Topic: Linearity of ABW (Read 14669 times)

Jeff-Grant · « **on:** October 23, 2015, 08:17:18 pm »

ABW has some appeal and it is well regarded by many on Lula and elsewhere, so a friend and I conducted an experiment to test the linearity of ABW. As a printer in the digital age, I want to know that my profile, curve or whatever is linear. It strikes me that linearity is an excellent starting point for printing B&W. It's what you'd want in a QTR curve. To ensure that we both had colour management off, he used Qimage on Windows, and I used Print Tool on OS X, printing on Epson Enhanced Matte. To do the test, we printed 21x4 test targets in GG2.2 using each of the five standard ABW settings - from light to darkest - and plotted the results.

What is abundantly obvious is that ABW does not produce a linear print. "Dark" is probably closest, but that's an on-average comment, and if this were in QTR I would relinearise it immediately. Which I can't for ABW.

I hope that I am missing something in what I am doing, and would love to hear what it is. Do others find this an issue with ABW?

I am attaching a screen grab showing the ABW modes as they were measured.

Tim Lookingbill · « **Reply #1 on:** October 23, 2015, 10:42:18 pm »

So you're expecting a linear rendering on matte paper which has a reduced dynamic range over your own display including the level of black density due to higher absorption.

Remember a printer is doing a very complex magic trick with a viewer's eyes in that different papers are expected to render the same image the same way despite the changing black density for shadows to gradually scale upward without abrupt transitions to lighter tones.

What happens when you test linearity printing on high gloss which provides the highest dynamic range with most dense blacks? I don't do these kinds of analysis with printers nor do I have expensive measuring equipment but I have acquired a knowledge of how ink and paper behave on different papers.

How do photos look on this less than linear matte paper? Can you see the non-linearity in the actual print of a photo?

Jeff-Grant · « **Reply #2 on:** October 24, 2015, 12:24:46 am »

Quote from: Tim Lookingbill on October 23, 2015, 10:42:18 pm

So you're expecting a linear rendering on matte paper which has a reduced dynamic range over your own display including the level of black density due to higher absorption.

Remember a printer is doing a very complex magic trick with a viewer's eyes in that different papers are expected to render the same image the same way despite the changing black density for shadows to gradually scale upward without abrupt transitions to lighter tones.

What happens when you test linearity printing on high gloss which provides the highest dynamic range with most dense blacks? I don't do these kinds of analysis with printers nor do I have expensive measuring equipment but I have acquired a knowledge of how ink and paper behave on different papers.

How do photos look on this less than linear matte paper? Can you see the non-linearity in the actual print of a photo?

Tim, I didn't start this thread to have a 'let the prints talk' discussion, rather, I am asking for feedback from folks who have experience in using ABW, and can comment on how this non-linearity expresses itself. This has nothing to do with screen-to-print matching, at least not in the first instance. It is simply about the performance of ABW, and the inability to linearise it.

With QTR, I can achieve linearity on any paper that I choose to measure. I see that as the starting point before I start to use a paper. It would appear that I can't do that with ABW.
Below are two grabs of printing using a QTR curve. The first is before linearisation and the second is after linearisation.

My question is, has anyone else found this inability to linearise ABW, and its fairly strong departures from linearity to be an issue, and if so, how have you dealt with them?

Tim Lookingbill · « **Reply #3 on:** October 24, 2015, 03:10:49 pm »

I get what you're doing, Jeff.

I was just trying to learn something myself and hoped you'ld fill in with an explanation of what you're attempting to achieve. Something that's been rolling around in my head about black density variation and how it changes the scaling of shadow detail the lighter or darker black density appeared to relate to those curves you posted.

Those curves appear as if there's some kind of compensation with regard to changing black density and its affect on contrast which is a part of the dynamic range of both the display and print.

Didn't want to hijack your thread by veering off onto print matching. I'll bow out of this and hopefully someone here can help you out with your issue.

Jeff-Grant · « **Reply #4 on:** October 24, 2015, 03:52:04 pm »

Thanks Tim. Mine is a much more basic question.

