Pages: [1]   Go Down

Author Topic: How many gray inks does it take?  (Read 1160 times)

NAwlins_Contrarian

  • Full Member
  • ***
  • Offline Offline
  • Posts: 225
How many gray inks does it take?
« on: June 29, 2018, 10:35:43 pm »

Recently in another thread it was claimed,

Quote
The Epson P 10k/ 20k have been out over a year now and have the first true quad monochrome inkset on board within a full color inkset .... Have they shown us how their new improved ABW or rip software with those inks allow for outstanding smoother monochrome? No. But I bet they do.

Over and above the statement I think being technically incorrect insofar as the Canon Pro-1 offered an OEM quad-monochrome inkset* since its 2011 introduction, has anyone done any real testing on how many grays does it take to get to the point of no more visible improvements? I have seen some great piezography prints, but to what extent was that because there were six or seven black/gray inks, versus other technical and/or artistic mastery?

I realize that it probably depends on the image, the paper, the specifics of the inks, the dot-resolution of the printer, the quality of the ICC profile or other process used to map image tones to printer function, and probably other factors. But has anyone taken a serious look at this issue recently? If so, who, where, and with what results?

Thanks!

* The same as the P10000 and P20000: photo black, matte black, dark gray, gray, and light gray inks.
Logged

deanwork

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2400
Re: How many gray inks does it take?
« Reply #1 on: June 29, 2018, 11:48:12 pm »


I’ve specialized in monochrome inkjet Imaging since the very beginning of it, in the late 90s. This was when the first  3rd party “quad” inksets came out. They came out because there was no interest in the part of Epson, HP, or Canon to even explore bw pigment printing in a professional way.  When I bought my first 44” pigment printer, the first one by Epson, the CG 10000, they had not even one gray. I had to do my Large bw prints with the black channel only which obviously was pretty sad. I then started working with the Lyson quads followed quickly by Cone quad Piezotones about 1999 and they were incredible compared to any mono inkjet pigments that preceded them. In the late 90s Jon Cone had set up people using Iris printers with a dedicated bw inkset that used multiple grays and software he had made for the purpose. With his piezotone pigment quads for Epson printers we used the Studio Print rip from Ergosoft which allowed ink limits to be controlled and partitioned on the level of each channel. It was and remains the most subtle and sophisticated software to print bw with. However, it is very expensive and printing color with it can be a pain in the ass, though very beautiful.

When Cone came out with the K7 ( and K6 ) Piezography inksets, somewhere around 2005, this took monochrome to a level of dimensionally that had not been seen before and has not been equaled in my opinion by any other oem or third party inkset.  I’ve done excellent bw work for 12 years with various inksets by Epson out of qtr, Canon tri tones with true black and white software, and the Hp z3200 tri tone ( with an added pk for extra dmax on Matt media ) and none of them reach the dimensionality, especially in the high values, of K6 and k7. And I’m still waiting.

Yes Canon did add one more light gray in one of their desktop units but the excellent true black and white rip by bowhaus never supported it and it was worthless to me anyway because of the size limit. Canons software was useless and amateurish. I was hoping they would add that light gray to the lastest large format models, but all they did was increase speed and up the gamut a tiny amount that cut their longevity in half. So - the only oem large format printer to offer a true quad inkset with equally spaced gray dilutions is the new Epson 10k and 20k. I believe Studio Print supports these and for a color printer that also has this kind of bw capability it could be an excellent combination, and with a very stable color inkset to tone with. I wish I could afford all that. It won’t equal K7 but it should be the next best thing. QTR currently does not support the 10/20k printers unfortunately.

All this is my opinion with my experience working with bw Photography since 1974. Everybody has their own opinions about what works for them. At this point it isn’t worth arguing about. The big companies are going to do what they think will sell the fastest. Many of use multiple printers and inks for multiple purposes. I’ll leave it at that.


Recently in another thread it was claimed,

Over and above the statement I think being technically incorrect insofar as the Canon Pro-1 offered an OEM quad-monochrome inkset* since its 2011 introduction, has anyone done any real testing on how many grays does it take to get to the point of no more visible improvements? I have seen some great piezography prints, but to what extent was that because there were six or seven black/gray inks, versus other technical and/or artistic mastery?

I realize that it probably depends on the image, the paper, the specifics of the inks, the dot-resolution of the printer, the quality of the ICC profile or other process used to map image tones to printer function, and probably other factors. But has anyone taken a serious look at this issue recently? If so, who, where, and with what results?

