Re: Back-exposing on plastic (was: Re: Gum transfer

Yves,

I agree with your entire post.... except for one thing. While color files
can have millions of colors and printers can reproduce a lot of them when
making a standard color inkjet print, there isn't currently a means of creating a
digital negative with more than 256 tones. No matter how you may try, there
isn't currently a way to do it with a single negative printing even a 16 bit
gray scale image. But if you discover one, I'll be the first to buy your book.

But for Tri Color Gum, I doubt if anyone would ever notice, because you have
a lot more possible tones due to printing 3 negatives.

Actually printing something like AZO would be most likely to give you
posterization—you are spreading 256 tones across a wider print density range. (log
.03—log 2.2?)

There are, however, colors that will give you less than 256 tones, because
Photoshop can't parse them into 256 tones.

It's Dan Burkholder, not Don Burkholder who wrote Making Digital Negatives
for Contact Printing. A great book—I have both editions and recommend the book
to my workshop students. I think Don Burkholder is the guy who took the
photos in New Orleans and made some comment about Cheney heheheheh

Best Wishes,
Mark Nelson
Precision Digital Negatives--The Book
PDNPrint Forum at Yahoo Groups
www.MarkINelsonPhoto.com

In a message dated 5/2/06 5:12:04 PM, gauvreau-yves@sympatico.ca writes:

> Katharine,
>
> say we'll take a little ride and explore a few things together. I think we
> agreed a while back that a gum emultion react to light in some proportional
> manner, more light equal more pigment get caught in the matrix of gum that
> is created by the interaction of dichromate and gum at the molecular level.
>
> Below you say: "No, you cannot, once you've got the right emulsion
> <basically obtain any tone you want between the DMax and the Dmin of the
> print> by adding a curve". Ok, your right it is absolutely true that I or
> anyone else for that matter, can't obtain a continuous tone "gradient" from
> Dmax to Dmin using a curve in a digital world, because in this world there
> is only a limited number of steps (in the general sence) one can achieve. If
> we speak of an 8 bit black and white image the maximum number of step is 256
> (0..255). In the digital world you can increase this to about 16 million
> step if you use a 24 bits color image.
>
> Basically, you can stop me about here if you think that 16 million steps if
> not enough to be called continuous in practice but in theory I agree with
> you it is far from continuous.
>
> Now another point, say we agree that as you said below only 10-12 step out
> of 21 can be obtain from the best emultion we can make in a single exposure.
> In terms of negative density this mean a maximum of 12 steps * 0.15/step =
> 1.8, wow! This is far out, it would seem only carbon and silver gelatin can
> better this according to Sandy. But lets say to please every one that we use
> is figure or 1.20 above paper white. I don't know about you but I don't own
> a densitometer, instead I use my Gossen light meter equiped with a fiber
> optic probe. The best I can read with it is about 1/6 of a stop or about
> 0.05 in density which is obviously insufficient but lets say we consider a
> density resolution of 0.001 instead, this is about 3/1000 of a stop. Can we
> agree that this is plenty enough to be call continuous in practice, you say
> NO because this is only 1200 steps from 0 density above paper white to 1.200
> above it and I kind of agree with you, we can do much better, say we use a
> 0.0001 resolution, this would make 12000 steps which is much better don't
> you think.
>
> I'll admit here, I haven't read the PDN book nor Don's book, both of which
> are about making digital negative and lets say we wont limit ourselve in
> anyway and try to reach for the top, meaning we'll have to find our own way.
> As I said earlier, a color image needs at least 24 bits of information to
> represent a sufficiently large amount of all the colors we can observe in
> nature or even man made ones. This is 16777216 different colors, this is way
> to much, we only need 12000 out of these but is it possible to find 12000
> colors out of these 16 million such that each of them give us a constant
> 0.0001 difference between each consecutive color when ordered properly of
> course. From here, there is a short way and a long way, the short way is
> accepting for a fact that it is possible to generate at least 12000
> increasing density steps with a constant step size of near 0.0001 value in
> density. The long way is to proove this using the Beer-Lambert law and this
> would take more time then both of us have for now.
>
> Another mistake I've made is I use "the curve" where I should have used "the
> curves", imagine for a moment that I choose a Hue at max saturation and max
> lightness that when given to the printer these values produce a density of
> 1.2000, keeping this same Hue we can vary both saturation and lightness in
> such a way as to produce a linear decrease in density with step of 0.0001
> from 1.2000 to 0.0000 and obtain our objective 12000 steps of which and
> unless you have very sophisticated equipment you wouldn't be able to discern
> a large amount either because of the limit of your instrument or because the
> printer hardware software isn't configured to produce various densities but
> various colors as best as possible mechanically. We can find all the excuse
> we want but in the end I have absolutely no reason to believe it is not
> possible to have a printer produce a large enough number of density values
> that the 256 different densities we can obtain from an imagesetter would
> seem like kid stuff.
>
> I think you would also agree with me that controling the density
> distribution on a negative is effectively controling the distribution of
> exposure we give a certain print and indirectly we also control the tonal
> distribution as well. If you limit yourself to 8 bit or 256 distinct values
> of which only a partial number would be useful to control our density
> distribution then I'm with you, there is no way I can produce any tone I
> want. It's like using a boxing glove to clean my nose. But if like me, you
> think that out of the 16 million plus colors a printer can produce, the
> exact figure is probably much less then that but even say 10% of this is
> still well above the million mark, we probably need only a relatively
> limited number of distinct tone to give the illusion of continuous tone. I'm
> sure the actual number is below 250 as examplified by imagesetter negatives
> where some values produce a much to high density to be useful.
>
> Still convinced I can't produce a negative with a large enough number of
> density values in the appropriate range to produce a print with an apparent
> continuous tonality gradient. Well think of this, with my Hue example I
> purposely limited myself to a subset of the potential 16 M colors because I
> used a fix Hue. Now there is a software call Gimp and with this little
> fellow I can create program that can use all 24 bits or all the 16 million
> colors. There would be obviously a long study to learn the relationship
> between these 16 million colors and the density they each produce. Enough of
> this nonsense, I'm sure we don't need no where near a million colors or
> densities and I think I've made clear enough that if an imagesetter can
> produce fine prints and this with less then 256 distinct densities, making a
> digi-neg from a 24 bit color file and requiring only 256 distinct densities
> from potentialy of much more, should be a walk in the park even in a wheele
> chair.
>
> Regards
> Yves
>
Received on 05/02/06-08:30:43 PM Z