'hybrid' digi negs

[Colin Graham]

I’ve been experimenting with a hybrid of 'colorized' negative processes and Quadtone RIP. Since the ink control with QTR is so great, theoretically it should be possible to produce any color that routines like RNP or PDN can produce, but from within the QTR interface.

I’ve long wanted to combine the smooth highlights of the colorized negatives with the even shadow gradations and the very minor necessary curve adjustments possible with QTR and the light ink set. Highlights can problematic with QTR when using all the dark inks because the UV blocking characteristics of each of the dark inks is so different, which effectively exaggerates the grain and dither of the printer. It seems like a matter of just finding the equivalent proportion of CMYK inks plotted of the best highlight blocking colors in an HSL array and plugging them into the QTR interface.

From my early experience with colorized negatives, the less black, the smoother the tonality so I've started with the 'purer' greens and reds, no black inks at all. So far the results are encouraging. I am still using LK along with LC and LM for the shadows and lower mids as they seem to have more uniform blocking characteristics.

Anyway, just getting started with this but thought I'd share the idea. I wouldn't be surprised if someone has tried this before but I've not seen any mention of it anywhere. Any thoughts and observations most welcome.

 

[sanking]

Ive been experimenting with a hybrid of 'colorized' negative processes and Quadtone RIP. Since the ink control with QTR is so great, theoretically it should be possible to produce any color that routines like RNP or PDN can produce, but from within the QTR interface.

Interesting approach. Please keep us advised of your progress. If I knew how to exactly duplicate the PDN color of G-255, R-110 with the Epson 3800 it would certain save time to print with QTR. As you say, it should be possible to get any color you want since in theory you have almost complete control of the ink distribution in QTR.


Sandy King

 

[donbga]

I’ve been experimenting with a hybrid of 'colorized' negative processes and Quadtone RIP. Since the ink control with QTR is so great, theoretically it should be possible to produce any color that routines like RNP or PDN can produce, but from within the QTR interface.

I’ve long wanted to combine the smooth highlights of the colorized negatives with the even shadow gradations and the very minor necessary curve adjustments possible with QTR and the light ink set. Highlights can problematic with QTR when using all the dark inks because the UV blocking characteristics of each of the dark inks is so different, which effectively exaggerates the grain and dither of the printer. It seems like a matter of just finding the equivalent proportion of CMYK inks plotted of the best highlight blocking colors in an HSL array and plugging them into the QTR interface.

From my early experience with colorized negatives, the less black, the smoother the tonality so I've started with the 'purer' greens and reds, no black inks at all. So far the results are encouraging. I am still using LK along with LC and LM for the shadows and lower mids as they seem to have more uniform blocking characteristics.

Anyway, just getting started with this but thought I'd share the idea. I wouldn't be surprised if someone has tried this before but I've not seen any mention of it anywhere. Any thoughts and observations most welcome.

Has anyone posed the question to Roy Harrington?

I'm not sure there is a way to deconstruct HSL values into CMYK or LAB values other than to pluck the values from PS's Color Picker.

How that would get plugged into ink limit values is most confounding. But yes I've pondered this question as well.

Don

 

[gmikol]

I think you're on to something here, Colin. If this were possible it would be like the "Grand Unified Theory of digi-negs".

If there was some easy way to overprint multiple colors to build an HSL-like array in QTR, it would be fantastic, but I don't know of any way to do that, other than multiple passes with different inks in various positions. And how do you build a 8-dimensional array, anyway?

Without those gymnastics, we're faced with a "black box" we don't have access to...the internal RIP of the printer.

I had wondered about whether it's possible to extract or generate the actual ink levels from the Gutenprint drivers. But I don't run a Mac, and have never used Gutenprint. And their color generation algorithms are not identical to the OEM's. But it's probably as close as you can get.

My $0.02.

--Greg

 

[Colin Graham]

My approach was admittedly a little rudimentary- I took the CMYK values from HSB sample and use similar percentages in the QTR setup. For example the green that Sandy mentions is 76% cyan and 100% yellow. I always has good luck with green negatives dating back to a HP 9180, so I started at a 'known' QTR density from previous profile for the shadows and tried a limit of 40 for yellow and 30 cyan inks for the highlights, then increased the densities until I get decent blockage at my basic exposures. Right now I'm at about a 60Y/30C mix of the dark inks and the steps are looking fairly linear already without any curving. It's certainly not a pure translation of a sampled HSB color, but it's workable.

