Profile/curve for ABW - fundamental question

[Ferp]

Work is busy but I've done some testing.  My hunch is that the cameratico article was written back in the day when ABW accepted ICCs, and so was a profile aware application.  It mattered what ICC you tagged your test patches in.  I never used this version of ABW so I've not seen this, and I have no way to access it now to check.  But that's my hunch, because the cameratico article workflow tells you to do this, and if I understand his results, then tagging the test chart with the grey lab space seems to have some effect.  If you are still using that printer (3800?) and the old driver then it may still work for you. He must have experimented with different ICCs until he found one that in conjunction with the color management engine of ABW created linear results.  Again just a hunch.

But that's not the case now.  I just followed his workflow as best I could (harder in current versions of Photoshop - see below) and I also printed his test chart tagged with gray gamma 2.2, and measured them.  The results were identical.  Tagging with different ICCs had no effect. 

So all my comments earlier still hold, including about options 1-5 in the original post.  Both the latest version of ABW and QTR are ICC agnostic applications and it doesn't matter how you tag your test charts.  You just need to ensure that you have color management off to print the test charts for creating an ICC.  Also for printing images via ABW if you wish to use the no color management workflow.

Some other observations.

The scripting tool that Alan refers to is just different terminology for the QTR-Create-ICC and QTR-Create-ICC-RGB applications that we've been discussing all through this thread.

Turning off color management in recent Windows of Photoshop is getting harder and harder.  In the past we've discussed here on Lula how to avoid the silent sRGB conversion by Photoshop that you'd get under the printer manages color setting on Windows.  See for example http://forum.luminous-landscape.com/index.php?topic=103432.0 .  It looked to me today like some of these tricks no longer work.  Qimage is then your best option.

Those who want to print to ABW on a Mac using and ICC should open their copy of Print Tool, and in the bottom right corner in the 'Print Color Management' section, ensure that 'Epson ABW mode is checked, and then next to it, that 'Print-Tool Managed' is selected in the drop-down list of options, and below that select you profile and rendering intent etc and proceed, and you should have the same functionality as the Chan profiles in the old driver used to provide.

[texshooter]

I need a drink.

[Doug Gray]

Turning off color management in recent Windows of Photoshop is getting harder and harder.  In the past we've discussed here on Lula how to avoid the silent sRGB conversion by Photoshop that you'd get under the printer manages color setting on Windows.  See for example http://forum.luminous-landscape.com/index.php?topic=103432.0 .  It looked to me today like some of these tricks no longer work.  Qimage is then your best option.

The null transform trick to print w/o color management continues to work fine in Photoshop under Windows, including the latest Win 10. I make a point of checking it because of its uncertain future.

http://forum.luminous-landscape.com/index.php?topic=117516.msg973997#msg973997

[texshooter]

Who cares if Photoshop does a silent conversion or not?  The ABW driver is going to convert it to sRGB anyways. Unless something screwy happens when an image is converted twice in succession to sRGB?

[Ferp]

I need a drink.

That was my reaction too, when I read that article and tried to understand and replicate the results.  Why doesn't someone try and replicate mine?  Just make sure that all color management is off.

The null transform trick to print w/o color management continues to work fine in Photoshop under Windows, including the latest Win 10. I make a point of checking it because of its uncertain future.

http://forum.luminous-landscape.com/index.php?topic=117516.msg973997#msg973997

I'll have to read that exchange between you and MHMG in detail later.  Yes the null conversion still works, but my impression is that there are less options than there used to be.  The author of the cameratico article suggests setting the printer profile to the QTR-gray-lab monochrome ICC that his test chart is tagged with, but monochrome ICCs don't appear in the list.  I'm not sure they ever did, as the printer driver presents to software as an RGB device. 

I think what I used to do was convert my gray gamma 2.2 image to AdobeRGB, since it's essentially the same thing, and then set it as the printer profile.  But AdobeRGB no longer appears in the list of printer profiles in the printer.  Perhaps my memory is faulty, but I thought it was there, along with ProPhotoRGB and sRGB, and without these the null transform is harder to use.  Not impossible, but harder.  That's what I meant.

Who cares if Photoshop does a silent conversion or not?  The ABW driver is going to convert it to sRGB anyways. Unless something screwy happens when an image is converted twice in succession to sRGB?

There are two options here.  One is to specify Photoshop Manages Colors in the print dialog.  Under this option you specify the printer profile to be the same as the document profile, if you can (see just above).  Under this option there's no silent conversion, and neither Photoshop nor ABW will silently convert to sRGB. 

