Ad
Ad
Ad
Pages: « 1 2 3 [4] 5 »   Bottom of Page
Print
Author Topic: ABW or Photoshop  (Read 10271 times)
Ernst Dinkla
Sr. Member
****
Offline Offline

Posts: 2824


« Reply #60 on: June 02, 2013, 06:59:09 AM »
ReplyReply


b*:

This is the blue-yellow axis, and deviations from neutral here would influence the extent to which one perceives the print to be relatively warm or cool. To preface the observations, Ilford has informed the public in one of a previous Lula thread that IGMS does contain OBA. This would influence a spectrophotometer to return results with a bluish bias, however, the Pulse spectro I am using is UV corrected, which should mitigate this bias to an extent I have not tried to isolate. It think it likely that the OBA presence in this paper contributes to the somewhat negative bias in the b* readings. From L*9 onward, the ICC profile produces more neutral outcomes than the ABW settings, except in the range of L*68 to L*75, where they are clustered very closely, and again at L*94 to L*97. Having said all this, it is important to retain the fact that these long bars are measuring small differences, insofar as the maximum variance of about -1.50 is only slightly more than one percentage point of the scale on one side of zero.

The main conclusion I come to is that all this measuring is inconclusive. One needs to make prints and use what one thinks looks best. For example, in the Northlight B&W test page, I found a slight margin of preference for ABW Normal given how well it opens shadow detail, but the ICC profile is not far behind. Based on these tests, previous tests, and previous test prints using both an ICC and ABW-based workflow (but no QTR), I would normally opt for the ICC-based workflow because I really do like the ability to soft-proof my actual photos before sending them to print.


Mark,

I do not share your opinion "all this measuring is inconclusive". Inkjet printing did improve a lot over the two decades that we more or less have access to it and part of that was in the tools we got for measuring. Calibrating the workflows offers us the consistent printing where we can build the individual corrections on.

Where you write that the Pulse Spectro is UV corrected, I would write UV-cut and delivering extrapolated/arbitrary numbers for the blue-UV range based on the longer wavelengths measurements. That is quite a difference. All UV-cut instruments + software act like that to my knowledge.

In the end your conclusion is correct about our eyes being the last check. For example your evaluation of the Dmax not being worth it as there is no discrimination possible between Dmax already that high. With enough light we can though but it is an interesting issue. Most of the time that high level of light is not available and not desirable. Printing for framing behind glass is subject to it, there is a loss of light already by putting glass in front and at a distance off the print. For matte papers we do get Dmax numbers that do create a real problem but in the lowest museum display conditions. With the matte papers the total dynamic range is more a limit than the display light level.  

It would be interesting to see at what light levels the average eye can not distinguish detail in the shadow parts etc set against the Dmax. The light sources and sensors of our spectrometers are also challenged near 2.5 Dmax. And they work with a consistent light level which does not represent what we as observers experience. There must be some information available on this; one could create perceptual curves that combine the two variables of dynamic range and expected display light level. The Kruithof curve already shows us that display light level has an influence on what we like as color temperature at that light level.

The eyes will be the last check but also tell us that there is a limit in what we can measure and deduct. One conclusion is to skip that scientific part, the other to improve it. It is hard to say where the line has to be drawn. Creating a good image is an art but creating a good workflow is an art too in my opinion.

--
Met vriendelijke groet, Ernst

http://www.pigment-print.com/spectralplots/spectrumviz_1.htm
December 2012, 500+ inkjet media white spectral plots.
« Last Edit: June 02, 2013, 10:14:33 AM by Ernst Dinkla » Logged
Alan Goldhammer
Sr. Member
****
Offline Offline

Posts: 1616


WWW
« Reply #61 on: June 02, 2013, 08:17:32 AM »
ReplyReply

Mark and I spoke on the phone the other day as he was finishing up his analysis as I had raised some of the issues following the nice review that he posted several weeks ago.  I thought it might be instructive to do something similar with Museo Silver Rag, the paper that I print on the most.  I decided to look at all the ABW tone settings from Light to Darkest (the dark tone settings are labeled according to increasing darkness on the data table).  The reference data set is the normal Epson driver with no color management (I have an ICC profile for this paper that I made which includes a 51 step B/W patch set and I'll do some work to see how that compares in a couple of days).  I've also calculated Dmax using Bruce Lindblom's on line converter.  Patches were generated using ArgyllCMS (21 B/W step wedge) and read using the same software with an i1pro.  I have not done the type to plots that Mark did nor how Andrea re-displayed them (I'm not the greatest Excel grapher and I just updated to Office 2013 which requires another learning curve; I'll try to get those done this week).  However, there is a plot of four of the 'L' data sets showing how the curve changes with tone setting.

This paper has a better Dmax than Ilford Gold Fiber Silk according to my measurements (IGFS routinely measures 2.2).

EDIT ADDED:  forgot to thank Mark for adding the descriptors to the chart!
« Last Edit: June 02, 2013, 10:50:50 AM by Alan Goldhammer » Logged

Alan Goldhammer
Sr. Member
****
Offline Offline

Posts: 1616


WWW
« Reply #62 on: June 02, 2013, 08:18:29 AM »
ReplyReply

Couldn't figure out how to post two things at once (it's Sunday after all).  Here is the data set.

Alan
Logged

Mark D Segal
Contributor
Sr. Member
*
Offline Offline

Posts: 6880


WWW
« Reply #63 on: June 02, 2013, 09:26:36 AM »
ReplyReply

Hello Mark,
many thanks for sharing your valuable experienced opinion,
I think that it is a very precious opportunity we have here and I really appreciate it a lot.

I'm glad you found interesting the way I analyzed/showed the data in the graph, by taking as reference the real potential straight line between printed black and paper white. I have adopted this approach as my standard evaluation tool and I always found it really useful for comparisons and for complementing my real world evaluations.

I have to admit that the Ilford ICC you used, even if not perfect, is a very good one, considering that it's not custom made. I dare to say is unusually good for a canned profile. Smiley
The Ilford ICC I have used in my test, with the same settings, is obviolusly less performing and too dark.
In my experience I have often observed this kind of deviations and even more.
For this reason usually I need/prefer to build a linearization curve: it provide me a reliable/consistent/comparable reasonably good starting point for all the work.

I fully agree with your points, the linear line target is not a holy grail, and the personal supervision is always the most decisive thing.

The reason why I said these kind of analysis are not inconclusive (probably misunderstanding the real "inconclusive" word significance, sorry for that) is because I think that these are really significant, providing some objective/repeatable information regarding how a real print could probably look on paper.
I know that one can figure it out by a visual inspection, but with a simple 18 strip and a graph you know immediately if the system is spot-on or not, how much, and where are the zones that are needing more attention, if any.
Based on my little experience by doing it consistently, it will greatly improve the personal knowledge allowing for less trial/error steps in the long run.
In addition, I full agree that each image needs a different "finished" look, and for this exact reason I prefer that the overall control regarding the related corrective actions, when needed, could be possibly performed by myself in the most transparent way, and not hidden in the ICC/print process in a not so consistent and/or easily predictable way.

In any case, as you perfectly said, the most important thing is to enjoy the photographs one makes.

Thanks again.
Ciao Smiley

Andrea

Hi Andrea,

Yes, the LinFit function you used for measuring departures from linearity is very useful, and of course the kind of technique one deploys all the time in doing regressions and analysis of variance using Excel, (editorial rant: which over the years has become more and more convoluted as Microsoft complexifies the application for marketing purposes without adding any real value to it). So I much enjoyed seeing you take the trouble to bring that perspective to bear on the issues.

