Part 3 of 3
OUTPUT DEVICES: ADJUSTING THE DIGITAL NEGATIVE TO FIT THE NEEDS OF YOUR
PRINTING PROCESS
What can be done to remedy this situation? In PhotoShop, you can apply a
curve to the image which will compensate for the inaccuracies of the
imagesetter (or any other output device). With the Curves command, you can
create a correction curve which will map any brightness value in an image to
absolutely any other brightness value. After you've finished creating an
image to your satisfaction on the screen, you simply apply a correction curve
to the image just before sending it to the service bureau. These corrections
can also be made at the service bureau with imagesetter software. If all goes
well, the resulting digital negative will have a range of optical densities
that will give proper tonal values in your contact print.
There are two ways to construct such an adjustment curve. With the Image =>
Adjust => Curves command (Command-M on the Mac), you can make a correction
curve by using the cursor to drag the curve to fit the values you want. A
more precise method is to use the Transfer Function. Use the File => Page
Setup => Transfer Button to get to the Transfer Function dialog box. There,
you will be presented with a curve identical to the one you had in the Adjust
Curves box. However, in addition you will have spaces for entering specific
numerical values for thirteen points on the curve. After you've entered
values for a curve, you can save the curve in a file on your hard disk.
There are two methods you can use for making and testing correction curves.
With the first approach, you use several experimental correction curves to
make digital test negatives, make contact prints from these negatives, and
evaluate your curves on the basis of the test prints. With the second
method, you evaluate test negatives with a transmission densitometer and
construct correction curves on the basis of the densitometer readings, with
no intervening test prints.
I have oriented the descriptions of both these methods to the unique problems
presented by imagesetters. However, the methods can be used to optimize
negatives from any digital output device for B&W contact printing.
METHOD 1: USING TEST PRINTS TO EVALUATE CORRECTION CURVES:
You will get the best results if you do all of your tests with digital step
tablets. For my test work, I use a digital step tablet with three segments.
First, there is a 21 step tablet, with densities ranging from 0% to 100% in
5% increments. Then, there are two additional 21 step tablets, one having a
single step for each of the values in the 0%-20% range and the other for the
the range 80%-100%.
How do you begin plotting correction curves for imagesetter digital
negatives? Here are the numerical values for seventeen test adjustment
curves. These curves are a good starting point for optimizing imagesetter
negatives for B&W printing. The right correction curve for your process
probably lies somewhere among these fourteen.
Curve DMax 0.5 Curve DMax 0.6 Curve DMax 0.7
============== ============== ==============
0%: 31 0%: 25 0%: 19.2
5%: 36 5%: 27.5 5%: 22.7
10%: 39 10%: 32 10%: 26.3
20%: 44 20%: 36.3 20%: 29.3
30%: 49 30%: 40 30%: 35
40%: 52.5 40%: 46 40%: 41
50%: 57.5 50%: 51 50%: 46
60%: 63.4 60%: 57.5 60%: 51
70%: 68.4 70%: 63.4 70%: 58.7
80%: 73.4 80%: 69.9 80%: 64.7
90%: 80.6 90%: 76.5 90%: 73.4
95%: 89.9 95%: 84 95%: 82
100%: 100 100%: 100 100%: 100
Curve DMax 0.8 Curve DMax 0.9 Curve DMax 1.0
============== ============== ==============
0%: 15 0%: 11.5 0%: 9
5%: 18.5 5%: 14.5 5%: 11.7
10%: 21.7 10%: 17.5 10%: 14
20%: 25 20%: 20 20%: 17.5
30%: 28.6 30%: 25 30%: 22.1
40%: 34 40%: 29.9 40%: 24.8
50%: 40.8 50%: 35.5 50%: 31.2