I'm starting to wonder though if the ratio of the C, Y or even M inks are as important as simply leaving the K out (speaking from strictly a rendering standpoint.) The red versions look just as smooth, but would require a lot more ink to be as effective as the cyan/yellow mix.

Given a long enough time line of testing, I bet there might be a way to extrapolate HSB coordinates empirically, but sure hope I don't have to prove it..

 

[Loris Medici]

Colin, first of all thanks for the thought / tests and sharing (present and future - in advance) results. I'm really interested with the results you're going to get in the future - even if I don't have an Epson printer, the information is definitely useful...

Just wanted to mention:
If the figures below were gotten by using the color sampler tool in photoshop, then they don't reflect the real proportion used by the printer driver / RIP; they're only correct according to the "default CMYK profile selected in Photoshop color settings dialog". Real / actual proportions may differ...

Regards,
Loris.

...
For example the green that Sandy mentions is 76% cyan and 100% yellow.
...

 

[Joe Lipka]

One of the forum members, mkochsch, did all this work a few years back. It should be here somewhere in the forum or archived. The basic theory is you can't put enough black ink on a digital negative to get the proper density. You can add a color layer to block the UV light which, when added to the black ink give you enough density to print correctly. I show a little bit of how this works on my web site.

Click on over to my web site, scroll down a bit on the right and click on the digital negatives link.

 

[Ben Altman]

Yes, I've been working on a system on-and-off for several years that uses Excel and QTR to define the use of each ink at every density. It's coming along.

Currently I'm straightening out a few minor issues and working on making it user-friendly enough for others to try. But I'm getting very good test strips and it does not take long to get into the ballpark.

Recent advances have come in two areas:

First, to tackle the problem of the differing densities of dark inks that Colin mentions, I'm printing the QTR ink separation sheet (slightly modified) with the process - POP palladium in my case - and using those densities as input. Then I match those densities to each other. So for a desired print density, the program will select dots of each ink that all look the same to the POP Pd and add them together to give the desired density. A light area on the print might be something like 25%Y, 26%LK, 19%C, and 14%PK on the negative, for example. In the near future I'll try building a profile with MK instead of PK and see if my method gives smooth highlights with that ink.
Interestingly this approach produces a pretty good green color for dense part of the negative, corresponding to what others have found by trial and error to be a good combination. For dark tones I use the less powerfully blocking inks and utilize the program to blend the transition in the mid-tones.
Earlier versions used direct measured UV densities of the QTR ink separation sheet, but I find the response of the POP process differs quite a lot from that of the UV densitometer. By measuring results from the actual process and the actual printing set-up, the accuracy is better.

Second big advance is using densities measured from a test wedge printed with a trial profile as input for generating the next trial profile. Essentially the program takes values from an "ideal" profile and finds those values on the test wedge data. It then uses those points to to build a new profile. A certain amount of hand smoothing is required, but I have developed tools to adjust and smooth the ink profiles individually and in groups.

One nice thing about this approach is that the user can select the "ideal" profile desired; if you prefer to emphasize micro contrast in the mid-tones, or get good separation in the dark areas, or whatever your needs, you can construct a profile to do what you want.

The system should adapt easily to other processes and light sources. In fact I plan to test with S/G soon.

Difficulties with the system: You have to have a good reflective densitometer. There's a certain amount of computer work to do, including in transferring the profile to QTR. The system works best if the initial input data is good, so you need a stable printing method and careful measurements of the ink separation print, preferably repeated over two or three trial prints and averaged (although I don't think the ink matching is very sensitive, so you can plough through with data that's a bit off and just smooth it out by hand). The effect of the 'thin' inks is hard to measure accurately, but fortunately they are less important to smoothness.

So - if one or two people would like to beta-test the system, let me know. It may be a few weeks before I have it ready.

Ben

 

[Colin Graham]

Loris, thanks, I am finding that out rather quickly- my initial percentages were even visually out of whack following the CMYK sample. It does take some playing around. I used an abbreviated step so I could print 5 or so estimates at once to save time and zero in more quickly. Like I said my approach was fairly clumsy so far, and the CMYK dropper won't give you pure HSB conversions as distinct drivers interpret them, but I'm hoping a little more work might start show a pattern to help with the conversions.

