Pages: [1] 2 3 4   Go Down

Author Topic: Profile/curve for ABW - fundamental question  (Read 15795 times)

unesco

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 254
Profile/curve for ABW - fundamental question
« on: July 24, 2017, 04:19:08 pm »

I conduct some extensive research on ABW behavior on my P800 and 3880 also making QTR profiles.
One fundamental question comes to my mind - I have read tons of information in the last 2 years about ABW linearization and have not found definitive answer - 21/51 patch set and its profile embedded for test print. Should it be:
1) as is the target e.g. in QTR folder, no profile attached,
2) attached GrayGamma2.2 as ABW assumes it,
3) attached linear gamma profile e.g. QTR GrayLab,
4) attached linear gamma profile e.g. QTR GrayLab and then converted to GrayGamma2.2,
5) any other?

1 and 4 give different results. That should be something obvious but now I am mixed-up and really appreciate valuable answers. In the Internet I have found different approaches. I am asking only about the target file to be printed, not the rest of the process.
Logged

Alan Goldhammer

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 4344
    • A Goldhammer Photography
Re: Profile/curve for ABW - fundamental question
« Reply #1 on: July 24, 2017, 04:54:30 pm »

I don't understand your question.  Are you printing the patch set to make a QTR profile?  If this is the case you need to print the patch set out with no color management in the same way you would profile a colored inkset.  I use the Adobe print utility program to do this and make sure that that your printer driver is correctly configured.
Logged

Ferp

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 295
Re: Profile/curve for ABW - fundamental question
« Reply #2 on: July 24, 2017, 08:29:49 pm »

How are you printing?  Which operating system and which program?  If under Windows and from a recent version of Photoshop then there's a problem that we've discussed here before, which is that Photoshop will silently convert all images printed using the printer manages colors setting to sRGB, which is the setting you need to use for ABW.  This is tricky to work around.  ACPU is one solution, but then you have the same problem again when you print images, and you can't use ACPU for that as its image placement settings are broken under Windows.  OS X also has its CM quirks, but at least there is Roy Harrington's Print Tool.

More generally, I'd measure linearity using the same workflow that I'd use to print images.  You print your images in gray gamma 2.2?  Do the same for the targets.  For ABW it shouldn't matter whether the image is unassigned or you assign it to grey gamma 2.2, as ABW ignores the ICC. 
Logged

texshooter

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 575
Re: Profile/curve for ABW - fundamental question
« Reply #3 on: July 24, 2017, 08:56:49 pm »

I Think I understand what Unesco is asking.  Researching the web yields contradictory recommendations on how to build and print a 21-step test chart for the purpose of creating linearized ICC profiles with QTR and for Epson printer destination.  He wants to know which method (he listed four) is best and more importantly WHY?  Perhaps they all work but in different situations?  Some say that the test chart to be printed should have no color working space assigned to it.  Some say it should have a 2.2 gamma space assigned.  And apparently QTR has their very own color space as well. . . Why would that even be necessary?

I have been struggling with this issue myself and can never get a satisfying answer. But let me take a crack at it, and someone can correct me.

Which color working space (e.g., aRGB 2.2, Prophoto 1.8, QTR Gray Lab 1.0, Gray Gamma 2.2, or NONE) that should be assigned to a test chart before it is sent to the printer depends on which color space that test chart was sourced from and intended to have by its author.    Just because an image has no color space assigned to it doesn't mean that it has no color space. All images have a color space baked into it even if it has no tag or sign that says: Hello world, my creator made me to be viewed in the sRGB color space!   Every pixel must have an RGB or L*a*b* or CMYK number value associated with it.  How else would the monitor know which color to display esch pixel with  or how to tell one patch on a test chart apart from another patch? 

So if a calibration software, such as QTR or Colormunki, provides their own test chart digital negative, they obviously don't want you to go assigning some god-knows-what color space to it. They designed the chart to be used with their software and nobody else's. That's why they usually instruct to print the target with no profile. Some calibration systems, on the other hand, will ask for a 2.2 gamma profile assignment. That's because they designed their linearization program to work with 2.2 gamma test charts.

Am I warm?

Logged

Ferp

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 295
Re: Profile/curve for ABW - fundamental question
« Reply #4 on: July 24, 2017, 10:17:54 pm »

Warm but not hot. 