60%: 47.2 60%: 41.6 60%: 37.1
70%: 54.4 70%: 50.2 70%: 45.6
80%: 60 80%: 56.2 80%: 53.8
90%: 68.1 90%: 67 90%: 66
95%: 80 95%: 78 95%: 76
100%: 100 100%: 100 100%: 100
Curve DMax 1.1 Curve DMax 1.2 Curve DMax 1.3
============== ============== ==============
0%: 7.2 0%: 5.9 0%: 4.6
5%: 10 5%: 8 5%: 6.3
10%: 11.5 10%: 10 10%: 8.2
20%: 14.5 20%: 12 20%: 10.1
30%: 18.2 30%: 15.4 30%: 13.4
40%: 22.3 40%: 19.2 40%: 17
50%: 27.4 50%: 24.7 50%: 22.3
60%: 33.5 60%: 31 60%: 27.8
70%: 41.3 70%: 38.6 70%: 36
80%: 50.4 80%: 47.4 80%: 45
90%: 61.7 90%: 59.4 90%: 56
95%: 76 95%: 72 95%: 71
100%: 100 100%: 100 100%: 100
Curve DMax 1.4 Curve DMax 1.5 Curve DMax 1.6
============== ============== ==============
0%: 3.6 0%: 2.9 0%: 2.1
5%: 5.2 5%: 4.3 5%: 3.2
10%: 6.8 10%: 5.3 10%: 4.7
20%: 8.5 20%: 7 20%: 6.1
30%: 11 30%: 9.5 30%: 8.5
40%: 14.7 40%: 12.8 40%: 11.3
50%: 20 50%: 17.5 50%: 15.5
60%: 24.7 60%: 23.4 60%: 20.8
70%: 32 70%: 30.4 70%: 27.2
80%: 41 80%: 39 80%: 35
90%: 54.8 90%: 52.4 90%: 50
95%: 70 95%: 68 95%: 66
100%: 100 100%: 100 100%: 100
Curve DMax 1.7 Curve DMax 1.8 Curve DMax 1.9
============== ============== ==============
0%: 1.9 0%: 1 0%: 0.9
5%: 2.7 5%: 2.1 5%: 1.8
10%: 3.9 10%: 3.1 10%: 2.6
20%: 5.4 20%: 4.4 20%: 3.7
30%: 7 30%: 6 30%: 5
40%: 9.4 40%: 8.3 40%: 7.5
50%: 13 50%: 12 50%: 10
60%: 18.9 60%: 16.9 60%: 15.6
70%: 25.6 70%: 24 70%: 22.4
80%: 33.8 80%: 32.5 80%: 30.6
90%: 47.5 90%: 45 90%: 42.5
95%: 62 95%: 62 95%: 57.9
100%: 100 100%: 100 100%: 100
Curve DMax 2.0
==============
0%: 0
5%: 1.5
10%: 2.1
20%: 3.1
30%: 4.5
40%: 6.8
50%: 9.5
60%: 13
70%: 19.7
80%: 28.1
90%: 42.5
95%: 57.9
100%: 100
With 300 LPI digital negatives made by Command-P with their Scitex
imagesetter, these curves will each result in a negative with a D/LogE curve
similar in shape to the "ideal" curve illustrated in the above graph. They
will differ only in the slope of the curve and the DMax of the negative. As
indicated by the names of the curves, they produce negatives with a DMax
ranging from 0.5 to 1.8, in increments of 0.1. The DMin of all the curves is
identical, at 0.03 to 0.05.
If you are not inclined to experiment and test these curves, you can use them
directly on images and have Command-P make negatives for you. You must first
make test prints with a 21-step tablet (from Kodak, Stouffer, etc.) to
determine the DMax required by your printing process. Using that
information, select the appropriate curve and apply it to your image before
sending it to the service bureau.
If you want to test the entire system on your own, you must make a test
negative using the Test Curves on a series of digital step tablets. My
digital step tablet as described above measures 1.33" x 5"; so, I can put
twelve step tablets on a single 8"x10" digital negative. For your first
8"x10" test negative, you might put twelve digital step tablets in one 8"x10"
image. With this negative you can testing twelve of the curves in the
series, excluding the four which are farthest from the DMax required by your
printing process. On each step tablet, apply one of the correction curves in
Figure 4: use the Marquee tool to select a single step tablet, and then use
the Adjust => Curves command (Command-M on the Mac) to load the file of a
test curve and apply it to the selected area.
By the way, I would be glad to send you copies of the Test Curves and my
1.33"x5" digital step tablet. Send me a blank formatted Macintosh 1.4 MB
floppy and a SASE and I'll mail the floppy back to you. Sorry, I don't have
the files in IBM-PC format.
Send this test image to a service bureau and have a digital 8"x10" negative
made. These are the instructions for sending a digital file to Command-P.