Joe, my post was mainly about trying to blend existing methods of negative printing to use the best of the available methods. But I am getting away from black inks altogether for the highlight blocking. I don't think they're necessary and the UV blocking differential seems to just exacerbate the dither, so why fight it if the other inks can give the density you need. I never got the best rendering and tones with K-rich colors in the RNP negs on my HP printer, and the same seems to be true on my 3800 with QTR.

 

[Colin Graham]

Just to be clear, I'm not trying to invent a new negative system or detract from the existing ones, just trying to find a workaround for a problem specific to my work. I sometimes get highlight dither with carbon transfers on unforgiving surfaces. It might not be worth the trouble for anyone else to pursue. But so far I like the results I'm getting so I'll keep messing with it and gives updates when warranted.

 

[Ben Altman]

Colin, I should have said in my previous post that I think your approach sounds promising. You are in fact working on a similar principle to me and with a much simpler approach. It's good to hear that you also see a problem I am attempting to solve, which is essentially balancing the inks in a good way.

I'm basically doing this because I find it a fun problem and I can use a good solution in my printing. Most people will find my approach too techie, I suspect. I could probably have made a lot of good prints by now with PDN or Ron Reeder's system or similar...

I should also say that the information people share here on Hybrid has been and continues to be very helpful. Having a lot of people thinking about a complex problem from different angles is incredibly powerful.

Best, Ben

 

[Colin Graham]

I agree completely Ben. Many thanks for your input, it's just the sort of info I hang around here for.

 

[sanking]

One of the things about QTR that has always perplexed me is that I have not found a way to anticipate which combinations of inks will print smoothly with a given process. It seems that you rather have to first figure out the inks to give you a certain UV blocking, and then do a second stage of testing to see if this prints smoothly.

With PDN the two things are done in one step. You print the tonal palette at the right exposure and if you have a step that prints with the right UV blocking you can quickly determine visually buy looking at the surrounding steps whether the color will print smoothly.

I spent several months trying to create a QTR profile that would print smoothly with carbon and never got it. I found a good color on the first try with PDN.

But clearly if you can print with QTR you save several steps and a fair amount of time.

Sandy

 

[pschwart]

But clearly if you can print with QTR you save several steps and a fair amount of time.

I don't get the fascination with doing things the hard way
Where are steps and time saved with a QTR workflow? More importantly, can anyone definitively demonstrate that a QTR negative produces better prints than a well-done neg produced using another method? This is the bottom line for me. Full disclosure: I have spent a huge amount of time over the years experimenting with QTR negatives, but I currently use other methods. It is hard enough for me to be productive without QTR complications:confused:

 

[sanking]

I don't get the fascination with doing things the hard way
Where are steps and time saved with a QTR workflow? More importantly, can anyone definitively demonstrate that a QTR negative produces better prints than a well-done neg produced using another method? This is the bottom line for me. Full disclosure: I have spent a hugh amount of time over the years experimenting with QTR negatives, but I currently use other methods. It is hard enough for me to be productive without QTR complications:confused:

Phil,

I generally agree with you about PDN versus QTR, but the fact is that there are several more steps that you have to go through to print a negative with PDN than QTR. It is not a deal breaker but several more steps for me mean several more possibilities to screw up!!

However, I am printing carbon transfer at this time with PDN and am very happy with the results.

Sandy

 

[Ben Altman]

One of the things about QTR that has always perplexed me is that I have not found a way to anticipate which combinations of inks will print smoothly with a given process. It seems that you rather have to first figure out the inks to give you a certain UV blocking, and then do a second stage of testing to see if this prints smoothly.

With PDN the two things are done in one step. You print the tonal palette at the right exposure and if you have a step that prints with the right UV blocking you can quickly determine visually buy looking at the surrounding steps whether the color will print smoothly.

I spent several months trying to create a QTR profile that would print smoothly with carbon and never got it. I found a good color on the first try with PDN.

But clearly if you can print with QTR you save several steps and a fair amount of time.

Sandy

Hi Sandy,

These are the problems my latest approach is designed to solve: first, matching the use of the color inks so they all look the same to the printing process at each print density (this should in theory give smooth tones automatically, but is not yet proven in practice), and then second arriving at a profile with smooth transitions and the desired internal contrast with fewest iterations of print-measure-refine-print.
I'll post again when I've cleaned it up a bit and done some more tests.