I will restrict myself to commenting on QTR.  It assumes nothing.  On Windows it is a stand alone application that is not color managed and whatever color profile is or isn't embedded in the image file will make absolutely no difference to how QTR prints the image.  It's up to you to manage the ICC - assign or convert (as per our other thread) - before you save the file and send it to QTRGui.  You can send an untagged image, or one where you've assigned any profile you like, to QTRGui and you'll get the same print.  The one thing that will make a difference is if you convert to another ICC, because then for an untagged file, Photoshop will assume your default grayscale working space and convert the file, changing it.  That's going to change the print, but not because QTR did it, Photoshop is the culprit.

On OS X, QTR is a printer driver and it too is color profile agnostic.  It's just a relatively basic printer driver.  It doesn't expect anything.  The printing software is another matter.  Both Photoshop and Print tool offer options to convert to an ICC on the way to QTR.  Since OS X 10.6.8 you can't print untagged or gray gamma 2.2 images from Photoshop using QTR without getting a silent profile conversion behind the scenes.  This was the precise reason that Roy Harrington created Print Tool - to allow you to do that.  Yes, ACPU does as well, but Print Tool is more functional.  So to get the same degree of control as in Windows, use Print Tool (or ACPU).

So QTR has no expectations.  The question of what ICC to use is really about what workflow and software you are using.  Most of the documented workflows assume gray gamma 2.2, so that is how you need to print the target.  Your print the target the same way you are going to print images, taking care to avoid any silent profile conversions.  If you print in Windows using QTRGui, or in OS X with Print Tool with CM turned off, you can print an unassigned target, or one with gray gamma 2.2 or AdobeRGB (same as gray gamma 2.2 but in RGB space), and it won't matter.  My impression is that Roy Harrington supplied his test charts as untagged images because he assumed that people would be printing them that way, in line with generally accpeted workflows, but that's just an impression, and it's about workflow, not software.

My understanding of the ICC profiles that ship with QTR, like gray-lab.icc, is that they were created by Roy Harrington as generic profiles for people to use long before devices like the i1 became widespread and people started creating their own.  They're mostly dated around 2005.

All images have a color space baked into it even if it has no tag or sign that says: Hello world, my creator made me to be viewed in the sRGB color space!

Yes and no.  An image file is just a bunch of numbers and an ICC profile is a guide to color managed software as to how to interpret it.  It's true that if someone sends me an untagged color image then I will want to know what color space they intended, because that will tell me how it is supposed to be viewed.  The same is true for a B&W image, but not so for test charts.  No-one asks what color space the x-rite test charts are in, for the purposes of creating a color profile.  And so it is for QTR under most workflows.

Logged

unesco

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 254
Re: Profile/curve for ABW - fundamental question
« Reply #5 on: July 25, 2017, 04:40:18 am »

Thank you all for your replies! Now, I am on my mobile so only a short answer.

I am on Windows, print patches using PS in the newest version and print using ABW on Epson 17" printers.
I use QTR in another workflow, so now QTR itself is not targeted in this discussion.

textshooter, you've got exactly my idea! That's what I mean and look for.
I will reply in more details when I get access to my PC
Logged

unesco

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 254
Re: Profile/curve for ABW - fundamental question
« Reply #6 on: July 25, 2017, 03:53:43 pm »

First, thank you all for such comprehensive replies.
The core of my problem should be defined as follows: "how to generate patch target for linearisation and ABW printing under Windows and PS (or maybe other software as ACPU or QImage)". It only looks as a simple question.

I had some discussions on local forum with a guy who has P10k and would like to make better ABW prints, and after long exchange of comments and print tests/measures on both sides I started to doubt I can answer such a question.

ABW mode on P800 looks very well, sometimes it is hard to distinguish it from QTR prints, so I would like to improve it, but first to understand its settings.

More generally, I'd measure linearity using the same workflow that I'd use to print images.  You print your images in gray gamma 2.2?  Do the same for the targets.
it only looks so simple - I would print the target that way, but how the target should look like? how to prepare the target? That could be my 1)-5) version of profile assingment/conversion in the first post.

He wants to know which method (he listed four) is best and more importantly WHY?  Perhaps they all work but in different situations?
Indeed, I can linearise the printer according to reference, but reference is different in each case.
Logged

Ferp

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 295
Re: Profile/curve for ABW - fundamental question
« Reply #7 on: July 25, 2017, 08:29:57 pm »

Your print the target the same way you are going to print images, taking care to avoid any silent profile conversions.
it only looks so simple - I would print the target that way, but how the target should look like? how to prepare the target? That could be my 1)-5) version of profile assingment/conversion in the first post.
Indeed, I can linearise the printer according to reference, but reference is different in each case.