Tim Lookingbill · « **Reply #5 on:** October 24, 2015, 04:54:43 pm »

Got to admit though taking a second look at your opening post those ABW curves sure are smooth as if they were designed that way intentionally compared to your gnarled straight line linear correction in your second sample. Wonder if it means anything.

Ferp · « **Reply #6 on:** October 24, 2015, 10:03:30 pm »

Quote from: Tim Lookingbill on October 23, 2015, 10:42:18 pm

Can you see the non-linearity in the actual print of a photo?

The answer to this question is yes, if you use a workflow based around linearity.

I think that a lot of folks using ABW have gotten used to the ABW look and have adapted their workflow and editing to suit that look.

Jeff-Grant · « **Reply #7 on:** October 24, 2015, 11:48:56 pm »

Linearity is the only workflow that I know. That's the way of QTR and Piezography as far as I understand it. How do folks adapt to the ABW look?

Tim Lookingbill · « **Reply #8 on:** October 25, 2015, 02:04:20 am »

Quote from: Jeff-Grant on October 24, 2015, 11:48:56 pm

How do folks adapt to the ABW look?

I don't know how those curves you posted affect the look of an ABW print. I don't have an ABW printer but from the shape of those curves it appears that mid and darker tones would appear lighter which would affect contrast.

TylerB · « **Reply #9 on:** October 25, 2015, 02:48:39 am »

Jeff, I don't use ABW, but have assisted others establishing ABW workflows for their printers and papers. I find the prospect of printing B&W without linearization all but unacceptable. Creating profiles using QTR profiling tools and then utilizing Print Tool to incorporate into an ABW print workflow works very well, and of course Print Tool has other very useful functions as well.
The option without Print Tool is to preconvert the file to said profile in PS, then print with printer manages color, admittedly some inconvenient additional steps.
I'm sure there is some viable Windows variant.

Of course, then, either option allows the use of same profile for soft proof during editing. Observing the A and B values of given ABW hue settings on given papers can be useful also.

A conversion to profile, one way or the other, is not technically linearization, but since QTR profiles only "correct" luminosity, it may be considered such.
My $0.02

RHPS · « **Reply #10 on:** October 25, 2015, 05:41:38 am »

In my experience it's more important to have a smooth curve than one which averages to perfect linearity. A "best fit" line can be very misleading - if there are short-range changes in slope these can give visible banding in gradients. I think Tim Lookingbill is right to question your "gnarled straight line linear correction". It would be interesting to see a more detailed curve, say from a 51-step wedge to show short-range changes a bit better.

Jeff-Grant · « **Reply #11 on:** October 25, 2015, 06:07:43 am »

The general consensus is that a 51 step gives odd results. It introduces all sorts of twists. I really don't want to go down that rabbit hole. I'm really asking how people work with ABW and the inherent non-linearity

I should have posted a smoother linearisation. That one is unusual. I only postef it as an example of a linearised curve for QTR

Jeff-Grant · « **Reply #12 on:** October 25, 2015, 06:11:31 am »

Quote from: TylerB on October 25, 2015, 02:48:39 am

Jeff, I don't use ABW, but have assisted others establishing ABW workflows for their printers and papers. I find the prospect of printing B&W without linearization all but unacceptable. Creating profiles using QTR profiling tools and then utilizing Print Tool to incorporate into an ABW print workflow works very well, and of course Print Tool has other very useful functions as well.
The option without Print Tool is to preconvert the file to said profile in PS, then print with printer manages color, admittedly some inconvenient additional steps.
I'm sure there is some viable Windows variant.

Of course, then, either option allows the use of same profile for soft proof during editing. Observing the A and B values of given ABW hue settings on given papers can be useful also.

A conversion to profile, one way or the other, is not technically linearization, but since QTR profiles only "correct" luminosity, it may be considered such.
My $0.02

Thanks Tyler. I hear what you are saying. I would just like to understand how folks work with ABW

Tim Lookingbill · « **Reply #13 on:** October 25, 2015, 02:28:34 pm »

Don't know if this will help but after digging around online (trying to find what the 'A' in ABW stands for..

