation of shadows and highlights.
The lith film, when processed to low contrast, has an S curve, but since you
are making the interpositive, you get the reverse S curve. If you use a
linear, contone film for interpositive but apply the above in the negative
stage, then you get the S curve. In other words, by careful control, you can
achive what you want without masking (except of course when the image
lighting specifically requires it).
I won't get into details or numbers because I don't know if you are actually
going to try it and because I know you are familiar with the zone system and
tone controls, so if you are going to try it, you will get it without the
need of detailed explanations anyway.
Dave Soemarko
This archive was generated by hypermail 2b29 : 06/13/00-03:10:21 PM Z CST