Sorry about the long post.  As the saying goes, I didn't have time to write you a short one.  Probably not today either.  When I said to print the target as you would images, that assumed a workflow where you print without any color management.  That's one of two possible workflows.  The other is to use color management and print images using an ICC, which if I recall correctly is what Alan Goldhammer was recommending.  You don't often see the difference between these two alternative workflows discussed and compared.  People just assume or or the other.  Two different workflows, resulting in slightly different prints and requiring different image editing and different approaches to soft-proofing (see below).

First, thank you all for such comprehensive replies.
The core of my problem should be defined as follows: "how to generate patch target for linearisation and ABW printing under Windows and PS (or maybe other software as ACPU or QImage)". It only looks as a simple question.  ...  I would print the target that way, but how the target should look like? how to prepare the target? That could be my 1)-5) version of profile assingment/conversion in the first post.

Possibly I'm not thinking straight, but I can't see the problem of patch generation and preparation.  If you make a color profile, you have pages of standard patches supplied by x-rite.  No generation or preparation.  You print them as they are, which is usually untagged, being careful to turn off all color management.  You create a profile.  You use that profile to print.  QTR is no different.  It comes with standard 21 and 51 untagged patches, you print one of them without any color management, you create a profile using either the QTR-Create-ICC or QTR-Create-ICC-RGB application, and print using the profile.  You do the same if If you want to use QTR profiles with ABW.  I can't see the problem.

We've been discussing the other workflow, printing without color management, over in the digital B&W forum:
http://forum.luminous-landscape.com/index.php?topic=118989.0
which as I noted above, will result in slightly different prints and requiring different image editing and different approaches to soft-proofing.  But the targets used to create the ICC are printed the same way.

What am I missing?  Are we still talking at cross-purposes?  In relation to your four options, I consider #1 and #2 to be the same, since you print the targets without any color management enabled and tagged profiles should be ignored..  I haven't seen anyone recommend #3 or #4 for printing the QTR test patches.  You got a link?  I can't see that #3 would be any different so long as you assigned QTR GrayLab and didn't convert to it.  I would not recommend #4, and I don't understand why someone would.
Logged

texshooter

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 575
Re: Profile/curve for ABW - fundamental question
« Reply #8 on: July 25, 2017, 11:10:53 pm »

I haven't seen anyone recommend #3 or #4 for printing the QTR test patches.  You got a link?

Here is a link to a source that advocates #3.
http://cameratico.com/guides/black-and-white-icc-profiles-and-soft-proofing/

Where I get confused is where the author criticizes the use of gray gamma 2.2 test charts and insists on a 1.0 gamma chart, particularly the QTR GrayLab variety.  Didn't  you say that it doesn't matter what gamma the chart is encoded with because  both the ABW driver and QTR are gamma agnostic? And because the ABW driver converts whatever it receives to sRGB gamma 2.2 anyways.  Something tells me I should be worried of "silent profile conversions" but not sure where to look.
« Last Edit: July 26, 2017, 12:34:50 am by texshooter »
Logged

unesco

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 254
Re: Profile/curve for ABW - fundamental question
« Reply #9 on: July 26, 2017, 01:59:11 am »

Sorry about the long post.  As the saying goes, I didn't have time to write you a short one.  Probably not today either.  When I said to print the target as you would images, that assumed a workflow where you print without any color management.  That's one of two possible workflows.  The other is to use color management and print images using an ICC, which if I recall correctly is what Alan Goldhammer was recommending.  You don't often see the difference between these two alternative workflows discussed and compared.  People just assume or or the other.  Two different workflows, resulting in slightly different prints and requiring different image editing and different approaches to soft-proofing (see below).
I think ABW is color managed under PS on one hand, on the other I do not want to use ICC, because I would like to minimise use of colour inks, especially Y on 3880, and with ColorMunki I cannot (except of Argyll) prepare good, neutral ICC.

