The aim in creating a QTR curve is to make it linear. That is, you want the luminosity values that it prints to lie on a straight line when you graph them. This gives you an even separation of tones in your print. You print out a test chart like the 21x4 chart that ships with QTR using the curve in question, and measure the luminosity results (using Measure Tool or similar software) and plot them. If you look at the two charts I posted in post 17 of this thread, you'll see a pink line. You want the luminosity values that you plot - the black lines - to lie on this pink line or fairly close to it.
If you create QTR curves using the curve creation tools that ship with QTR, then this is easy to do, assuming that you have a spectrophotometer like an i1. Printing and measuring the 21x4 gives you the linearisation numbers, and you can append them to the curve. Reprint using the updated curve and remeasure, and you'll get a the black line to sit on or close to the pink. There are various tutorials around about how to do this.
But K7 curves don't work like this. They are created by Jon Cone and InketMall using their own proprietary software, so you can't use the QTR curve creation tools to re-linearise them.
So what to do if you need to re-linearise a K7 curve that needs it? Well, (i) you can pay IJM $99 per curve to create custom curves for your printer. (ii) An alternative that various people here have adopted is to create their own custom spreadsheets to hack the .quad files that contain the curves and do some approximate re-linearisation.
Roy is suggesting (iii) - when you print the 21x4 and measure the patches (Measure Tool again), you can also use the results to create an ICC profile. QTR ships with QTR-Create-ICC which does this. You drop the measurements on this program and you get an ICC. Again, there are instructions around about how to do this. You've got a shiny new i1 Photo, so give it a try. You can at least use the ICC for soft-proofing (see below).
Now the question is, how you use the profile? Well, one thing you can do is use the ICC to soft-proof how the image will print. In the Jon Cone workflow, you *do not* convert to this profile, so in Photoshop you must select "preserve numbers" in your soft-proofing options (I'm not sure whether you can do this in Print Tool, even though that's what you'll print with). This works, as you can see how the image will print, including the effects of any non-linearity in the curve. You can edit the image to compensate for the non-linearity, but your curve is still non-linear.
What Roy is suggesting is to *convert* the image to the ICC for printing. What effect will this have? Well in that post 17 of mine, I demonstrated this. The first chart is from measuring my curve, and as you can see it's not very linear. For the second chart, I converted the 21x4 to the ICC I created from the initial measurements, printed it and remeasured. You can see that doing this has evened out all the wiggles in my curve - it's nice and smooth. But it's a long way from that pink line. Since it's below it this means that the image will print darker after the conversion. You can use your new ICC to soft-proof this, but this time leave "preserve numbers" unchecked.
Why is the ICC designed to have this effect? Because Roy designed the ICCs to enable you to print the file in a "perceptually linear" manner. I.e. one that looks roughly like it does on the screen. But to do this it is compressing the shadows quite a lot, and even Roy conceded above (post 24) that this compresses the shadows, and you may need to edit them to compensate.
So that's why you'd tell Print Tool (not PS) to convert the image to the ICC rather than print in GG2.2 - to straighten out the wiggles in a non-linear curve. And that's also why you wouldn't do it - because it darkens the print and looses shadow detail (how much it does this will depend on the image).
So as I said in my interim reply, there are two quite different philosophies in play here. Jon Cone places a premium on preserving shadow detail and opening the shadows, and tells you *never* to convert to the ICC, only to use it for soft-proofing. But if you follow this workflow, you've got no (easy) way to re-linearise the K7 curve. You've got to pay $99. Or develop a spreadsheet to hack the .quad file.
Roy on the other hand suggests that you create an ICC and use Print Tool to convert the image to it for printing. If you've got an i1 or similar then this is easy, and it gives you a form of re-linearisation. But you'll get quite a different (darker) print to the intentions of the piezography system.
Does linearisation matter? Well It gives you a known starting point and reference point for printing. It means you're not editing every image to compensate for curve non-linearity.
This is long, but I've tried to make it simple.
---In QuadtoneRIP@yahoogroups.com, <tracy@...> wrote :
"For those on Mac's instead of trying to re-linearize K7 .quad files which is hard to do,
go the ICC route. Just create a custom grayscale ICC built on top of the existing .quad file.
Then print using this ICC.
...The MeasureTool issue is still a pain -- I run Parallels on the Mac with WinXP running so
"
OK: I hate to be the dullest blade in the drawer, but here goes...
I'm using Piezography. Right now I create my image and print it thru PrintTool w/NCM. Seems OK to me, but if I can make it better. (I just got a shiny new i1 Photo Pro, so I want to play...)
Does "... create a custom grayscale ICC built on top of the existing .quad file" mean to print out a target thru a .quad file for a given paper, just as if I were printing a photo? Then measure that result into an .ICC profile?
If so, then will such a profile make use of all the K7 inks when I print? Exactly where do I choose to use that profile? In the PS dialog, or in the PT driver?
Doesn't an ICC profile run off to the Profile Connection Space, which is a color space? Are not the Piezography curves just curves (amounts of ink) and doesn't that completely bypass the PCS?
I feel like this 70 year old brain is beating around the edges of what's being said here. Just not wrapping my brain around how this works, out here in the real world, one step after another.
and then I get to "The MeasureTool issue is..." Now I'm really confused. I thought the Measure Tool wasn't involved in the creation of an ICC profile. Wouldn't that be used if I were trying to directly linearize the curves instead of generate a profile?
Likely someone is slapping their forehead right now, but if I don't ask, I won't learn...
Thanks for your patience.