To be clear - I do think the measurements are worthwhile doing - for the reasons you state. My comment about them being inconclusive is that when I went back to the test prints of the Northlight B&W printer test page and compared what I saw in those photos relative to what the numbers were telling me, indeed the ABW Dark has a more linear L* outcome *by the numbers*, but the version printed with ABW Normal just looked a lot more interesting because the shadow detail was better revealed. So it's that kind of consideration - back and forth between the numbers and the photos for this particular exercise that prompted the comment I made. I'm the last one to declare a moratorium on arithmetic.

Turning to Ilford profiles - there was a time some years ago that they were truly quite poor. All that has changed. The latest crop (at least the ones for papers reviewed in our latest article) are probably hard to beat, so they have really gotten their act together on this - probably in response to feedback a number of us have been giving them over time. If you have not upgraded your canned profile for IGFS in the past while, I recommend you do so and see what results you get from it. It may or may not surpass what you can generate with the Colormunki - that I have no idea. I do most of my printing on IGFS and I use my own custom profile, generated with X-Rite Pulse Elite. I tested it relative to another profile I paid for from a very high quality external provider using the latest i12 profiling equipment and the results are very close - almost interchangeable, so I can use either. Depending on the printer model, the use of a canned profile can actually be quite satisfactory. The Epson professional printers (the Pro line) are manufactured to tight tolerances such that one unit is pretty much a clone of the other, explaining why many people are satisfied to use Epson's profiles supplied with the driver for Epson professional printers. I think for very exacting work, it still makes sense to do custom profiles, but the real need for this has become decreasingly acute over the past decade - again depending on the printer model and the acuity of the viewer's eyesight and expectations.

By the way, at level 100 in my Excel, I think I may well have switched the data between a* and b* for ABW Normal. Very sharp of you to pick that up. Thanks.
« Last Edit: June 02, 2013, 10:25:36 AM by Mark D Segal » Logged

Mark D Segal (formerly MarkDS)
Author: "Scanning Workflows with SilverFast 8....." http://www.luminous-landscape.com/reviews/film/scanning_workflows_with_silverfast_8.shtml
Mark D Segal
Contributor
Sr. Member
*
Offline Offline

Posts: 6880


WWW
« Reply #64 on: June 02, 2013, 09:55:35 AM »
ReplyReply

Mark,

I do not share your opinion "all this measuring is inconclusive". Inkjet printing did improve a lot over the two decades that we more or less have access to it and  part of that was in the tools we got for measuring. Calibrating the workflows offers us the consistent printing where we can build the individual corrections on.

Where you write that the Pulse Spectro is UV corrected, I would write UV-cut and delivering extrapolated/arbitrary numbers for the blue-UV range based on the longer wavelengths measurements. That is quite a difference. All UV-cut instruments + software act like that to my knowledge.

In the end your conclusion is correct about our eyes being the last check. For example your evaluation of the Dmax not being worth it as there is discrimination possible between Dmax already that high. With enough light we can though but it is an interesting issue. Most of the time that high level of light is not available and not desirable. Printing for framing behind glass is subject to it, there is a loss of light already by putting glass in front and at a distance off the print. For matte papers we do get Dmax numbers that do create a real problem but in the lowest museum display conditions. With the matte papers the total dynamic range is more a limit than the display light level.  

It would be interesting to see at what light levels the average eye can not distinguish detail in the shadow parts etc set against the Dmax. The light sources and sensors of our spectrometers are also challenged near 2.5 Dmax. And they work with a consistent light level which does not represent what we as observers experience. There must be some information available on this; one could create perceptual curves that combine the two variables of dynamic range and expected display light level. The Kruithof curve already shows us that display light level has an influence on what we like as color temperature at that light level.

The eyes will be the last check but also tell us that there is a limit in what we can measure and deduct. One conclusion is to skip that scientific part, the other to improve it. It is hard to say where the line has to be drawn. Creating a good image is an art but creating a good workflow is an art too in my opinion.

--
Met vriendelijke groet, Ernst

http://www.pigment-print.com/spectralplots/spectrumviz_1.htm
December 2012, 500+ inkjet media white spectral plots.

Hi Ernst, please see my comment above to Andrea on the *inconclusiveness* matter. I fully support scientific analysis and the measurements that go along with it. I was talking specifically about the relationship of measurements to visual outcomes in the context of the particular piece of work I did.

I agree with you that there are issues with UV-cut instruments. It's a one-size-fits-all and some have argued cogently that they can confuse more than clarify. But the one I have is like that, so "faute de mieux" that is what I used. Better than nothing.

I too have been intrigued by the question of the level and degree of difference in L* numbers one needs before seeing differentials of deep shadow detail under normal viewing conditions, and I have tried to relate numbers to my visual perception. I find that much depends on the layout of the gradient; placing bars side by side one perceives much smaller differences than one does looking up or down for differences over a smooth array of gradually differing tonality. One thing for sure - at least for my level of visual acuity - I cannot see a difference of blackness between about L*2 and L* 3.7 read off of prints.
« Last Edit: June 02, 2013, 10:26:35 AM by Mark D Segal » Logged

Mark D Segal (formerly MarkDS)
Author: "Scanning Workflows with SilverFast 8....." http://www.luminous-landscape.com/reviews/film/scanning_workflows_with_silverfast_8.shtml
NeroMetalliko
Jr. Member
**
Offline Offline

Posts: 64



« Reply #65 on: June 02, 2013, 03:10:56 PM »
ReplyReply

Couldn't figure out how to post two things at once (it's Sunday after all).  Here is the data set.

Alan

Hello Alan,
thanks for your measurements set,
I can tell you I can try to put your data in my type of graph if desired,
but I kindly ask you a big favor:
can you put the data in a txt or csv or xls file instead of the jpeg one?
Because there are a lot of numbers, and from the jpg I have to write it manually one to one...

I was so curious that I done it manually for the Mark data set,
but it was time consuming (and error prone),
so if you could share a more numeric readable file I will do it more easily.

I assume your data are referenced to Gray gamma 2.2,
please confirm it because it is a required information for having a correct graph.

Many thanks for the attention.

Ciao Smiley
Andrea
« Last Edit: June 02, 2013, 03:13:12 PM by NeroMetalliko » Logged
Mark D Segal
Contributor
Sr. Member
*
Offline Offline

Posts: 6880


WWW
« Reply #66 on: June 02, 2013, 03:34:49 PM »
ReplyReply

Hi Andrea,

So sorry you went to all that trouble with my data. I could easily have sent you the Excel. If there is a next time, please just ask. :-)

Regards,

Mark
Logged

Mark D Segal (formerly MarkDS)
Author: "Scanning Workflows with SilverFast 8....." http://www.luminous-landscape.com/reviews/film/scanning_workflows_with_silverfast_8.shtml
Alan Goldhammer
Sr. Member
****
Offline Offline

Posts: 1616


WWW
« Reply #67 on: June 02, 2013, 04:03:19 PM »
ReplyReply

Hello Alan,
thanks for your measurements set,
I can tell you I can try to put your data in my type of graph if desired,
but I kindly ask you a big favor:
can you put the data in a txt or csv or xls file instead of the jpeg one?
Because there are a lot of numbers, and from the jpg I have to write it manually one to one...