Possibly I'm not thinking straight, but I can't see the problem of patch generation and preparation.
[...]
I can't see the problem.
I also have no problem to generate untagged patch file of any size e.g. using Argyll, but should it be untagged? (option #1)

What am I missing?  Are we still talking at cross-purposes?  In relation to your four options, I consider #1 and #2 to be the same, since you print the targets without any color management enabled and tagged profiles should be ignored..  I haven't seen anyone recommend #3 or #4 for printing the QTR test patches.  You got a link?  I can't see that #3 would be any different so long as you assigned QTR GrayLab and didn't convert to it.  I would not recommend #4, and I don't understand why someone would.
what you mean cross-purposes? there is no hidden agenda here ;-).
#1 and #2 should be the same, but because of another reason than your mentioned (PS do manages colors in this print somehow, since it converts images to sRGB for "Printer manages colors"). sRGB has gamma = 2.2 so the same as natively assumend in my PS setup for B&W.
#4 I have seen somewhere, and also it looks logical. what we want to make L linear so we need to have linear target (GrayLab assingment), but ABW expects GG2.2 so there is conversion to it later (am I saying stupid things?).

what is the most logical here is to prepare smooth gradient under PS in sRGB and then to limit it to 21 or 51 steps. this would require to switch-off PS dithering as well as work on 16 bits first, then to convert it to 8 at the end. what do you think?
Logged

Alan Goldhammer

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 4344
    • A Goldhammer Photography
Re: Profile/curve for ABW - fundamental question
« Reply #10 on: July 26, 2017, 08:43:21 am »

Here is a link to a source that advocates #3.
http://cameratico.com/guides/black-and-white-icc-profiles-and-soft-proofing/
That link is just the standard approach to using QTR for generating ABW profiles.  Note that this approach can only be used with WinOS as MacOS does not permit it (long thread several years ago here on LuLa when Apple eliminated this approach.  Eric Chan stopped making ABW profiles at that time.
Logged

Alan Goldhammer

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 4344
    • A Goldhammer Photography
Re: Profile/curve for ABW - fundamental question
« Reply #11 on: July 26, 2017, 08:46:51 am »


what is the most logical here is to prepare smooth gradient under PS in sRGB and then to limit it to 21 or 51 steps. this would require to switch-off PS dithering as well as work on 16 bits first, then to convert it to 8 at the end. what do you think?
You have a ColorMunki, correct?  The easiest thing is to use Argyll to generate either a 21 or 51 step B/W patch set, print it using the Adobe utility, read the data using Argyll and then use the QTR scripting tool to generate your ABW profile.
Logged

Ferp

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 295
Re: Profile/curve for ABW - fundamental question
« Reply #12 on: July 26, 2017, 09:38:08 am »

That link is just the standard approach to using QTR for generating ABW profiles.  Note that this approach can only be used with WinOS as MacOS does not permit it (long thread several years ago here on LuLa when Apple eliminated this approach.  Eric Chan stopped making ABW profiles at that time.

In what way is using a QTR-generated ICC to print ABW on a Mac using Roy Harrington's Print Tool any different to what Eric Chan's profiles used to allow us to do directly in ABW?
Logged

Ferp

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 295
Re: Profile/curve for ABW - fundamental question
« Reply #13 on: July 26, 2017, 09:55:11 am »

This is a preliminary comment on the cameratico article.  I don't understand how it could work.  It seems to imply that somehow printing a 21 step chart tagged with (assigned to) the  'QTR – Gray Lab' color space via ABW and creating an ICC via QTR should give different results to leaving it untagged or tagging it with gray gamma 2.2.  That's not my understanding and experience.  I thought that ABW was also ICC agnostic, at least the current version of it.  The original draft of the article was from July 26th, 2012 - was ABW ICC aware back then, as Alan pointed out that it used to be?

I've also got a few other uncertainties about that article and some apparent gaps in it.  I drafted a long and rambling post, but I think it's better if I find some time to test before I comment further.
Logged

Alan Goldhammer

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 4344
    • A Goldhammer Photography
Re: Profile/curve for ABW - fundamental question
« Reply #14 on: July 26, 2017, 10:37:24 am »

In what way is using a QTR-generated ICC to print ABW on a Mac using Roy Harrington's Print Tool any different to what Eric Chan's profiles used to allow us to do directly in ABW?
You cannot use ABW profiles on a Mac.  Don't ask me why as I have always been on Windows.  There is a whole thread on this from a couple of years ago that discussed the matter.  Had to do with an Apple decision
Logged

Alan Goldhammer

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 4344
    • A Goldhammer Photography
Re: Profile/curve for ABW - fundamental question
« Reply #15 on: July 26, 2017, 10:38:44 am »

