[alt-photo] Re: New Platinum Prints [pretty long]

[etienne garbaux]

Davidh wrote:

>Etienne,
>Thanks for replying. Just so I understand what you're saying. The
>curve I apply to my digital negative is actually destroying some of
>the image content of the scanned negative? If so, why do we apply a
>curve at all? Why can we not merely print the negative as it is
>scanned? Assuming of course that the original negative is correctly
>developed for platinum?

You could, if you were able to translate the transmission densities 
at every point of the original negative exactly to the transmission 
densities at each corresponding point on the digineg.  But you can't, 
at least not by simply scanning and printing.  If your whole workflow 
is calibrated (scanner, monitor, and printer), you can scan a print, 
then make a reasonably matching print without any manipulation (other 
than the calibrations you did initially).  But when it comes to 
matching the UV transmission densities of a negative with those of a 
printed transparency, much fiddling is required.  For starters, the 
scanner isn't "reading" the UV transmission densities in the first 
place.  Second, the printer doesn't know how to print for 
transmission density.  Finally, the printing inks have very different 
UV properties than the original negative.  Furthermore, people who 
are making diginegs for Pt/salt/cyano typically aren't starting with 
negatives of suitable density range in the first place.  They may not 
even be starting with negatives at all, but rather with in-camera 
digital captures.

Think of it as mapping.  The scene you photograph has a certain range 
of luminance values.  You want to translate, or "map," these 
luminance values to useful negative densities, which can in turn be 
mapped to the available reflection densities of your chosen printing medium.

To do this, we start with the exposure scale ("ES") of the printing 
medium.  Using calibrated step wedges, we see what range of exposures 
gives the full range of printed tones the medium is capable of 
producing -- any more exposure is indistinguishable from the blackest 
tone, and any less exposure is indistinguishable from the lightest 
tone (paper white, or close to it).  Now, if we want the full range 
of tones the process is capable of producing to be represented in our 
print [which may not always be the case], we know that our negative 
must have a density range ("DR") equal to the printing medium's 
ES.  So, we arrange to translate the luminance values in the scene to 
the particular negative densities that will produce the print tones 
we want to represent each scene luminance value.

Not so long ago, we did this by adjusting our exposure and film 
development, and then perhaps reducing or intensifying the negative 
or masking it for printing, and finally by dodging and burning as we 
printed.  It sometimes took all that, because we have to condense or 
compress the 1,000,000:1 luminance range of the scene we photographed 
(this is about normal for a sunlit scene) down to the 100:1 (or less) 
density range that a print can reproduce.

These days, people who print digitally can use Photoshop to adjust 
the mapping from the as-captured (or as-scanned) image file to the 
negative that will be used to make the print.  This is much easier, 
and also much more flexible, than doing it chemically or with 
masks.  However, the goal is the same -- to translate or map certain 
luminance values in the scene to the desired print density values, 
using the negative as an intermediary.  Applying curves is how we do 
this.  It's fundamentally the same as using the Photoshop tonal 
controls (levels, brightness/contrast, whatever) to adjust a digital 
capture for the desired print values (i.e., if you are just printing 
digital images on paper), with two added wrinkles: (i) you have to 
understand how the characteristic curve of the printing process 
responds to the negative values, and (ii) you have to be able to 
imagine how to "do it in reverse" since you are working on a negative 
-- if you want the shadows to have more contrast in the print, you 
have to increase the contrast in the lightest parts of the negative.

In theory, you can map any scene luminance value (or digital 
representation of a scene luminance value) to any available print 
tone (with the caveat that the curve should be monotonic -- never 
reversing slope -- unless you are after special effects reminiscent 
of the Sabattier effect).  In my view, there is no substitute for 
learning enough sensitometry/densitometry to really understand how 
the mapping works, and then to test your processes to see how they 
distort the mapping so you can correct for it.  It's not very 
difficult, and once it is mastered you will truly have the chops to 
get what you want out of your photographs.

So, all that said, back to curves destroying negative 
content.  Mapping is mapping -- the person who wants to represent the 
surface of the earth on a flat surface has choices to make, because 
there is no way to linearly reproduce the surface of a sphere on a 
flat surface.  And as we learned in grade school, cartographers have 
come up with hundreds of different ways to do it, each one good for 
some particular task.  If you want to be able to visualize 
comparative land areas, you use a different projection than if you 
want to determine bearings from one place to another.  The same is 
true for mapping tonal values in photography.  So, the "right" curve 
is the one that produces the results you want -- i.e., the one that 
maps the scene luminance values to the print densities you want.  In 
general, one way I'd advise folks NOT to do this is to copy someone 
else's curve "because you're using the same process."  No two people 
ever use the "same" alt process, because there are way too many 
variables to control.  And no two monitors are the same, or scanners, 
or printers, or Pt "emulsions," or coating techniques, or anything 
else you use to make prints.  So, the only way to end up with a 
useful curve for your process flow is to test and figure it out for yourself.

Generally, one would like to calibrate one's monitor, then build a 
curve for each printing process one uses so one can just adjust the 
image on the monitor (as a positive), then let the computer figure 
out what negative densities are required to map the monitor view to 
the final print (though once again, the monitor has a considerably 
greater luminance range than a print has density range, so it will be 
a "rendition" of the monitor image, not a literal copy).  Only you 
can build such a curve, after doing the sensitometry/densitometry on 
your equipment and printing processes.  There are aids available, but 
IMO one is much better off gaining an understanding of the 
fundamentals and just doing it -- just as people who really 
understood what they were doing always got better results than people 
who "learned" the "zone" system by rote.

So, if you have a curve that really does translate (transliterate ??) 
from your monitor to your prints, great -- it is not destroying 
anything, but rather helping you to map values from your digital 
image file to the final print, thereby allowing you to do your image 
adjustment by eye rather than by figuring.  But if your curve doesn't 
produce prints that are pleasingly rendered based on the screen 
image, you need to change something.  You can futz around with the 
process to try to match it to the curve you have, but that's the hard 
way (and you may well not ever find a variation that works as you'd 
like).  Better to adjust your printing process until you get the most 
linear scale you can (for reasons I won't go into here, having to do 
with producing the smoothest tonal range), then developing a curve 
that translates from your monitor to your prints.

The problem with the "short-scale" versions of Pt is that they have 
much less linear tonal ranges than long-scale Pt.  You can 
successfully map this, if you work at it, and thereby get correct 
overall tonal rendering by using a curve that compensates for the 
nonlinearity.  However, you still won't get the smooth transitions 
that long-scale Pt can produce.  And since the gorgeous tonal 
rendering is the real draw of Pt in the first place, why settle for 
something less just because one would prefer to avoid dealing with 
how to make digital negatives of sufficient DR?  Particularly given 
the cost of the Pt process, I just can't see why one wouldn't use it 
to its full advantage -- which IMO requires using the long-scale process.

Best regards,

etienne