I was so curious that I done it manually for the Mark data set,
but it was time consuming (and error prone),
so if you could share a more numeric readable file I will do it more easily.

I assume your data are referenced to Gray gamma 2.2,
please confirm it because it is a required information for having a correct graph.

Many thanks for the attention.

Ciao Smiley
Andrea
Check your LuLa message box and get back to me with an email address.  Yes, the data should be in gray gamma 2.2.  Targets were generated in ArgyllCMS and printed out with Adobe Color Print Utility with no color management at all.

Alan
Logged

NeroMetalliko
Jr. Member
**
Offline Offline

Posts: 64



« Reply #68 on: June 02, 2013, 04:17:36 PM »
ReplyReply

Hi Andrea,

Yes, the LinFit function you used for measuring departures from linearity is very useful, and of course the kind of technique one deploys all the time in doing regressions and analysis of variance using Excel, (editorial rant: which over the years has become more and more convoluted as Microsoft complexifies the application for marketing purposes without adding any real value to it). So I much enjoyed seeing you take the trouble to bring that perspective to bear on the issues.

Glad you liked it (btw I use Libreoffice Calc and not Excel). It is not a "real" LinFit in math terms, I labeled it Linfit in the graph but it is a simple line from measured black to measured white points.

Quote
To be clear - I do think the measurements are worthwhile doing - for the reasons you state. My comment about them being inconclusive is that when I went back to the test prints of the Northlight B&W printer test page and compared what I saw in those photos relative to what the numbers were telling me, indeed the ABW Dark has a more linear L* outcome *by the numbers*, but the version printed with ABW Normal just looked a lot more interesting because the shadow detail was better revealed. So it's that kind of consideration - back and forth between the numbers and the photos for this particular exercise that prompted the comment I made. I'm the last one to declare a moratorium on arithmetic.

I agree with you, but nonetheless I like to add this additional tought:
if the Northlight B&W printer test image printed in ABW-normal was (and probably it is) more appealing, usually I ask myself this question:
- it is more appealing because of technical reasons (it is more similar to what I see on screen when compared to the others two prints)?
- or it is more appealing because of artistic reasons (I have done it in that way but after comparing the prints I have discovered I like it less dark in the midtones, even if this is not exactly what I see on screen)?

If the reason is mainly technical, I could accept to settle to ABW-normal as default setting, because maybe this means that there is a slight perceptive deviation between the on-screen image and the ideal L* linear print process. So, once confirmed, ABW-normal could become my overall good default setting for all the images I will print on that paper/printer/ink combination even if this profile is not the more linear one (as I know from my measurements).

But if the reason is something related to my personal artistic taste for that particular image, the most correct way I should address it is by editing the source image before printing, and not getting it in the print process as non-linearity side effect. In this case I tend to think I will stick to ABW-dark as general default setting for that that paper/printer/ink combination and edit that particular image to my tastes before printing.

Quote
Turning to Ilford profiles - there was a time some years ago that they were truly quite poor. All that has changed. The latest crop (at least the ones for papers reviewed in our latest article) are probably hard to beat, so they have really gotten their act together on this - probably in response to feedback a number of us have been giving them over time. If you have not upgraded your canned profile for IGFS in the past while, I recommend you do so and see what results you get from it. It may or may not surpass what you can generate with the Colormunki - that I have no idea. I do most of my printing on IGFS and I use my own custom profile, generated with X-Rite Pulse Elite. I tested it relative to another profile I paid for from a very high quality external provider using the latest i12 profiling equipment and the results are very close - almost interchangeable, so I can use either. Depending on the printer model, the use of a canned profile can actually be quite satisfactory. The Epson professional printers (the Pro line) are manufactured to tight tolerances such that one unit is pretty much a clone of the other, explaining why many people are satisfied to use Epson's profiles supplied with the driver for Epson professional printers. I think for very exacting work, it still makes sense to do custom profiles, but the real need for this has become decreasingly acute over the past decade - again depending on the printer model and the acuity of the viewer's eyesight and expectations.

Unlucky I tend to think that is difficult to make general assumptions on canned profiles and/or ABW settings.
My Ilford profiles are recent (I will re-check again in any case), but in my experience there is not valid proof that, if a particular canned ICC profile works good for a given printer/paper combination it will be perform equally good for another one. As you cans see My ilford ICC is far less satisfactory that yours.

If you look to Ernst thread "Gold standard",
http://www.luminous-landscape.com/forum/index.php?topic=78352.0
my last post is clearly showing that ABW-normal was my chosen best overall starting point (ABW-dark being too dark) and still it was not so good for the Gold Cotton Smooth paper as it is ABW-dark on Gold Fibre Silk.

If you desire, I can show you a graph of the Ilford ICC for that Gold cotton Smooth paper that is even more dark than the one of the Gold Fiber Silk I already showed.

But, as showed, ABW-dark was very similar for your 4900+MonoSilk and my R3000+FibreSilk.

For these reasons, I never trust in blind mode, and I always want to see what happens in my setup. Then, if the canned ICC (or a given ABW settings) it shows to be already good, this is a appreciated bonus for me.

As visible reading my few posts in this forum, at the moment I'm more focused on B&W print with related linearization/neutralization, and this is a simplified reality.
I know that for color ICC the things are way more complex and my ColorMunki could be really challenged (or become a limiting factor too) for really good results. I will see.

Quote
By the way, at level 100 in my Excel, I think I may well have switched the data between a* and b* for ABW Normal. Very sharp of you to pick that up. Thanks.

Thanks, I was simply lucky enough to notice it in the table and graphs Smiley

Ciao Smiley
Andrea
Logged
NeroMetalliko
Jr. Member
**
Offline Offline

Posts: 64



« Reply #69 on: June 02, 2013, 04:25:09 PM »
ReplyReply

Hi Andrea,

So sorry you went to all that trouble with my data. I could easily have sent you the Excel. If there is a next time, please just ask. :-)

Regards,

Mark

Hello Mark,

no problem,
it was my personal curiosity to look inside your data and so I decided to do it.
It was not a big trouble, only a little bit annoying, so in the future if readable data are available, it could be easy for me to do the task.

Thanks for your willingness.

Ciao Smiley
Andrea
Logged
Mark D Segal
Contributor
Sr. Member
*
Offline Offline

Posts: 6880


WWW
« Reply #70 on: June 02, 2013, 05:12:41 PM »
ReplyReply

Glad you liked it (btw I use Libreoffice Calc and not Excel). It is not a "real" LinFit in math terms, I labeled it Linfit in the graph but it is a simple line from measured black to measured white points.

I agree with you, but nonetheless I like to add this additional tought:
if the Northlight B&W printer test image printed in ABW-normal was (and probably it is) more appealing, usually I ask myself this question:
- it is more appealing because of technical reasons (it is more similar to what I see on screen when compared to the others two prints)?
- or it is more appealing because of artistic reasons (I have done it in that way but after comparing the prints I have discovered I like it less dark in the midtones, even if this is not exactly what I see on screen)?

If the reason is mainly technical, I could accept to settle to ABW-normal as default setting, because maybe this means that there is a slight perceptive deviation between the on-screen image and the ideal L* linear print process. So, once confirmed, ABW-normal could become my overall good default setting for all the images I will print on that paper/printer/ink combination even if this profile is not the more linear one (as I know from my measurements).