Thanks!

* The same as the P10000 and P20000: photo black, matte black, dark gray, gray, and light gray inks.
Logged

MHMG

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 1285
Re: How many gray inks does it take?
« Reply #2 on: June 30, 2018, 12:52:13 am »

... has anyone done any real testing on how many grays does it take to get to the point of no more visible improvements? I have seen some great piezography prints, but to what extent was that because there were six or seven black/gray inks, versus other technical and/or artistic mastery?

I realize that it probably depends on the image, the paper, the specifics of the inks, the dot-resolution of the printer, the quality of the ICC profile or other process used to map image tones to printer function, and probably other factors. But has anyone taken a serious look at this issue recently? If so, who, where, and with what results?

Thanks!


Interesting question, and yes, FWIW, I have been working for some time on an objective quantification of that very aspect of B&W printing quality. In upcoming reports on the Aardenburg website (coming soon), I will be adding an initial "Grayscale neutrality" plot into the reports. It plots CIELAB a* and b* values along the L* value tonal scale for the initial grayscale print quality before any light fade testing begins. In a "perfect hue constancy" system the a* and b* values remain the same as those determined for the paper white color across the whole L* tonal scale from paper white to Dmax. However, in the real world, systems will deviate from that perfect hue constancy goal, especially when the artist actually desires to invoke some "split toning" across the tonal scale, say for example, with highlights being "warm" and shadows being "cool". That said, the a* and b* curves in any visually more appealing system will always follow very gentle and smooth curves along the total tone scale. When the grayscale neutrality scale shows increasing noise or discontinuities along the grayscale neutrality scale, it's an excellent indication the human observer will like the B&W response less and less as the jaggedness of the grayscale neutrality curve gets more and more pronounced.

The best systems, e.g. tradtional silver gelatin print processes or Cone Piezograpy K6 and K7 inkjet output, generate very very smooth grayscale neutrality curves. The visually poorest B&W output generates bumpy, jumpy, irregularly abrupt plots. So, I think this objective quantification approach to B&W tonality performance has a lot of merit, and I'm going forward with it as a published evaluation metric at Aardenburg Imaging & Archives.

BTW, the Canon Pro-1 with its quad black inkset and a well made profile also does noticeably better in the Aardenburg "grayscale neutrality" test than other 3 level gray ink sets, not quite as good as Cone K6 and K7, but once again giving me confidence that the metric does indeed have practical merit to other discerning printmakers. Also, even with a good B&W printer like the Canon Pro-1, a poor ICC profile can degrade the "B&W printed in color mode" performance to levels much worse than the printer itself is actually capable of producing...just sayin'.   Conversely, software like QTR can improve a printer's B&W performance objectively in the grayscale neutrality plots as well as subjectively in the visual judgement of discerning viewers.

kind regards,
Mark
http://www.aardenburg-imaging.com
« Last Edit: June 30, 2018, 09:11:27 am by MHMG »
Logged

deanwork

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 2400
Re: How many gray inks does it take?
« Reply #3 on: June 30, 2018, 10:00:10 am »

All this should have been on the bw group, oh well.

Mark, have you seen any obvious differences between the results measured from the new Epson ABW
workflow and the previous iterations? They also have a darker black within the inkset to work with.

It’s important that there are at least four important things to evaluate when comparing one bw system to another. All of them have great dmax these days which makes for better dynamic range.  For me ( I’m not a split toner for the most part ) they are: 1. Resolution, a function of print head design, software used, the inkset, and the paper selected. 2. Subtlety of tonal ramp, smoothness across the tonal scale, also determined by the same factors and accurate profiling 3. Print color, does the print produce the same hue in all the values ( the inks, the software, and the media used all play important roles there.) Also it should be noted that the more color ink dots that are used the more metameric failure is introduced and the greater chance of color shift under various lighting situations and that is not good.

After making that one quick test of a dead neutral very smooth bw print out of QTR on Platine with the Epson P800, it just seemed like a real improvement to me and is not just dmax improvement. But at the same time since I’ve been using other systems, I can’t say I’ve really tweaked the previous Epson inkset with either studio print or qtr. But I’ve looked at a lot of Epson bw prints in various portfolios and exhibitions and I’ve never seen any that looked that clean in regard to print color. All of this is what makes me wonder about the potential monochrome capability of the P 10k within really good linearizing software, it should be better than the P800.