If you use another service bureau, you will have to work out these details
with that bureau:
1. File Format: PhotoShop 3.0 grayscale file, 635 PPI at your final
negative size.
2. Send file as a positive image.
3. Specify whether you want stochastic screening or 300 LPI linescreen
(Diffusion Dither Bitmap screening is no longer available at Command-P).
4. Request output to film as a negative with emulsion down.
5. Call ahead to make sure that they can read your removable hard disk
(SyQuest, Jaz, Zip, etc.).
When you get the test negatives back, make test prints from them. You should
standardize your printing with a single combination of paper, coating
material, developer, contrast agent, etc. The curve which you develop will
be optimal for this single combination of materials only. Determine the
minimum exposure time to produce a pure black with the 100% step. Each test
print should be exposed for this minimum time, and no longer.
I cannot emphasize enough the importance of standardizing your printing
methods and minimizing the number of variables in your test system. There
are enough variables in PhotoShop and service bureau equipment to make this
testing procedure quite complex under the best of circumstances. Don't
complicate matters further by fiddling with your printing methods half way
through these tests!
How do you evaluate your test prints of the digital step tablets? There are
three steps in the process:
1. Look first at the 21 step tablet with 5% increments. Visually, there
should be even gradation through all the steps. In particular, there should
be good separation between the steps at the extremes of the tablet (0-5-10%
and 90-95-100%). 0% should be absolutely white; compare it against uncoated
paper to make sure that the step is white rather than light gray. The 50%
step should be a neutral gray.
2. After you identify an acceptable 21 step tablet with 5% increments,
inspect the negative of that tablet. The extremities of the curve should be
your main concern. Make sure that each individual step in the 0-20% and
80-100% 21 step tablets is different from its neighbors. You are looking
for evidence of "posterization": a few (or several) adjacent steps having
identical density. Posterization usually isn't a problem in the 80-100%
range but often occurs in 0-20% values when one of the more extreme
correction curves has been used. If you have access to a densitometer, it
can be quite helpful in confirming that there is separation among these
values, which by visual inspection may appear very similar. Posterization in
an occasional pair of adjacent values won't be apparent with most negatives
of real images. But, any more than a few pairs or triplets of identical steps
with the same density will make a curve unacceptable for some images.
You will notice that the highest DMax of the family of curves in Figure 4 is
2.0. I realize that for some alternative processes you may want a denser
negative. However, in my experience with imagesetters, posterization in the
0-20% range becomes a significant problem with DMax's of 1.8 and higher if
the DLog/E curve is relatively linear, like the ideal curve in Figure 1. If
possible, you should limit your DMax to 1.8 or less and use
contrast-enhancing agents with your printing process to decrease the optical
density you need in the negative.
This posterization in the 0-20% range is the thorniest problem I've had in
making negatives with imagesetters. I would appreciate the input from list
members regarding ways of dealing with this issue. In particular, I'd like
to hear the thoughts of those who have had hands-on experience with
imagesetters. Is posterization simply inherent in this output device, or is
there some way around it using imagesetter software (as opposed to my
technique of doing all the curve corrections in PhotoShop)?
3. If the print of the step tablet with 5% increments is visually satisfying
and there is separation of most steps in the 0%-20% and 80%-100%
negatives, you probably don't have to worry to much about the 20% to 80%
range. As long as the extremities of the tonal range are satisfactory, the
intermediate tones are usually acceptable.
You can consolidate values from several curves into a new test curve. For
example, if 0-20% from one curve looks good, 25-75% from another is ideal,
and 80-100% from yet another is best for your process, you can construct a
new curve using the optimal segments of each. In the Transfer Function box,
simply use the numerical values for the preferred segment of each individual
test curve (refer to the tables above for these values) and build your new
test curve. After you've patched together these segments, visually inspect
the new curve in the Transfer Function box to make sure that it is smooth,
with no abrupt transitions; you may have to make minor adjustments to some of
the thirteen data points to insure that the new curve is smooth.
If you have a flatbed scanner, scan the print of your ideal step tablet to
use for reference in future testing. Comparing the electronic densities from
scans of your test step tablet prints with those of your benchmark step
tablet print takes much of the judgment and guesswork out of the testing
procedure.