But if the reason is something related to my personal artistic taste for that particular image, the most correct way I should address it is by editing the source image before printing, and not getting it in the print process as non-linearity side effect. In this case I tend to think I will stick to ABW-dark as general default setting for that that paper/printer/ink combination and edit that particular image to my tastes before printing.

Unlucky I tend to think that is difficult to make general assumptions on canned profiles and/or ABW settings.
My Ilford profiles are recent (I will re-check again in any case), but in my experience there is not valid proof that, if a particular canned ICC profile works good for a given printer/paper combination it will be perform equally good for another one. As you cans see My ilford ICC is far less satisfactory that yours.

If you look to Ernst thread "Gold standard",
http://www.luminous-landscape.com/forum/index.php?topic=78352.0
my last post is clearly showing that ABW-normal was my chosen best overall starting point (ABW-dark being too dark) and still it was not so good for the Gold Cotton Smooth paper as it is ABW-dark on Gold Fibre Silk.

If you desire, I can show you a graph of the Ilford ICC for that Gold cotton Smooth paper that is even more dark than the one of the Gold Fiber Silk I already showed.

But, as showed, ABW-dark was very similar for your 4900+MonoSilk and my R3000+FibreSilk.

For these reasons, I never trust in blind mode, and I always want to see what happens in my setup. Then, if the canned ICC (or a given ABW settings) it shows to be already good, this is a appreciated bonus for me.

As visible reading my few posts in this forum, at the moment I'm more focused on B&W print with related linearization/neutralization, and this is a simplified reality.
I know that for color ICC the things are way more complex and my ColorMunki could be really challenged (or become a limiting factor too) for really good results. I will see.

Thanks, I was simply lucky enough to notice it in the table and graphs Smiley

Ciao Smiley
Andrea


Andrea,

I don't usually do ABW. It comes up only when I'm writing an article where I think it appropriate to include for sake of thoroughness. I use a standard ICC workflow, my display is a high quality product (NEC PA271W) calibrated and profiled with what I think (based on profile analysis) for this display is now the preferable monitor calibration/profiling software on the market (BasicColor 4) and a custom-calibrated colorimeter provided for the display (this I could most likely improve by spending 800 dollars on a Discus, but I can't see the justification just now). The predictability of my outcomes between what I see on the display and what comes out of the printer using Ilford Gold Fibre Silk in an Epson 4900 is about 95% in terms of retained prints versus total prints. I think that's not too bad. With ABW I could not achieve this because it can't be soft-proofed and I simply don't have time to go through hoops simulating this. And I'm not all that hung-up on linearity because precise linearity is unimportant for the photos I make. These days I'm hardly rendering files. They come from the camera, into Lightroom and out from the printer in one seamless workflow, unless they need Photoshop for adjustments LR can't do, and then I still send them back to LR for printing. This is as true for B&W as it is for colour. For B&W I think LR's conversion option is fine and offers a lot of tonal editing flexibility with predictable outcomes. Using NIK Silver Efex expands into more exotic possibilities with very localized fine-tuning that doesn't need complex masking, so it's nice. I would add that most people probably hope they could use ABW to advantage without needing to go through all the hoops you have gone through - I saw from your DPReview forum article - months of work to get linear results. ABW is a respectable option - it has somewhat higher DMax (a doubtless technical quality but of dubious value as far as my eyes can see) and *can be* more linear than an ICC profile, not clear how much more neutral - depends on the ICC profile compared and the paper; so all in all, yes, it's an option, but perhaps less compelling than some make it out to be. I'm sure not everyone will agree with me about a lot of what I've just written here, but "vive la difference". :-)

By the way, with your interest in B&W printing, if you don't have it yet, you may find Vincent Versace's book "From Oz to Kansas" of interest. It's the most comprehensive book on Black and White conversion I've ever seen; however, I must caution - in all 250+ pages, there is not a mention of Epson ABW mode. Maybe that's why the sub-title of the book is "Almost Every Black and White Conversion Technique Known to Man".

All that said, you appear to be doing pioneering work with this tool, and if you have succeeded in making ABW perform better than what it does in its native state, congratulations and respect are in order, and for those who want to use ABW, they could perhaps benefit from what you have done.
Logged

Mark D Segal (formerly MarkDS)
Author: "Scanning Workflows with SilverFast 8....." http://www.luminous-landscape.com/reviews/film/scanning_workflows_with_silverfast_8.shtml
NeroMetalliko
Jr. Member
**
Offline Offline

Posts: 64



« Reply #71 on: June 02, 2013, 05:55:13 PM »
ReplyReply

Andrea,

I don't usually do ABW. It comes up only when I'm writing an article where I think it appropriate to include for sake of thoroughness. I use a standard ICC workflow, my display is a high quality product (NEC PA271W) calibrated and profiled with what I think (based on profile analysis) for this display is now the preferable monitor calibration/profiling software on the market (BasicColor 4) and a custom-calibrated colorimeter provided for the display (this I could most likely improve by spending 800 dollars on a Discus, but I can't see the justification just now). The predictability of my outcomes between what I see on the display and what comes out of the printer using Ilford Gold Fibre Silk in an Epson 4900 is about 95% in terms of retained prints versus total prints. I think that's not too bad. With ABW I could not achieve this because it can't be soft-proofed and I simply don't have time to go through hoops simulating this. And I'm not all that hung-up on linearity because precise linearity is unimportant for the photos I make. These days I'm hardly rendering files. They come from the camera, into Lightroom and out from the printer in one seamless workflow, unless they need Photoshop for adjustments LR can't do, and then I still send them back to LR for printing. This is as true for B&W as it is for colour. For B&W I think LR's conversion option is fine and offers a lot of tonal editing flexibility with predictable outcomes. Using NIK Silver Efex expands into more exotic possibilities with very localized fine-tuning that doesn't need complex masking, so it's nice. I would add that most people probably hope they could use ABW to advantage without needing to go through all the hoops you have gone through - I saw from your DPReview forum article - months of work to get linear results. ABW is a respectable option - it has somewhat higher DMax (a doubtless technical quality but of dubious value as far as my eyes can see) and *can be* more linear than an ICC profile, not clear how much more neutral - depends on the ICC profile compared and the paper; so all in all, yes, it's an option, but perhaps less compelling than some make it out to be. I'm sure not everyone will agree with me about a lot of what I've just written here, but "vive la difference". :-)

Hello Mark,

many many thanks for sharing your opinions,
I'm a newcomer here, and I really appreciate it.

I don't want to steal further time to you, I perfectly understand that you have a very effective/proven workflow and this is what matters.

I have invested some month of time in my tool development (which is still under development) but I like to highlight that currently the time needed for single a ABW or ICC linearization is minimal.
The big time limiting factor is the wait for the ink to settle in the printed strips, but even for a 3 step process you are done in 3 days of wait (1 day for each pass due to ink settlement) and only some tenths of minutes of active work for the measurements, data manipulation, DeviceLink ICC creation.

Quote
By the way, with your interest in B&W printing, if you don't have it yet, you may find Vincent Versace's book "From Oz to Kansas" of interest. It's the most comprehensive book on Black and White conversion I've ever seen; however, I must caution - in all 250+ pages, there is not a mention of Epson ABW mode. Maybe that's why the sub-title of the book is "Almost Every Black and White Conversion Technique Known to Man".

Thanks for the book hint, I will look at it for sure.

Quote
All that said, you appear to be doing pioneering work with this tool, and if you have succeeded in making ABW perform better than what it does in its native state, congratulations and respect are in order, and for those who want to use ABW, they could perhaps benefit from what you have done.