Philip - doing things the hard way is more fun! I do enjoy the process, but I guess I do hope that I'm close to a point where I can just push the buttons and make good negatives, and sometime soon...

Best, Ben

 

[pschwart]

Phil,

I generally agree with you about PDN versus QTR, but the fact is that there are several more steps that you have to go through to print a negative with PDN than QTR. It is not a deal breaker but several more steps for me mean several more possibilities to screw up!!

However, I am printing carbon transfer at this time with PDN and am very happy with the results.

Sandy

I don't get it -- I have a saved layer group with a step tablet, an inverse layer, a color layer, and a correction curve. I drag this to the top of my Photoshop layer stack and print. Why is this more steps than having to flatten a copy of the image, save it as a tif, then print it from QTR?
Not being argumentative, just curious ...

 

[Colin Graham]

Sorry, I just don't see the difficulty with QTR, it's just a different tool. You control the ink carts separately to get the densities you want- that's just very cool if you ask me. But I understand that many might not want or need that level of control.

But I hope this thread doesn't devolve into an argument about the different negative systems. It was started in a spirit of how to combine them for the best possible results.

 

[sanking]

I don't get it -- I have a saved layer group with a step tablet, an inverse layer, a color layer, and a correction curve. I drag this to the top of my Photoshop layer stack and print. Why is this more steps than having to flatten a copy of the image, save it as a tif, then print it from QTR?
Not being argumentative, just curious ...

Phil,

Well, you appear to have automated things compared to the way I work.

But here is what I do.

PDN
1. Load the image file
2. Change mode from Grayscale to RGB
3. Invert the file
4. Apply .acv curve
5. Create new screen color layer
6. Fill the layer with the PDN color
7. Print

QTR
1. Load the image file
2. Invert
3. Print

Sandy

 

[pschwart]

Sorry, I just don't see the difficulty with QTR, it's just a different tool. You control the ink carts separately to get the densities you want- that's just very cool if you ask me. But I understand that many might not want or need that level of control.

But I hope this thread doesn't devolve into an argument about the different negative systems. It was started in a spirit of how to combine them for the best possible results.

I have said this many times before -- I am not a QTR basher. But I am the guy who likes simplicity, as long as it is not at the expense of print quality.
Here's what settled it for me: I can profile a new printer from scratch to print colorized negs using the Epson driver in a couple of hours. These discussions about QTR profiling demonstrate that there are lots of variables and considerable complexity in the process. That doesn't mean it's not worth investigating or can't provide excellent results; it just hasn't turned out to be my chosen method. I should also say that I own PDN but I use my own workflow for my colorized negs, and I don't use colorized negs for all my printers. OK, no more comments from me in this thread

 

[Colin Graham]

I have said this many times before -- I am not a QTR basher. But I am the guy who likes simplicity, as long as it is not at the expense of print quality.
Here's what settled it for me: I can profile a new printer from scratch to print colorized negs using the Epson driver in a couple of hours. These discussions about QTR profiling are demonstrate that there are lots of variables and considerable complexity in the process. That doesn't mean it's not worth investigating or can't provide excellent results; it just hasn't turned out to be my chosen method. I should also say that I own PDN but I use my own workflow for my colorized negs, and I don't use colorized negs for all my printers. OK, no more comments from me in this thread

Hey, that's fantastic- I'm really happy for you Phil. But if you're not here to share, why are you here?

 

[sanking]

I should also say that I own PDN but I use my own workflow for my colorized negs, and I don't use colorized negs for all my printers. OK, no more comments from me in this thread

I think discussions about the relative merits of the two systems is not a bad thing. At the very least the information should be useful to those not currently making digital negatives, and even those who have already used both systems may learn a thing or two.

However, I definitely see pros and cons with both systems and look forward to hearing more from Ben or Colin, or others who may pursue this type of experimentation.

And BTW, I too am a "Keep it Simple, Stupid" kind of person. But a bit of an adventure outside of our comfortable range rarely does permanent damage.

Sandy

 

[Ron-san]

Friends-- This thread started out being a discussion of how to determine the best blocking color before morphing into QTR versus the world.