METHOD 2: USING THE TRANSMISSION DENSITOMETER TO
EVALUATE CORRECTION CURVES:
If you know in advance the desired D/LogE curve of your digital negative
(such as I have given for Pt/Pd in the graph) and have access to a
transmission densitometer, there is another method you can use to evaluate
the test curves. This technique is quicker and more precise than the method
just described. It is the approach I used to develop the family of curves in
the tables. It is rather complex to explain, but in practice is quite simple
to do.
Make a digital step tablet with 101 segments, one for each PhotoShop density
value from 0% to 100%. Have your service bureau make a negative of this step
tablet, with no correction curve applied.
With a transmission densitometer, determined the optical density of each of
the 101 steps in your digital negative, and make a table of these values.
Make another table giving the desired density of your negative at each of the
thirteen set points in the Transfer Function box (0, 5, 10, 20, 30, 40, 50,
60, 70, 80, 90, 95, and 100%).
In the following manner, construct a new test correction curve based a
comparison of these two tables: For each of the thirteen set points in the
new curve, find in the Uncorrected Curve the step whose density most closely
matches the desired new value. For example, suppose that for Pt/Pd the ideal
optical density for 30% is 1.25, and that the closest density in the
Uncorrected Imagesetter Curve is at PhotoShop density of 7%. In the Transfer
Function dialog box, use 7% as the new value for 30% in your new correction
curve. In this manner, determine values for the other twelve set points.
The slope of the uncorrected D/LogE curve is very steep in the 0-15% range,
resulting in large changes in optical density for each 1% change in PhotoShop
density. Consequently, you may have to do some interpolation and guesswork
to get the correct value for the densest areas in you new negative. In
making interpolations, you can use fractions of percents (10.7%, etc.) in the
Transfer Function dialog box.
Using your new correction curve, have a digital negative made of a digital
step tablet. With the transmission densitometer, measure the density at the
thirteen set points to see how close you came to your desired curve. As I
discussed above, make sure that there is only minimal posterization in the
0%-20% and 80%-100% segments of your test curves. You may have to make
another generation or two of curves to fine tune your correction curve.
------------------------------------------
Because of space constraints, I have been able to give only a brief
description of these calibration processes. Also, it is quite difficult to
describe these techniques using only ASCII files and EMail, with no
illustrations. I'll try to answer any of your questions you may have about
all of this.
CONCLUSION:
If you are just beginning to use computerized imaging, the learning curve is
steep. I would compare the effort required to learn digital techniques to
the task of mastering your chosen alternative process in the first place. To
anyone who has a technical bent (which includes most of us working with
alternative processes), it shouldn't be a problem. The biggest difficulty is
the lack of information regarding the use of PhotoShop techniques with B&W
materials. The Adobe manuals and the aftermarket books leave much to be
desired for B&W photographers. There are several PhotoShop tricks for working
with monochromatic images which I do not have room to describe here. As
PhotoShop is designed mainly for color work, many of these are poorly
documented; some of the most useful are completely undocumented. It is true
that with a little help you can pick up the PhotoShop basics pretty quickly;
but, the subtleties and tricks take a while to learn, especially if you're
working by yourself.
Sceptics point out the technical problems and high costs of digital methods,
the necessities for software updates, planned obsolescence, and the like. At
least for now, we are pushing the limits of the imagesetter to make good
negatives for alternative processes, forcing the machine to do a task it was
not designed to do. All of this is true; but, my response is that no
powerful and complex method is without its difficulties. I have had all
sorts of technical snags with my digital work, but they are no worse than the
troubles I have making Pt/Pd printing work in a consistent manner. Take, for
example, the problems created for Pt/Pd printers by paper manufacturers.
Talk about planned obsolescence! Just when you think you've found the
perfect paper, they change it without telling you, and you are once again
looking for a new one.
Despite the acknowledged difficulties in applying digital methods to
alternative processes, is it worth the trouble? Absolutely! When everything
is working, my digital Pt/Pd prints are indistinguishable from conventional
Pt/Pd prints. And most importantly, we should not forget what attracted us to
computerized imagery in the first place: digital techniques provide control
and freedom which are simply unimaginable with conventional methods.
Charles Palmer
1506 Park SW
Albuquerque NM 87104
USA
CHPalmer@aol.com