Many thanks for your appreciation.
At the moment, in this field I'm only an humble enthusiast hobbyist, luckily I have some engineering skills, and I have simply tried to use it for this purpose.
Please, note that the tool is usable not only for ABW but for ICC linearization too (with optional gray tone neutralizing or grading in this case, if desired).

That said, I don't know if what I'm trying to do is something really uncovered or not, I tend to doubt it.
Maybe I could be simply one who has published/asked something on this regard, but I think that the big hardware/software makers in the color-management world have smart ideas and smart developers, so it could easily be that all (or part) of this is something already there and only hidden inside their products to protect their know-how.
Frankly, I don't know if I will ever further benefit from what I am developing more than what I'm already doing, i.e. by using this tool for my joy and my photo prints,
but time will tell. Smiley

Obviously any further opinion/suggestion will be always welcome and appreciated.
Thanks again.

Ciao Smiley
Andrea

« Last Edit: June 02, 2013, 06:01:10 PM by NeroMetalliko » Logged
NeroMetalliko
Jr. Member
**
Offline Offline

Posts: 64



« Reply #72 on: June 03, 2013, 05:42:09 PM »
ReplyReply

Mark and I spoke on the phone the other day as he was finishing up his analysis as I had raised some of the issues following the nice review that he posted several weeks ago.  I thought it might be instructive to do something similar with Museo Silver Rag, the paper that I print on the most.  I decided to look at all the ABW tone settings from Light to Darkest (the dark tone settings are labeled according to increasing darkness on the data table).  The reference data set is the normal Epson driver with no color management (I have an ICC profile for this paper that I made which includes a 51 step B/W patch set and I'll do some work to see how that compares in a couple of days).  I've also calculated Dmax using Bruce Lindblom's on line converter.  Patches were generated using ArgyllCMS (21 B/W step wedge) and read using the same software with an i1pro.  I have not done the type to plots that Mark did nor how Andrea re-displayed them (I'm not the greatest Excel grapher and I just updated to Office 2013 which requires another learning curve; I'll try to get those done this week).  However, there is a plot of four of the 'L' data sets showing how the curve changes with tone setting.

Alan, thanks for having sent me the readable data.

I have tried to put it on my type of graphs.

Because your reference column is in K values and not in L* ones, initially I didn't know how to manage it.
There is a long time debate on the net regarding what kind of gamma Epson driver expect for ABW.
To convert from K values to L* values I need to know what kind of gamma to apply.

I have made a first group of graphs assuming a gamma 2.2 conversion.
Looking into it, I had the feeling something was not right.
In addition I have noticed that in "Epson standard" color mode the Epson driver is defaulted to sRGB (at least in my R3000 drivers), so I thought it could make sense for Epson to use the same default for expected encoded images printed via ABW mode.
For these reasons I decided to plot again all the data using a sRGB gamma curve for L* x axis reference calculations, instead of the gamma 2.2 previously used.

I have attached both the graph group for public evaluation, because I think it could be interesting to know all the opinion here.

As visible, the "sRGB" group is clearly more linear than the "g22" group. ABW-D1 (dark setting) is always the best of the bunch.

Usually I don't get the irregular deep-shadows zone behavior visible here in the gamma 2.2 test,
for this reason it could be useful to remark that these results are valid for this kind of printer/paper/settings combination, and, as already stated, it could not be convenient to draw more general conclusions from these data only.
In addition, looking to the b* shapes,  the gray tone is not perfectly neutral (it's a little bit "cool"), and I have seen better results in ABW in other tests.

Quote
This paper has a better Dmax than Ilford Gold Fiber Silk according to my measurements (IGFS routinely measures 2.2).

This is strange, I get Dmax sometimes exceeding 2.4 with Gold Fibre Silk and I think I'm not the only one.

However,
I hope this was useful.
Any further opinion/comment is always welcome.

Ciao Smiley
Andrea
« Last Edit: June 04, 2013, 03:25:09 PM by NeroMetalliko » Logged
NeroMetalliko
Jr. Member
**
Offline Offline

Posts: 64



« Reply #73 on: June 04, 2013, 05:21:59 AM »
ReplyReply

(I have an ICC profile for this paper that I made which includes a 51 step B/W patch set and I'll do some work to see how that compares in a couple of days).

Alan,
if you have the data related to the above mentioned custom ICC built including 51 B/W steps, please, let me know, that I will try to compare it to the "Eposn std" and the "ABW" we have already seen.
Thanks.

Ciao Smiley
Andrea
Logged
MHMG
Sr. Member
****
Offline Offline

Posts: 600


« Reply #74 on: June 04, 2013, 08:47:57 AM »
ReplyReply


I have attached both the graph group for public evaluation, because I think it could be interesting to know all the opinion here.


Far from being of academic value only, graphs like these provide considerable insight to the photographer/printmaker if one understands how to interpret them. First, let me say to forum members here who are not familiar with this type of technical evaluation of photographic and printing systems that Andrea's plotting procedure is indeed the correct way to evaluate this data. He has appropriately plotted L*input on the x axis and L* output on the y-axis combined with a* and b* data plotted at + and - minus Y axis values along a horizontal mid axis and also corresponding to the L* input values for the patches from which they were measured.  Such a graph as Andrea has plotted is the modern digital era's colorimetric approach to what classically was called an H&D curve in the film era of photography and follows in a tradition of photographic sensitometry dating back to well over a century in applied usage.  It's difficult, however to teach folks how to interpret them without the benefit of being in a classroom where a pointer can be used to highlight aspects of each graph easily and where questions and answers can can flow more quickly. That said, I will attempt to point out a couple of features to a few of the graphs, and that perhaps may help.

1). A perfect system for L* tone reproduction would follow a 45 degree straight line stretching from L* 0 to L*= 100. Imagine (or draw) a dashed line like this on all of these graphs. The slope or gamma if you will of this line = 1 and that's the ideal aimpoint reproduction. However, no printer/ink/paper achieves an L* max= 100 and no printer/ink/paper reaches a black ink on paper value of L* = 0, which means  Andrea's choice of straight line linearization as the goal is one logical approach, but visual contrast over all tones is nevertheless forced lower in order to achieve. Glossy papers with high whitepoint and rich blacks get closest which is why producing "prints that match my monitor" is easier with said papers compared to mat fine art papers (or worse yet plain uncoated papers). Note that compared to the ideal gamma 1 objective, all of the graphs from Epson Standard to ABW light and dark all follow a rendering theme dancing in and around Andrea's full linearization approach. This result is logical because all are useful as starting points for various print reproduction objectives, with some better for some images and viewing conditions than others. It depends on the application objective and what parts of the global curve one wants to tackle with additional image edits (more curves and selective dodging and burning to emphasize certain areas of a particular image. I will try to highlight a few examples that hopefully may help folks come to their own useful interpretations.

2). The curve for ABW Light: This aimpoint curve provides the lightest print and with the most open shadow detail by virtue of of L*output running higher than L* input values over all of the curve except anchored at L* max and L* min to what the ink/media can provide. Such a curve is a good starting point for image reproduction if one Knows the print will be displayed under relatively dim lighting conditions, Viewers will judge print quality best under these conditions, but the downside is when the print is displayed under better lighting conditions it will look more washed out (light and lacking image contrast in highlights and midtones albeit with good shadow detail).