Interesting question, and yes, FWIW, I have been working for some time on an objective quantification of that very aspect of B&W printing quality. In upcoming reports on the Aardenburg website (coming soon), I will be adding an initial "Grayscale neutrality" plot into the reports. It plots CIELAB a* and b* values along the L* value tonal scale for the initial grayscale print quality before any light fade testing begins. In a "perfect hue constancy" system the a* and b* values remain the same as those determined for the paper white color across the whole L* tonal scale from paper white to Dmax. However, in the real world, systems will deviate from that perfect hue constancy goal, especially when the artist actually desires to invoke some "split toning" across the tonal scale, say for example, with highlights being "warm" and shadows being "cool". That said, the a* and b* curves in any visually more appealing system will always follow very gentle and smooth curves along the total tone scale. When the grayscale neutrality scale shows increasing noise or discontinuities along the grayscale neutrality scale, it's an excellent indication the human observer will like the B&W response less and less as the jaggedness of the grayscale neutrality curve gets more and more pronounced.

The best systems, e.g. tradtional silver gelatin print processes or Cone Piezograpy K6 and K7 inkjet output, generate very very smooth grayscale neutrality curves. The visually poorest B&W output generates bumpy, jumpy, irregularly abrupt plots. So, I think this objective quantification approach to B&W tonality performance has a lot of merit, and I'm going forward with it as a published evaluation metric at Aardenburg Imaging & Archives.

BTW, the Canon Pro-1 with its quad black inkset and a well made profile also does noticeably better in the Aardenburg "grayscale neutrality" test than other 3 level gray ink sets, not quite as good as Cone K6 and K7, but once again giving me confidence that the metric does indeed have practical merit to other discerning printmakers. Also, even with a good B&W printer like the Canon Pro-1, a poor ICC profile can degrade the "B&W printed in color mode" performance to levels much worse than the printer itself is actually capable of producing...just sayin'.   Conversely, software like QTR can improve a printer's B&W performance objectively in the grayscale neutrality plots as well as subjectively in the visual judgement of discerning viewers.

kind regards,
Mark
http://www.aardenburg-imaging.com
Logged

Paul Roark

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 398
Re: How many gray inks does it take?
« Reply #4 on: June 30, 2018, 12:08:49 pm »

I've designed and used a lot of different inkset approaches over the last decade and, for a variety of reasons, I now have all but one of my printer positions containing 100% carbon pigments.  Carbon is, of course, warm by nature.  That leaves one position for a bluish toner that can pull the warm carbon to neutral (and slightly beyond for some parts of the grayscale). 

How many "gray" inks does it take?  I think the concept most economists can relate to is that there is decreasing marginal utility to more inks.  Very likely, the more the better, but in any case, with only a single color "toner" position, I'm maximizing the number of channels in the profiles while still having the ability to print with tones from neutral to carbon warm.

What I do is currently is "hard wire" the toner's ratio of color inks -- in my case cyan and blue.  For a typical setup and the toner formula I use, see http://www.paulroark.com/BW-Info/3880-Eboni-Variable-Tone.pdf, page 3.  People can vary this ratio to taste rather easily.  So, as a practical matter, the the color "profiling" is split between the mixing and the printer profile that determines how much of the toner to use.  For me, and from what I can tell from those who use my formulas, this makes a good compromise between flexibility and simplicity.

One of the factors I have considered is simply ease of profiling the inkset.  I found that most people cannot achieve a smooth Lab A & B with more than a single "color" ink they are varying.  That is, I doubt the typical B&W photographer can balance a CMY inkset.  With only a single color toner to offset the carbon warmth, I have simplified profiling about as much as is possible and still allow a range of useful print tones.

I'm agnostic on the OEM's "ABW" approach simply because I want more control and also want the least amount of color pigs in the blend as possible.  That color is going to fade a lot faster than carbon, and the color inks will fade at different rates, causing a color shift over time.  So, for B&W, my view is the less color the better.

Also, note that my selection of the color used was based in part on achieving the smallest hue angle between the colors.  In a CYM setup, the carbon is warm yellowish.  As such, C & M are used to offset this.  Some inksets have a "blue" ink.  The hue angle between that B and C is much less than between M and C.  The lower hue angle reduces the extent to which the colors are simply offsetting each other.  As such, it takes less total color ink to achieve the carbon offset.  Also, I believe, the narrowest possible hue angle reduces the extent to which the fade path will be pulled toward the stronger color as they fade faster than the carbon.

