What I mean is .... If you do as you suggest for QTR and print a equally spaced gray wedge target using QTR, read it with a spectro and process the measurements through QTR, you will get a set of linearization numbers that you can input into the QTR curve to linearize it. So you can linearize QTR, If you do as you suggest for ABW, you can't directly input those numbers into ABW as you can with QTR, so you can't directly linearize it.
I don't use QTR at all, I have tried it in the past but then I have developed my own tool which is the only I use. I print in Photoshop using my DeviceLink linearization profiles.
Now, my gray wedge targets are equally spaced in L*, not in RGB values, and this is a first difference (this means that I have a set of target for each RGB working space and a related DeviceLink correction profile for each paper/printer combination).
I print my BW target with ABW (using the proper paper/printer/settings) and then I read the target, in this way I have a map of the non-linearity of the ABW printing process. Then I use some math to calculate a compensation curve in order to get a L* straight line form max black ink to paper white. This curve is finally implemented building a DeviceLink profile which I do myself too.
At this point you can print the BW target with the DeviceLink applied (or use a second target set with twice the patches, if preferred):
the final result is a near perfect L* line. I can iterate with a second compensation and refine the DeviceLink further if I think it is needed, and check the result again (usually just the first pass is enough, I never go over two steps because it often becomes overkill).
In my workflow the DeviceLink curve IS the way you can directly inject the linearization numbers inside ABW (embedding it in the image prior to print it), and this is exactly a L* linearization, as the verification plot can easily confirm. The shadows are never compressed, quite the opposite.
I have read reports elsewhere of people who instead have use the linearization output from QTR to create curve corrections to achieve the same effect, which is what I assume you're doing as a first step. I didn't mention it because my impression is that people use ABW for its simplicity. Once you need to adopt such a complex workflow using workarounds, then it's no longer simple, and those looking for a more complex workflow will probably switch to QTR.
With ABW if you want it simple you still have it.
But if you want it to be linearized you can do it too, it is less simple but it is possible.
And ABW is not only for simplicity, ABW dMax values (and lightfastness) for example can easily be the best you can achieve. With all the other things on par by using an ICC workflow in Photoshop for B&W instead of the ABW you will probably end with inferior dMax values. If the tone is neutral enough (and often it is) I'm really maximizing the results with ABW and, in my opinion, the efforts required for the further enhancements I get are well worth.
Please, note I still think that even with a good use of QTR or Qimage you can get really top results in any case, it is less simple than using the standard ABW workflow but it's possible.
The same probably applies to DeviceLink profiles. I confess that I don't know a lot about them, despite a little Googling. I'd be interested to learn more, but the same point probably applies, i.e. at what point does this become too complex for most of the intended ABW target audience?
The "complex" part obviously is to get the DeviceLink profile done, but this kind of "complexity" is the same for every ICC creating process too, nothing more, maybe it is more simple because the targets are with 18 - 54 patches only, so not big targets like for a full ICC build.
Indeed I think you are right in the sense that if the hassle to have any kind of extra step over the straight ABW use is considered too much, this approach is not better (but not worse) than any other kind of approaches, from this point of view.
Once you have the profile, only few mouse clicks are required in Photoshop to apply the DeviceLink profile to the image before printing it: it is very easy and simple.
I don't want to lose sight of the OP, for whom all this discussion is probably mumbo-jumbo. For such people ABW is either works simply it doesn't. I standby my comment that QTR would be a step up in terms of print quality, split-toning, and ease of linearization.
I agree, ABW is simple and it works, so it's a good solution if no extra step are preferred, and I think that it could be often better than a straight Photoshop + canned ICC workflow for B&W.
QTR could be a step up if you want something more, but it's less simple.
My tool simply it doesn't exist as released product: it's an experimental work of myself, so it cannot be considered a public alternative at all, and for this reason my observations here are more an "academic" talk at this point.
By linear in luminosity I mean that if you print a equally spaced gray wedge target using either QTR or ABW, read it with a spectro and plot the luminosity then the values lie on a straight line. Perceptual linearity is a term used by the creator of QTR, Roy Harrington, for what happens when you convert an image to a QTR-derived ICC, and if you search the QTR forum then you will find his references to it and explanations of it. I mentioned it because if you convert your image to an ICC produced by QTR then your print will no longer be linear in luminosity - the shadows are compressed. You sometimes read suggestions to convert your image to the ICC to linearize it, including in this thread, but it doesn't work that way.
Ok, you have to pay attention to the way you plot the data, however in my case I can confirm you that the results will be a straight L* line, so we can agree that this IS linear in luminosity and, in my case, it IS possible to achieve it even using ABW.
Many thanks for your answer, I really appreciate it.
Ciao,
Andrea