2). ABW Normal, ABW Dark, ABW darker: The curve progression between these rendering choices goes in the predictable direction it needs to in order to produce gently increasing darkness in the overall tonal rendering. ABW Normal nicely splits the difference between ABW light and ABW dark but still errs slightly in the direction of an overall lighter print. Epson as well as other manufacturers typically regard this type of tone reproduction as "normal" because many studies show that the majority of prints out there in the real world are still displayed under less than optimal lighting, hence will be helped by a "normal" tonality erring on the lighter printing side of the straight linearized tone reproduction curve.

3). Example 3). The Epson Standard result: This is a bit surprising but may be the result of the media settings used. Typically, for Epson consumer printers printing sRGB color images, the Epson standard controls (Epson's hardwired controls in the driver for consumers who don't know about ICC profile) produces a  curve that is shaped like this one, but again typically errs in the mid tones through the 3 quarter tones slightly lighter and thus above a fully linearized curve like Andrea chooses to use. This test result for Epson Standard mode boosts mid tone contrast at the expense of shadow reproduction as I would have expected, but I would also have expected the mid tone region boosted also lighter than what we see here. Note again that shadows are compressed more hence lower contrast, but midtones are getting a boost in contrast. One will have to dodge shadows to open them up more if this curve is used as a starting point, but less work would have to be done to mid tones and highlights to achieve good visual contrast overall. What's more interesting in this plot is the b* dip which occurs at about 45 L. See item 4 below.

4). To have perfect monochromatic neutral under a specified illuminant condition (e.g. D50) across all tones in in a B&W print requires a flat curve parallel to  the horizontal zero line for both a* and b*. Many inkjet papers on the market today vary from bright white to warm white color, and this range of media color typically plots between b* = -10 to +5. Also, many PK and MK inks don't always achieve a perfect b*=0 on all media. Hence, the a* and b* curves must also bend to accommodate printing system realities, but gentle gradual bending is visually better than lots of bumps and dips. That said, one shouldn't get alarmed at + or - deviations less than or equal to approximately 1.0 because much of that is simple instrument and patch measuring repeatability. However, when you see rapid dips or bumps at 2.0 a* or b* difference over smalL L* difference on these plots as seen in the Epson Standard curve plot at the 45 L* tone, such dips start to become noticeable as unwanted "color ripple".  Some people are more sensitive to these color variations in B&W prints than others, but ABW mode in these graphs shows it's handling the hue gradation across the tone range better than the Epson standard mode in this test. Note also that even ABW mode shows a relatively rapid change in b* value as shadow tones near L* minimum. This is where the Epson driver is switching over to PK or MK black ink and feathering out the other gray inks and colors. This would be a noticeable issue too, but the saving grace is that it's occuring at the lowest L* values in the print. The often quoted rule that humans can see delta E changes as small as 1.0 is a good general rule of thumb but it's not perfect. As tones get darker, our visual color discrmination gets poorer, so a 3 0r 4 unit discrepancy in the very darkest part of the print will generally not be likely seen as that dip 2.0 b* dip occuring at the 45L location in the Epson Standard graph. Lastly, that the hue of Silver Rag gradually moves from slight positive 2.0 b* value at media white point (i.e., it's a slightly warm colored paper) to a -2.0 value over the tone range and then back up again as dmax (l*min) is approached is providing an ever so subtle split tone effect for this ink on this paper with this Epson driver, but overall would be judged as a very neutral B&W print with no unwanted color ripple (except for possibly in the Epson Standard mode at the b* dip occurring at L45).

My post has gotten very long which is the nature of this type of analsis, but I hope it helps more folks reading this thread to get a better sense of how to interpret these plots. They are indeed very informative of tone reproduction behavior of the different rendering modes for the chosen printer/ink/media.

regards,
Mark
http://www.aardenburg-imaging.com
« Last Edit: June 04, 2013, 09:00:34 AM by MHMG » Logged
Alan Goldhammer
Sr. Member
****
Offline Offline

Posts: 1616


WWW
« Reply #75 on: June 04, 2013, 02:24:32 PM »
ReplyReply

3). Example 3). The Epson Standard result: This is a bit surprising but may be the result of the media settings used. Typically, for Epson consumer printers printing sRGB color images, the Epson standard controls (Epson's hardwired controls in the driver for consumers who don't know about ICC profile) produces a  curve that is shaped like this one, but again typically errs in the mid tones through the 3 quarter tones slightly lighter and thus above a fully linearized curve like Andrea chooses to use. This test result for Epson Standard mode boosts mid tone contrast at the expense of shadow reproduction as I would have expected, but I would also have expected the mid tone region boosted also lighter than what we see here. Note again that shadows are compressed more hence lower contrast, but midtones are getting a boost in contrast. One will have to dodge shadows to open them up more if this curve is used as a starting point, but less work would have to be done to mid tones and highlights to achieve good visual contrast overall. What's more interesting in this plot is the b* dip which occurs at about 45 L. See item 4 below.

Mark, the data that Andrea worked up was from me.  The Epson Standard result was simply printing the monochrome patch set from ACPU with no color management in the Epson driver.  I will be printing out the same patch set using an ICC profile that I made that includes a full B/W patch set.  I would assume this will give a different result than the one displayed.

Alan
Logged

MHMG
Sr. Member
****
Offline Offline

Posts: 600


« Reply #76 on: June 04, 2013, 03:41:03 PM »
ReplyReply

Mark, the data that Andrea worked up was from me.  The Epson Standard result was simply printing the monochrome patch set from ACPU with no color management in the Epson driver.  I will be printing out the same patch set using an ICC profile that I made that includes a full B/W patch set.  I would assume this will give a different result than the one displayed.

Alan

Ok, that explains the dark tone scale rendering.  I guess I wouldn't have called it Epson std mode. It's NCA, i.e., "no color management" mode with the Epson driver and represents the full ink ramp of the chosen media setting, thus no changes in RGB data values being sent to the printer. I tend to reserve the term Epson "Standard mode" for the way Epson uses it, which is when sending a straight sRGB tagged image (in this case perhaps preconverted to RGB neutral triplets) to the Epson driver set to "standard" not "vivid" or "Adobe RGB" mode using the Epson color control menu. This menu is Epson's built in "easy" color management pathway for amateur photographers shooting sRGB jpegs and not wanting to learn how to set up a full color managed workflow using ICC printer profiles.

cheers,
Mark
Logged
NeroMetalliko
Jr. Member
**
Offline Offline

Posts: 64



« Reply #77 on: June 04, 2013, 04:28:39 PM »
ReplyReply

3). Example 3). The Epson Standard result: This is a bit surprising but may be the result of the media settings used. Typically, for Epson consumer printers printing sRGB color images, the Epson standard controls (Epson's hardwired controls in the driver for consumers who don't know about ICC profile) produces a  curve that is shaped like this one, but again typically errs in the mid tones through the 3 quarter tones slightly lighter and thus above a fully linearized curve like Andrea chooses to use. This test result for Epson Standard mode boosts mid tone contrast at the expense of shadow reproduction as I would have expected, but I would also have expected the mid tone region boosted also lighter than what we see here. Note again that shadows are compressed more hence lower contrast, but midtones are getting a boost in contrast. One will have to dodge shadows to open them up more if this curve is used as a starting point, but less work would have to be done to mid tones and highlights to achieve good visual contrast overall. What's more interesting in this plot is the b* dip which occurs at about 45 L. See item 4 below.