The other option is to specify Printer Manages Colors in the print dialog.  Under this option, you will get a silent conversion to sRGB.  According to some comments by an Adobe software engineer over on TOP last year, this is an intentional design decision by Adobe in relation to Windows, and nothing to do with the Epson driver.  I seem to recall is has to do with protecting less expert users, and something to do with likely future changes in the Windows printing pipeline.  The workaround for this workflow is to convert from gray gamma 2.2 to AdobeRGB and then assign sRGB.  At this point Qimage is looking good.

Need another drink?

(Edited for a typo)

[Doug Gray]

I'll have to read that exchange between you and MHMG in detail later.  Yes the null conversion still works, but my impression is that there are less options than there used to be.  The author of the cameratico article suggests setting the printer profile to the QTR-gray-lab monochrome ICC that his test chart is tagged with, but monochrome ICCs don't appear in the list.  I'm not sure they ever did, as the printer driver presents to software as an RGB device. 

The null transform works fine with any printer profile. At least in RGB space. Just assign any paper profile. It doesn't have to even be one for a paper you have or even printer you have as long as the profile is installed and a printer driver exists for that printer.

Here's the workflow I use:

For patch target sheets, I just assign an arbitrary printer profile. It doesn't even have to be a profile for the printer you use but can be any RGB profile for any model printer. Then, I select Photoshop manages color and print using ICC disabled with the same printer device settings used for printing color managed images. The settings in Photoshop such as Perceptual, Relative, Absolute, and black point have no affect at all since Photoshop simply assumes that the RGB values, which are already in printer space, Are the same as whatever you select. Otherwise the null transform wouldn't work.

I can't say how this would affect ABW operation. I don't do B&W often but when I do, I have a special patch set with several hundred extra neutral, and near neutral, patches. I only use a profile made from this for B&W (in RGB color mode)  though since it can create some odd profile characteristics in more saturated areas. At least with I1Profiler.

An ongoing concern I have with Adobe is that they will either prohibit this outright, or worse, round trip from device space back to Lab PCS then back to device space. So I test this with new updates just in case though I think it unlikely they will start doing it.

I suspect this approach of using a printer profile to effect a null transform would work fine on Apples but I don't have an Apple to test it. Wouldn't surprise me at all if the Apple issue isn't something that only shows up in conventional (matrix) RGB spaces. Perhaps that's why Photoshop has removed most of them from being selecting when Photoshop manages color. Perhaps someone with IOS could test this.

[Alan Goldhammer]

For patch target sheets, I just assign an arbitrary printer profile. It doesn't even have to be a profile for the printer you use but can be any RGB profile for any model printer. Then, I select Photoshop manages color and print using ICC disabled with the same printer device settings used for printing color managed images. The settings in Photoshop such as Perceptual, Relative, Absolute, and black point have no affect at all since Photoshop simply assumes that the RGB values, which are already in printer space, Are the same as whatever you select. Otherwise the null transform wouldn't work.

I can't say how this would affect ABW operation. I don't do B&W often but when I do, I have a special patch set with several hundred extra neutral, and near neutral, patches. I only use a profile made from this for B&W (in RGB color mode)  though since it can create some odd profile characteristics in more saturated areas. At least with I1Profiler.

An ongoing concern I have with Adobe is that they will either prohibit this outright, or worse, round trip from device space back to Lab PCS then back to device space. So I test this with new updates just in case though I think it unlikely they will start doing it.

Why not just use the Adobe Color Print Utility which is designed to print without color management?  I only use Argyll to do profiles (both color and ABW) and that's my method of choice.

[Doug Gray]

Why not just use the Adobe Color Print Utility which is designed to print without color management?  I only use Argyll to do profiles (both color and ABW) and that's my method of choice.

ACPU works fine as does I1Profiler direct, but I prefer being able to position multiple targets and control borders precisely which is easy to do in Photoshop. Once I'm in Photoshop, I prefer to print without leaving to run another app. And I have to be careful to replicate exactly the same printer device settings so more room for mistakes. Also, ACPU, for some reason, doesn't accurately size the images.

[unesco]

[...] Also, ACPU, for some reason, doesn't accurately size the images.

I usually enlarge image by additional 3 or 4% in printer driver to get the right size on page

[Alan Goldhammer]

Also, ACPU, for some reason, doesn't accurately size the images.

Yes, a well known bug that Adobe knows about and shows no signs of wanting to fix.  I think the patches from Argyll end up about 4% smaller than they should be but it never is an issue reading them.