) I found this blog...

http://www.ronmartblog.com/2010/08/how-to-using-epsons-advanced-b-photo.html

...that might explain the reason for the curve's smoothness and affect on the image according to their namesake.

Scroll down to "ABW Test Image 2 - Zena" it describes by the author how it changes the look of individual elements in the test image. Some interesting tidbits including Eric Chan's take on ABW along with some other well known photographers as impetus for the blog topic are mentioned.

Note the last line...

Quote

UPDATE: These values correspond to gamma changes (i.e., 1.8, 2.0, etc…) , which is why it is okay to use something besides Normal.

Jeff-Grant · « **Reply #14 on:** October 25, 2015, 07:09:34 pm »

Thanks Tim. It is an interesting read. I guess that he spells it out quite clearly. It's a process of trial and error with a lot of knob twiddling until you find a solution that works. I find the limited soft proofing to be the main sticking point for me. I can certainly see the advantages of ABW in the things that I can't do today with Piezography such as toning. It occurs to me that I can achieve the same result with QTR and soft proof more easily.

Alan Goldhammer · « **Reply #15 on:** October 26, 2015, 04:40:16 pm »

Sorry I'm a bit late to this discussion. What you have observed is not news and was documented with the release of the Epson 3800. Giorgio Trucco has a lot of information here: http://www.outbackphoto.com/printinginsights/pi045/essay.html#20070123 Kieth Cooper, a frequent contributor to LuLa, has lots of B/W stuff here: http://www.northlight-images.co.uk/digital_black_white.html including information on creating QTR ABW profiles. There are others as well. It looks like you already read the material that Eric Chan has posted. I believe Eric used a 51 step patch set when he created his profiles (I had him do some for me some years ago when I had an Epson 2880).

I've done QTR profiles for all the main papers that I use and found that a 21 step B/W patch set is more than adequate for getting a good linear output. I generate the patch set and read using ArgyllCMS. You need to only focus on either the Dark or Darker tone setting to find out which works best for the particular paper you print on. the other settings don't give decent results IMO. I know for the 3880 Eric Chan only used those two settings in the profiles he created.

Finally, I collaborated with Mark McCormick-Goodhart over at Aardenburg to look at the stability of Hahnemuhle Photo Rag Ultra Smooth and various tone combinations of the ABW driver. You can access those results on the website.

Jeff-Grant · « **Reply #16 on:** October 26, 2015, 04:48:46 pm »

Thanks Alan. I'll take a look at it.

John Lytton · « **Reply #17 on:** October 30, 2015, 06:21:37 pm »

I’m very late to this discussion, but I’ve learned a lot as a long-time lurker in LuLa and would like to finally contribute by sharing my procedure for linearizing a printer for ABW.

Basically, I use the Ideal Densities chart from Tom Moore’s Quadtone RIP guide (http://www.quadtonerip.com/User%20Guide.pdf) . I make a 13 step density strip in Photoshop (0%, 5%, 10%, 20%, 30%, 40%, 50%, 60%, 70%, 80%, 90%, 95%, 100%), measure the resulting densities, compare to the Ideal Density chart, make an adjustment curve with those percentage points, tweak it, and print again. Iterate until there’s a close match. I can usually achieve a very close match after 3-4 iterations.

There may be better and/or more efficient ways, perhaps with QTR itself, but this has worked for me and is pretty straightforward for me as well. Apologies if I'm reiterating a procedure that already evident to many.

Jeff-Grant · « **Reply #18 on:** October 30, 2015, 07:41:33 pm »

Thanks John. Your technique and other variants come up fairly regularly on ten QTR forum. I assume that you only do this once per paper?

John Lytton · « **Reply #19 on:** October 30, 2015, 08:30:44 pm »

Ah yes, Jeff. I presumed the this sort of technique was used by others - I just hadn't seen it since I don't visit the QTR forum.

And yes, I do it once per paper/ink-media setting. But every now & then I'll run another density test to see if there's a significant batch to batch variation.

Regards,

John

Author Topic: Linearity of ABW (Read 14669 times)