4). To have perfect monochromatic neutral under a specified illuminant condition (e.g. D50) across all tones in in a B&W print requires a flat curve parallel to  the horizontal zero line for both a* and b*. Many inkjet papers on the market today vary from bright white to warm white color, and this range of media color typically plots between b* = -10 to +5. Also, many PK and MK inks don't always achieve a perfect b*=0 on all media. Hence, the a* and b* curves must also bend to accommodate printing system realities, but gentle gradual bending is visually better than lots of bumps and dips. That said, one shouldn't get alarmed at + or - deviations less than or equal to approximately 1.0 because much of that is simple instrument and patch measuring repeatability. However, when you see rapid dips or bumps at 2.0 a* or b* difference over smalL L* difference on these plots as seen in the Epson Standard curve plot at the 45 L* tone, such dips start to become noticeable as unwanted "color ripple".  Some people are more sensitive to these color variations in B&W prints than others, but ABW mode in these graphs shows it's handling the hue gradation across the tone range better than the Epson standard mode in this test. Note also that even ABW mode shows a relatively rapid change in b* value as shadow tones near L* minimum. This is where the Epson driver is switching over to PK or MK black ink and feathering out the other gray inks and colors. This would be a noticeable issue too, but the saving grace is that it's occuring at the lowest L* values in the print. The often quoted rule that humans can see delta E changes as small as 1.0 is a good general rule of thumb but it's not perfect. As tones get darker, our visual color discrmination gets poorer, so a 3 0r 4 unit discrepancy in the very darkest part of the print will generally not be likely seen as that dip 2.0 b* dip occuring at the 45L location in the Epson Standard graph. Lastly, that the hue of Silver Rag gradually moves from slight positive 2.0 b* value at media white point (i.e., it's a slightly warm colored paper) to a -2.0 value over the tone range and then back up again as dmax (l*min) is approached is providing an ever so subtle split tone effect for this ink on this paper with this Epson driver, but overall would be judged as a very neutral B&W print with no unwanted color ripple (except for possibly in the Epson Standard mode at the b* dip occurring at L45).

Hello Mark,
many thanks for your valuable contribution and the articulated comments, I think it is very useful and appreciated.

Adding another straight line to the plot for the 0-100 (45 degrees) ideal L* function is clearly possible, I have done it in the past, but then I decided to remove it in order to keep the graph more clean and easily readable.

The most puzzling aspect for me is the strange L* "bump" in the deep shadows visible in the gamma 2.2 group.
It is something that usually I don't get, and consider that the big part of my personal past test are done in Adobe RGB (gamma 2.2), so this is very strange.
From what I can see in this case it could seems very reasonable to argue that the ABW mode is best served from a sRGB gamma than a 2.2 gamma, even if I know, as previously stated, that this is not always the case for sure, and at least it was not the case in the vast majority of my personal test.

I don't know if the fact that, in this test, the Alan printed strip was not colorspace assigned (it was kept unmanaged) before printing it, could have something to do with this phenomenon.

Another curious observation is the clear difference in the a* shape between "Epson-std" and all the "ABW" modes. In ABW modes the a* is very neutral across L* range, in Epson std is a little bit shifted toward green (negative values).

The last thought is related to the b* negative "bump" you rightly pointed out, which is more clearly defined in "Epson std" but it's still present in the ABW modes, even if somewhat masked from the more smooth (and extended in the shadows) shape of the b* curve.
Now, if you take in account that the "Epson std" L* values in the shadows zone are by far the darkest of all the 6 set (including ABW-darkest), and if you notice that moving across the ABW modes in a dark increasing order (from LT to D3) the b* shape in the shadows zone tend to "shift" to the right and to reveal/define more the negative "bump", in my opinion the b* shape of "Epson std" is not so inconsistent from the ABW ones, if you extrapolate the observed b* "right shift" from ABW-D3 to Epson std in relation to the increased L* darkness of the shadows zone.
I don't know if I was able to well explain what I intend to explain in this latest thought or not, so please, if something is not clear, let me know.

However, given the scale difference between L* (0 100) and a*, b* (-5 +5) adopted, the L* deviations from linearity visible in the graphs are more noticeable in real prints than the a*, b* (tone neutrality) ones, in my opinion.

Thanks again.

Ciao Smiley
Andrea
« Last Edit: June 04, 2013, 04:36:14 PM by NeroMetalliko » Logged
royvharrington
Newbie
*
Offline Offline

Posts: 7


« Reply #78 on: June 04, 2013, 06:16:40 PM »
ReplyReply

The graphs are interesting info about several different aspects.  So I have a few comments.

First, I've used similar graphs a lot over the years in making QTR curves.  It's the best info about
how well you are doing with the linearity for both tones and neutrality.

1) about the initial data and methods.  I have to agree with I think Mark A. the Epson STD being done without
an ICC profile shows just the response of the the driver and really isn't a good comparison of what you should
get printing normally.  The ABWs however are supposed to be tuned from the get-go so they are indicative
of what you will actually get.  (I wonder about the ABW-D3 data in both cases because it seems to go to L=100
rather than paper white of about L=96, but not too big a deal).  I am surprised at how cool the darks are (b=-2.5)
in all the ABW cases.  I don't think you've mentioned the ABW paper that was selected in printing but it may
have different characteristics than the Silver Rag paper you actually used.

2) the difference shapes of the curves in the dark shadows illustrate the difference between the different embedded
profiles.  AdobeRGB and Gamma 2.2 are true gamma curves -- pure exponentials.  sRGB and Mac's special
Generic Gamma 2.2 are pretty close for most of the range but you can see they are pretty different in the very
darkest range.  These two have the same tonal curve which is mathematically more complicated but suffice to say its
similar to gamma 2.2 until the dark shadows.  If you look at the L values for a step in the dark range you'll see that
all compress the shadows a bit more than linear L values. 
(aside: in QTR I made a special ICC profile that is exactly linear in L -- called QTR-GrayLab and QTR-RGBLab -- this
essentially makes K (or RGB) values line up directly with L values).

If you print in Photoshop with Printer Manages Color and ABW in driver what I believe the system does is convert
your data to sRGB before sending to the driver.  So what you see in the two sets of data is this conversion of
AdobeRGB (or Gamma2.2) to sRGB (or Generic Gamma 2.2).

3) Mark mentioned the "perfect system for L* tone reproduction".  Ideally a straight-line from L=0 to L=100 would be
the goal.  But since neither pure black not pure white is achievable we need to connect L(dmax) to L(dmin) somehow.
The simple straight-line seems like the obvious compromise (and in fact this is what linearization in QTR curves does).
But as I discovered with help from others is that what's more important is that the prints that come out match the
screen better.  You have to have the mid tones match -- in other words the ideal curve has to go through (50,50).
With the examples given here that happens pretty close because dmin and dmax of photo paper is very close to the
white and black.  The L=4 to L=96 line does go through (50,50).  But with matte paper the much lower dMax
illustrates this better.  A L=16 to L=96 line would go through (50,56) -- resulting in a noticeably lighter print.

Scaling for various dMax's is where Black Point Compensation (BPC) came about.  Adobe has a paper about it in all
the gory mathematical detail.  Bottom line is you should scale in the Luminance Y values rather than Luminosity L values.
See http://www.brucelindbloom.com if you want more detail about the math.

This is the main reason for using the CM of the system to match L values from source file to L values of the print
in such a way to give the best perceptual match given dmax/dmin limitations.