[Doug Gray]

As an aside, since printer profiles are RGB in device space but PCSLAB on the other side. If Adobe decides to roundtrip the RGB values to PCS and back they have to go through the conversion. Since roundtripping is not error free the way to detect if this occurs is to compare a printed patch set against the stated, not color managed, results from ACPU. The simplest way to do this which would also show the greatest error is using a small color patch set along the 3D gamut boundaries. I use the 56 patch set generated by patchtool for this purpose. This patch set is all the RGB values with steps of 85 from 0 to 255 and the 3D boundary subset are the ones with at least one of the RGB values either at 0 or 255.

The reason this is effective is that the conversion from PCSLAB to RGB is particularly bad at the gamut boundaries. The cubes in the highest resolution I1Profiler profile are in steps of 7.1 along the a* and b* axes and errors are unavoidable. They average about 1 dE with a max of somewhat over 3 purely from the mathematical conversions. The printer itself adds a bit to this, but under .5 dE so it is quite easy to see if the printed, null transform set from Photoshop is consistent with that obtained from ACPU using this small set of patches.

I've attached the patch set if any are interested.

[Ferp]

The thread is starting to wander, and we shouldn't lose sight that it started from a request for advice on ABW workflow.

The null transform works fine with any printer profile. At least in RGB space. Just assign any paper profile. It doesn't have to even be one for a paper you have or even printer you have as long as the profile is installed and a printer driver exists for that printer.  ..... I can't say how this would affect ABW operation.

I was aware of that approach.  The reason I didn't adopt it in this case was that I was trying to test the thesis, in that cameratico article, that the choice of profile to assign did make a difference to ABW.  I didn't think that it did make a difference, and my results confirm that, but I needed to do the test under specific conditions.  There are instances where the choice of tagged profile may make a difference.  When we have discussed the null transform previously, Printfab was also mentioned as needing it to operate, and without rereading those threads I don't recall whether the choice of assigned profile has any impact on it.

I suspect this approach of using a printer profile to effect a null transform would work fine on Apples but I don't have an Apple to test it.

You don't need it on an Apple.  This issue blew up in the last year or two because C'tein over on the TOP blog controversially advocated Printer Manages Color over a color managed workflow for color printing.  In the comments section it was pointed out that this didn't work so neatly on Windows for the reasons that we've been discussing, and that the behavior also affected ABW. The different behavior between Windows and Mac remains puzzling, but it ain't going to change.  It means ABW on Windows is less simple.

An ongoing concern I have with Adobe is that they will either prohibit this outright, or worse, round trip from device space back to Lab PCS then back to device space.

Your concerns are well founded because an Adobe engineer made that threat in the TOP thread.  He was not at all sympathetic about the problems that the current behavior was causing, and the further problems that removing the null transform would create.

I regard ACPU as sufficiently broken on Windows as to make it useless for most purposes.  If you are printing targets you can print from your profiling software.  I only recommend it to those people whose printers I am creating profiles for remotely, as it is simplest for those unfamiliar with the intricacies of color management.

[Alan Goldhammer]

You don't need it on an Apple.  This issue blew up in the last year or two because C'tein over on the TOP blog controversially advocated Printer Manages Color over a color managed workflow for color printing.  In the comments section it was pointed out that this didn't work so neatly on Windows for the reasons that we've been discussing, and that the behavior also affected ABW. The different behavior between Windows and Mac remains puzzling, but it ain't going to change.  It means ABW on Windows is less simple.

I believe you have this backwards.  It is Apple that broke the printer pipeline several years ago and there was a long thread about this on LuLa.  Eric Chan stopped making profiles for ABW driver because MacOS was not going to support their use.  ABW on Windows has not changed one iota in the whole time I've been printing with Epson printers (9 years now).

I regard ACPU as sufficiently broken on Windows as to make it useless for most purposes.  If you are printing targets you can print from your profiling software.  I only recommend it to those people whose printers I am creating profiles for remotely, as it is simplest for those unfamiliar with the intricacies of color management.

I'm sorry but this is another puzzling statement.  Why do you say it is broken on Windows.  I use ArgyllCMS and it only generates targets; you need to print them outside of PS or LR and ACPU works just fine for that purpose.  ACPU was designed to do one thing and one thing only, print out non-color managed targets for profiling.  Of all Adobe products, the directions for ACPU use are the clearest I've seen:  https://helpx.adobe.com/photoshop/kb/no-color-management-option-missing.html

[texshooter]

I'm running in circles. If the whole purpose of printing a test chart is to objectively measure (with a spectrometer) how accurately the printer reproduces tone (I'm only speaking of Epson's ABW printing right now, and only speaking of Windows), then why in the world would you turn off the ABW driver before printing the test chart?  How can we evaluate how well the ABW driver performs if we turn it off?