Typically, the C is much stronger than the M.  Assume the profile is neutral.  Ultimately, as the color fades, the print will end up carbon warm.  Before that point, however, if the M is fading faster than the C, the print tone gets pulled into a greenish look.  This takes a long time, but if one is selling art as "fine art," then I think for reputation (not to mention pride in one's craft/art) considering the very long run is worth the effort.

Unfortunately, I have found no inkjet pigment that is a perfect offset for carbon warm.  In watercolor pigments, the Smith Indanthrone blue is very close.  However, as an individual, I don't have the scale economies to do all that is needed to prepare a pigment for inkjet use.  I did do a quick fade test of the pigment, and it appeared competitive.  There may well, however, be good reasons the ink makers have not used it.  As a practical matter, I stick with the OEMs for the best, well-tested colors.  Carbon pigments, on the other hand, are readily available to third party sellers.

As such, finding the inkset with blue ink that looks the best in Mark's fade tests was part of my ink selection process. 

In short, in a field that is characterized by decreasing marginal utility to number of gray inks, ease of profiling was a significant factor.  Happily, I found a single toner ink blend could do it.  That left me with all the rest being able to be 100% carbon.  This allows enough inks to have two partitions of carbon and a single toner channel.  With that in the Y position, this opens the systems up to ICC control and printing.  (See QTR's Create ICC-RGB)  While I don't usually use that approach, it's there.

FWIW

Paul
www.PaulRoark &
http://www.paulroark.com/BW-Info/
Logged

Eric Myrvaagnes

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 22814
  • http://myrvaagnes.com
    • http://myrvaagnes.com
Re: How many gray inks does it take?
« Reply #5 on: June 30, 2018, 12:44:22 pm »

Mark and Paul,

Thank you both for the history and insights into B&W printing. This is valuable information and begins to take a bit of the mystery out of it, for those of us who work mostly in black-and-white.

Eric
Logged
-Eric Myrvaagnes (visit my website: http://myrvaagnes.com)

NAwlins_Contrarian

  • Full Member
  • ***
  • Offline Offline
  • Posts: 225
Re: How many gray inks does it take?
« Reply #6 on: July 02, 2018, 11:15:12 pm »

Thank you all for your thoughtful responses, you have given me a good bit to consider, and look forward to.

I realize that a bunch of factors affect B&W print quality, including at least Dmax, Dmin (paper), neutrality, linearity / smoothness, ability to smoothly make visible changes near the limits of Dmax and Dmin (is that a quality separable from linearity?), profile (or alternative mapping) quality, etc. Y'all have addressed most or maybe all of these to some degree.

Mark, your working on an objective quantification of "that very aspect of B&W printing quality" is especially interesting. I read a fair bit of the stuff you post at Aardenburg, and it sounds like another (alas, small) donation will soon be in order. Not clear though whether you mean that gray is actually untinted gray (which is what it sounds like to ignorant me), or what gray output is smoothly proportional to input.

Quote
the Canon Pro-1 with its quad black inkset and a well made profile also does noticeably better in the Aardenburg "grayscale neutrality" test than other 3 level gray ink sets, not quite as good as Cone K6 and K7

And that makes sense. I would be quite curious to see double-blind experimental tests to see whether people see this difference--serious B&W folks, general hobbyist photographers, members of the general public?

As for how many "gray" inks does it take, marginal cost / marginal utility, and more inks being better, no doubt that's true. But all else being equal (not that it often is!), what can the average person see? What can the average semi-expert see? What can you Paul see? I recently took an online color acuity test, which I realize is quite limited, but was interesting. Lots of relatively knowledgeable users are quite happy with B&W prints even from 'K2' printers like the Canon Pro-10. I have been pleasantly surprised by the quality of the B&W prints I can get with my regular little old Epson R280 (six colors but no grays, dye ink); the Canon Pro-100 ('K3', albeit dyes) is visibly better, but not as starkly so as I would have thought. (Yes, I'm sure that on some papers pigments would give me higher Dmax.)

Quote
Everybody has their own opinions about what works for them.

No doubt! I think experience with a workflow, adjusting a workflow to personal preferences, and even personal taste are big factors, and of course confirmation bias is a big issue.

Quote
All this should have been on the bw group, oh well.

That thought occurred to me, but my thoughts were, first, that this is a printing-specific question instead of a generalize B&W vision / conversion / workflow question; and second, the traffic in Printing is far greater than the traffic in Digital B&W.

Anyway, thanks all!
Logged
Pages: [1]   Go Up