I would like to return to the issue of blocking color. Before any of you embark on another six months of testing for the perfect color, consider the experiment I did this morning. And note that this experiment does NOT require a densitometer.

I used the Calibration Mode in QTR, and a sheet of Pictorico, to print out each ink in my Epson 3800 in steps from 0 to 100% of what the printer can lay down. I made this printout first using Photoblack as the dark black ink, then repeated the printout using Matte black. Each printout was then contact printed onto an emulsion of pure palladium (with a touch of NA2 to keep the whites bright), exposing for my standard exposure which just gives maximum black through a sheet of clear Pictorico.

First look at the contact print made with Matte Black as the dark black ink. Among the dark inks (mK, Y, C, and M) Yellow has almost the same blocking power as mK. The other two (C and M) are at least 10-fold weaker and are nearly useless. In QTR you could increase the limit of the weak C and M inks (relative to mK and Y) but long before you got them balanced you would have so much liquid on the Pictorico as to cause puddles. If you turn off mK you only make things worse because there would then be only one strong ink (Y) and a near certainty of getting grainy high tones. My own approach would be to keep mK and Y equal and increase both C and M until there was a danger of liquid overload. I do not see how you could do any better with this particular inset.
As for the light inks, only LK has enough blocking power to be of much use. By looking at the original print I conclude that I would turn off LM as useless, and increase LC and LLK relative to LK until near liquid overload. That would be about as good as you could do.

The Matte black contact print was a shock to me. I have done this experiment many times before but with quite different results. Previously Y was only 1/2 as strong as mK and both C and M were strong enough to make a contribution. I can only conclude that Epson in its wisdom has made a change in the ink formulas which adversely affects their value as UV blockers. This might explain why I keep having to redo my QTR profiles from time to time. You really ought to check your own inks.

Now look at the contact print made with Photo black as the dark black ink. Everything is the same (thankfully) except that pK is a weaker UV blocker than Y (50% pK gives about the same density as 30% Y). Currently I am using pK, throwing in enough Y to get near the density I need, and then upping C and M as much as I dare. You can probably think of other schemes based on these contact prints.


To my mind, the theory behind all this is that for maximum tonal smoothness you want the inks to fill in the holes around each other as much as possible. Ideally I think you would want four dark inks of equal UV blocking and four light inks of lesser but all equal blocking power. Nothing in that theory or in this experiment would lead you to green as the preferred color. And I have never seen any data to support the choice of green. And the choice would depend upon the exact ink set anyway. Epson seems lately to have gone out of their way to make their inks less than optimal for UV based printing (for the Ultrachrome K3 inset, anyway).

Cheers to one and all, Ron Reeder

 

[Colin Graham]

Interesting..That would make sense about the reformulation. In my tests plain yellow, even set at 70 limit, was fairly weak as a blocker unless there was a little cyan to the mix, even only 15 limit on dark cyan added to the yellow made all the difference.

In 2 days of testing, getting rid of the K in the highlight blocking set has made quite a difference is both the visible smoothness of the printer's output and the smoothness of the final print as well. Clearly individual results can vary quite a bit.

About green- in my experience with colorized negatives and the epson driver/ps color management green was generally the smoothest rendered, both in printer output and in the print as well. That's why I chose to begin my test within that hue, nothing more to it than that.

Just to repeat though, I generally get wonderful results with the full compliment of inks with most processes and papers. Just high key carbon prints on some surfaces show slight printer artifacts.

 

[Ron-san]

Interesting..That would make sense about the reformulation. In my tests plain yellow, even set at 70 limit, was fairly weak as a blocker unless there was a little cyan to the mix, even only 15 limit on dark cyan added to the yellow made all the difference.

Colin-- Maybe there is something deep that I am missing. But I assume that all the various colored inks just look like different shades of gray to a UV sensitive emulsion. And if that is so, how can adding a small amount of cyan, which is about 10-fold weaker already, to a much stronger yellow--how can that make much of a difference to the overall UV blocking? The numbers do not seem to compute.
And of course, since I print in palladium (matte surface) instead of a less forgiving medium like carbon or silver gelatin, I probably could not see the difference anyway.
Although I and a friend have recently made some albumen prints that look very smooth to my eye.

Everyone to his own taste as the farmer said when he kissed the cow. Cheers, Ron Reeder