Roy
Logged
NeroMetalliko
Jr. Member
**
Offline Offline

Posts: 64



« Reply #79 on: June 05, 2013, 03:09:04 AM »
ReplyReply

The graphs are interesting info about several different aspects.  So I have a few comments.

First, I've used similar graphs a lot over the years in making QTR curves.  It's the best info about
how well you are doing with the linearity for both tones and neutrality.

Hello Roy,
it's very nice to have your comments here,
many thanks for this.

As you perfectly know, the plot I do are nothing new, in fact they are only a slightly rearranged looking versions of the ones you still provide in text mode as output of your  QTR-tool Smiley

Quote
1) about the initial data and methods.  I have to agree with I think Mark A. the Epson STD being done without
an ICC profile shows just the response of the the driver and really isn't a good comparison of what you should
get printing normally.  The ABWs however are supposed to be tuned from the get-go so they are indicative
of what you will actually get.  

Maybe Alan (who made the test) could help to better clarify what exact settings were used in the labeled "Epson std" data set.

Assuming the strip was kept untagged and the ACPU set to no color management, what need to be confirmed is whether in the Epson driver panel,
under the "mode" menu it was selected "Epson std" or "Off (no color management)" before printing (they are mutually exclusive).
Only in the second case we could tell that the measured data are related to a raw fully unmanaged workflow, but labeling it as "Epson std" was confusing, if the case.
Having read "Epson std" as label in the provided data I have interpreted it as the above menu option was set to "Epson std" and not "Off (no color management)", but a final confirmation from Alan could be useful to finally clarify the issue.

Quote
(I wonder about the ABW-D3 data in both cases because it seems to go to L=100
rather than paper white of about L=96, but not too big a deal).  

You are right, this is strange in fact, it seems that the 5-6 readings near L* 100 in the ABW-D3 data set are abnormally lifted, this is something I forgot to ask to Alan about, and I'm still not easily able to figure out how it could happens. In any case it affects only the ABW-D3 plots, so we can even decide to keep these plots out of our thoughts without big impact.

Quote
I am surprised at how cool the darks are (b=-2.5) in all the ABW cases.  I don't think you've mentioned the ABW paper that was selected in printing but it may
have different characteristics than the Silver Rag paper you actually used.

I was surprised too, for this reason I have remarked that the results should be considered tied to the given printer/paper/settings combination.
Maybe Alan could specify what media setting was used, it could be nice to know.
 
Quote
2) the difference shapes of the curves in the dark shadows illustrate the difference between the different embedded
profiles.  AdobeRGB and Gamma 2.2 are true gamma curves -- pure exponentials.  sRGB and Mac's special
Generic Gamma 2.2 are pretty close for most of the range but you can see they are pretty different in the very
darkest range.  These two have the same tonal curve which is mathematically more complicated but suffice to say its
similar to gamma 2.2 until the dark shadows.  If you look at the L values for a step in the dark range you'll see that
all compress the shadows a bit more than linear L values.  
(aside: in QTR I made a special ICC profile that is exactly linear in L -- called QTR-GrayLab and QTR-RGBLab -- this
essentially makes K (or RGB) values line up directly with L values).

If you print in Photoshop with Printer Manages Color and ABW in driver what I believe the system does is convert
your data to sRGB before sending to the driver.  So what you see in the two sets of data is this conversion of
AdobeRGB (or Gamma2.2) to sRGB (or Generic Gamma 2.2).

During the development of my linearization tool I have written some math script in Octave, based on LindBloom and Adobe tech info, in order to precisely calculate the RGB to Lab (and inverse) transformations for different colorspaces (currently sRGB, Adobe RGB and ProPhoto RGB). I have exactly implemented the ideal 2.2 gamma curve, the sRGB gamma curve (with their linear part in the deep shadows and a 2.4 exponent) and even the Adobe adopted gamma slope limit of 32 used in all ACE engines conversions when pure gamma curves colorspaces are involved.
I perfectly know that you are well aware of all of this stuff Smiley
I have written something regarding these arguments in my linearization thread here (Reply #18):
http://www.luminous-landscape.com/forum/index.php?topic=78142.0

That said, I tend to agree with you regarding the fact that the shape of the extra dark zone in the gamma 2.2 plots is probably the result of an untagged K strip, being plotted as a gamma 2.2 L* x axis, but being printed assuming to be in sRGB-gamma format from the Epson driver (not only in ABW but even in the so labeled "Epson std" mode).
This was the reason why I decided to plot again all the data using a sRGB-gamma conversion for L* x axis.

What I find interesting is that if this is true, this could be considered a big proof of the fact that by default Epson driver expect all the inages to be encoded in sRGB and this is valid for ABW mode too.
Since there was a long time debate on the net if this was the case or not, with different opinions, and due to the fact that in my (limited) test I rarely, if ever, have seen this effect in my gamma 2.2 (Adobe RGB) assigned strips, I was one of the people starting to believe that the default expectations of Epson driver was gamma 2.2.
Now I'm much less sure, and it could really be that Epson driver, as you correctly point out (and matching Eric Chan opinion for example), by default, is expecting a sRGB encoded image in all their modes (ABW included).

Quote
3) Mark mentioned the "perfect system for L* tone reproduction".  Ideally a straight-line from L=0 to L=100 would be
the goal.  But since neither pure black not pure white is achievable we need to connect L(dmax) to L(dmin) somehow.
The simple straight-line seems like the obvious compromise (and in fact this is what linearization in QTR curves does).
But as I discovered with help from others is that what's more important is that the prints that come out match the
screen better.  You have to have the mid tones match -- in other words the ideal curve has to go through (50,50).
With the examples given here that happens pretty close because dmin and dmax of photo paper is very close to the
white and black.  The L=4 to L=96 line does go through (50,50).  But with matte paper the much lower dMax
illustrates this better.  A L=16 to L=96 line would go through (50,56) -- resulting in a noticeably lighter print.

I agree with you, this is something we have already covered in my Linearization thread.
Currently my approach is to match the linear line between real ink black and paper white readings as general target because I considered it a good starting point for my early development.
It could be absolutely possible to implement a second options, by adopting an "S" shape target, instead of the pure line, still anchored to real black and white readings, which could mimic the ideal 45 degree 0-100 L* linear in the midtones zone with good match, at the expense of shadows and highlights compression.
As you rightly highlight this "S" target could produce better perceptive results in matte papers more suffering from limited Dmax. In this case the shadows compression should be carefully monitored to avoid excessive "crushing" potentially occurring, in my opinion.

Quote
Scaling for various dMax's is where Black Point Compensation (BPC) came about.  Adobe has a paper about it in all
the gory mathematical detail.  Bottom line is you should scale in the Luminance Y values rather than Luminosity L values.
See http://www.brucelindbloom.com if you want more detail about the math.

This is the main reason for using the CM of the system to match L values from source file to L values of the print
in such a way to give the best perceptual match given dmax/dmin limitations.

I agree with you, and, as already told, I am often using the wonderful Bruce website as one of my most important technical information source.
I will look for the Adobe paper you mentioned for sure.

Many thanks again for sharing you precious thoughts, it is very appreciated.

Ciao Smiley
Andrea
« Last Edit: June 05, 2013, 04:06:43 AM by NeroMetalliko » Logged
Pages: « 1 2 3 [4] 5 »   Top of Page
Print
Jump to:  

Ad
Ad
Ad