[unesco]

The thread is starting to wander, and we shouldn't lose sight that it started from a request for advice on ABW workflow.

Frep, thank you very much for noting that :-).
During coming weekend I will have more time to test new approach of generating new sRGB patch set within PS and printing it through ABW what should be the right workflow - then compare it to options #1-#5.

[Ferp]

I believe you have this backwards.  It is Apple that broke the printer pipeline several years ago

When I said "you don't need it on an Apple", I was referring to the null conversion trick. There isn't a silent profile conversion if you select printer manages colors on OS X.

What follows is something of a thread diversion, my apologies to the OP. There are different printing quirks on both Windows and Mac. My recollection is that the trio - Epson, Apple & Adobe - are all variously involved. One of the frustrating things is that these issues are not well documented. I'm not going to derail this thread by trying to rectify that here, and I'm not the best person to do that anyway.

We are talking principally about the quirks affecting Windows in this thread. The best place to read about them are in some comments by senior Adobe software engineer Dave Polaschek in these threads on TOP:

http://theonlinephotographer.typepad.com/the_online_photographer/2015/07/are-profiles-obsolete.html

http://theonlinephotographer.typepad.com/the_online_photographer/2015/07/andrew-rodney-on-the-importance-of-custom-printer-profiles.html

http://theonlinephotographer.typepad.com/the_online_photographer/2015/09/product-review-epson-surecolor-p800-printer.html

http://theonlinephotographer.typepad.com/the_online_photographer/2015/10/photoshop-vs-printer-managed-color-printing.html

Although I warn you in advance that trying to synthesise his comments in all these is a struggle. Dave is the most authoritative voice you'll find on the silent profile conversion problem on Windows. It's real. I've measured it. If you've not seen it in your workflow, well, I can't explain that.

On OS X, I wasn't assigning any responsibly for the removal of ICCs from ABW. There was a change made by Apple in 10.6.8 that caused Roy Harrington to write Print Tool. You can research this yourself if you want to want to understand it. As I said, both OS have their quirks, but they're different.

My comment on ACPU was too brief.  Its placement on the page is broken under Windows, and I found that enough of a pain not to use or recommend it.

why in the world would you turn off the ABW driver before printing the test chart?

I've lost track of which comment this in is response to. I hope not one of mine. You don't turn it off. You turn off color management, and that is done in the Photoshop print dialog. In the Epson Dialog you select ABW and any options related to it.

@unesco: I look forward to your measurements.

[Alan Goldhammer]

Here is one link to the discussion several years ago about ABW, PS, and the OS:   http://forum.luminous-landscape.com/index.php?topic=62900.0  I think my first post also has a link to the first discussion.  Enjoy reading "old" history.  I'm going to graciously exit this discussion as I really don't have anything else to offer.

[unesco]

Here is one link to the discussion several years ago about ABW, PS, and the OS:   http://forum.luminous-landscape.com/index.php?topic=62900.0  I think my first post also has a link to the first discussion.  Enjoy reading "old" history.  I'm going to graciously exit this discussion as I really don't have anything else to offer.

Thank you Alan - worth reading!

[texshooter]

I decided to abandon ABW QTR  ICC printer profiling in favor of a more seat-of-the pants approach.  I wouldn't be surprised if the results are just as good, either.  Here's my plan.

Print out a variety of 2.2 gamma grayscale step wedge targets ( e.g. Keith's new chart  )

Then fashion and save to disk a PS contast curve to make the test chart as seen on my monitor to match as close as possible to the chart as seen on print.  All by eye and by hand only.  No profiling software, no spectrophotometers, no OS pipeline workarounds. Just  me and my reading glasses.  If the printer is not 100% linearized straight out of the box, then so what.  As long as the monitor matches the print, I'm good to go. I can soft proof 'till I'm happy.

Wish me luck.

[unesco]

I decided to abandon ABW QTR  ICC printer profiling in favor of a more seat-of-the pants approach.  I wouldn't be surprised if the results are just as good, either.  Here's my plan.

Print out a variety of 2.2 gamma grayscale step wedge targets ( e.g. Keith's new chart  )

Then fashion and save to disk a PS contast curve to make the test chart as seen on my monitor to match as close as possible to the chart as seen on print.  All by eye and by hand only.  No profiling software, no spectrophotometers, no OS pipeline workarounds. Just  me and my reading glasses.  If the printer is not 100% linearized straight out of the box, then so what.  As long as the monitor matches the print, I'm good to go. I can soft proof 'till I'm happy.

Wish me luck.

It will be 100% perceptually linear approach!