  I thought that ABW was also ICC agnostic, at least the current version of it.  The original draft of the article was from July 26th, 2012 - was ABW ICC aware back then, as Alan pointed out that it used to be?
Yes it was.  I was using the scripting tool to make ICC profiles once Eric Chan stopped doing so.
Logged

Ferp

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 295
Re: Profile/curve for ABW - fundamental question
« Reply #16 on: July 26, 2017, 10:45:37 am »

You cannot use ABW profiles on a Mac.  Don't ask me why as I have always been on Windows.  There is a whole thread on this from a couple of years ago that discussed the matter.  Had to do with an Apple decision

You keep saying this, but I believe you are incorrect.  Are you on a Mac?  Have you tried to print using Roy Harrington's Print Tool, and enabling the ABW in the software and specifying an ICC profile and printing this way?  It's a bit hard to compare empirically now to the Eric Chan option, as that option has gone, but I am fairly certain that this is conceptually the same.  Perhaps that old thread pre-dated the arrival of Print Tool.  But before you keep repeating that claim, you should check the approach I suggest, because I don't think that your statement is correct.
Logged

texshooter

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 575
Re: Profile/curve for ABW - fundamental question
« Reply #17 on: July 26, 2017, 11:37:48 am »

Yes it was.  I was using the scripting tool to make ICC profiles once Eric Chan stopped doing so.

What is this scripting tool you speak of? If I got my hands on it, would I be able to make ABW ICC profiles for my Epson 3800 (Windows 7)  today? The same profiles Eric Chan provided (link below).  The kind of profiles meant for double color management.  By that I mean PS Manages Color turned on (with ICC selected) at the same time Printer Manages Color turned on (ABW mode selected). 

http://people.csail.mit.edu/ericchan/dp/Epson3800/printworkflow.html

Scroll down to the header title:   How to print B&W images with the ABW driver (with ABW profiles)
« Last Edit: July 26, 2017, 11:43:45 am by texshooter »
Logged

unesco

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 254
Re: Profile/curve for ABW - fundamental question
« Reply #18 on: July 26, 2017, 12:48:33 pm »

You have a ColorMunki, correct?  The easiest thing is to use Argyll to generate either a 21 or 51 step B/W patch set, print it using the Adobe utility, read the data using Argyll and then use the QTR scripting tool to generate your ABW profile.
Thank you for your kind answer.

Yes, I use Color Munki (I do not need anything more for B&W printing right now).

I think, I have pointed out my objectives not clearly enough - again, I DO NOT want to make ICC profile for ABW but would like to research ABW behavior with different settings. thus I need reference patch set.

I know how to generate UNTAGGED 21 or 51 step B/W patch set, I know how to measure it either with color munki software or Argyll, BUT do not know if the untagged version is correct (option #1 from my initial email) or maybe other target setup for patches is needed for PS/ABW print and WHY (PS/ABW is different in print technique and results so I suppose the patch set should be different)?
Is it clear enough now?

BTW, I have measured some of Eric Chan profiles for untaggest patch set under ABW print in PS - they are not linear.

So still now answer how the patch set should look like: options #1-#5...
« Last Edit: July 26, 2017, 12:53:11 pm by unesco »
Logged

Alan Goldhammer

  • Sr. Member
  • ****
  • Offline Offline
  • Posts: 4344
    • A Goldhammer Photography
Re: Profile/curve for ABW - fundamental question
« Reply #19 on: July 26, 2017, 01:03:05 pm »

What is this scripting tool you speak of? If I got my hands on it, would I be able to make ABW ICC profiles for my Epson 3800 (Windows 7)  today? The same profiles Eric Chan provided (link below).  The kind of profiles meant for double color management.  By that I mean PS Manages Color turned on (with ICC selected) at the same time Printer Manages Color turned on (ABW mode selected). 

http://people.csail.mit.edu/ericchan/dp/Epson3800/printworkflow.html

Scroll down to the header title:   How to print B&W images with the ABW driver (with ABW profiles)

http://www.quadtonerip.com/html/QTRiccprofile.html is the tool.  IIRC you use the RGB version of the tool to make the profile.  Keith Cooper has a description of how to do this here:  http://www.northlight-images.co.uk/black-and-white-print-linearisation-with-i1profiler-and-qtr/  these are the same type of profiles that Eric used to make.
Logged
Pages: [1] 2 3 4   Go Up