QTR-Quadtone RIP

Does QTR support printing with a LAB color space? I tried to do this with Print Tool as a test for my calibration procedures with LINEARIZE and see that LAB space is not supported.

Sandy

My method in linearizing with QTR is, I understand it, consistent with what others do. My procedure, basically is to print step wedge targets in the Gray Gama 2.2 color space with Print Tool using No Color Management. I have gotten very good results in practice by linearizing with these targets, using both the Measure Tool with ProfileMaker 5 and iProfiler.

There was in interesting discussion yesterday on the carbon forum in this thread,

https://groups.yahoo.com/neo/groups/CarbronTransfer/conversations/topics/8930?reverse=1

Some of the discussion seemed to conflict with my procedures so I asked Calvin for permission to use his comments on this forum, and I attach them directly as he stated, his words in italics. Thanks in advance for comments.

Message One from Calvin

Looking at the Eye-One-Readme text, Roy explains the linearization process with the 21 step tablet he provides. The problem is he is linearizing a file in the Dot gain 20 profile space with lab values. In the end he shows an ideal linearization and has an L value of 57 for his 50% patch, when the L value should be 65. Following your book, it looks like this is how you are doing it as well. This would explain why yours and Peter's print is coming out too dark. If one is going to linearize with L values directly, then don't print a test chart with patches calibrated to the profile spaces of dot grain 20 or gray gamma 2.2. Make a test chart with the values corresponding to L values.

Message Two from Calvin

So let me get this correct, you are printing a test chart in gray gamma 2.2 then reading the patches and calibrating to evenly space or linearize the L values? If this is what you are doing, your print is going to come out dark. L values are not evenly spaced in either dot gain 20 or gray gamma 2.2. If you try to evenly space or linearize the L values for a test chart in either profile, the calibration will be wrong. For example the L value for a 50% patch in gray gamma 2.2 is not 50, if you were to have a perfectly white paper and pure black, but rather 54. The 90% patch is not 20, but rather 6 and so on. The difference in dot gain 20 is even greater. The 50% patch has an L value of 62. Let's look at a real world example of calibration- L values for paper white might be 95 and black might be 10. If you take a test chart in dot gain 20 and read the 50% patch, it should read 61 which is not the midpoint between 95 and 10. The midpoint is 52.5. The difference being 8.5, so your print would come out 8.5% too dark.

A linearized curve prints evenly spaced luminance values (actual measured luminance values of ink) for a given target that is encoded with a given grayscale colorspace. Those luminance values are from dMax to dMin. (Dark to white) of the ink/paper.

So, if you have a dot gain 20 target and you linearize it, your dot gain 20 images will print linear when you print it again with the linearized curve.  If you have a gray gamma 2.2 target and you linearize for that, your gray gamma 2.2 images will print linear.

Now, the actual printed “L” values are not exactly going to be something like L50* corresponding to L*50 in the image because you don’t have a paper that is tile-white and a dMax that is black-hole of L*0. Only then would your middle L* value as seen in Photoshop correspond to the same L* value on the print. Instead L*50 in photoshop is going to print (on matte paper with Piezography Pro Ink for example) half-way between say L* 12.67 and L* 96.6.  So L*50 in photoshop would be L*60.97 on a given linearized curve with matte paper and pro ink.

The L* value of your pixel data as seem by the info palette is (usually) color-space independent. Colorspaces (when it come to grayscale) are basically virtual encodings at this point used only to make images backwards compatible with CRT monitors and older presses and output devices. The industry has settled on the gray gamma 2.2 standard for that reason, but really if history could be re-written we should probably just use a gamma of 1.0 to keep everything mathematically easy.

Long story short: keep your target in the grayscale space of your images. If you images are Dot Gain 20, your target should be. If you images are gray gamma 1.0, your target should be. We, as an industry, have settled on gray gamma 2.2.

Trouble happens when the target is gray gamma 2.2 and images are dot gain 20 (Photoshop still defaults to dot gain 20 for it’s color settings which is a huge problem btw.)

Best,
Walker

> On Oct 19, 2017, at 8:59 AM, sanking@... [QuadtoneRIP] <QuadtoneRIP@...m> wrote:
> 
> 
> My method in linearizing with QTR is, I understand it, consistent with what others do. My procedure, basically is to print step wedge targets in the Gray Gama 2.2 color space with Print Tool using No Color Management. I have gotten very good results in practice by linearizing with these targets, using both the Measure Tool with ProfileMaker 5 and iProfiler. 
> 
>  
> There was in interesting discussion yesterday on the carbon forum in this thread,
> 
> https://groups.yahoo.com/neo/groups/CarbronTransfer/conversations/topics/8930?reverse=1 <https://groups.yahoo.com/neo/groups/CarbronTransfer/conversations/topics/8930?reverse=1>
>  
> Some of the discussion seemed to conflict with my procedures so I asked Calvin for permission to use his comments on this forum, and I attach them directly as he stated, his words in italics. Thanks in advance for comments.
> 
>  
>  
> Message One from Calvin
> 
>  
> Looking at the Eye-One-Readme text, Roy explains the linearization process with the 21 step tablet he provides. The problem is he is linearizing a file in the Dot gain 20 profile space with lab values. In the end he shows an ideal linearization and has an L value of 57 for his 50% patch, when the L value should be 65. Following your book, it looks like this is how you are doing it as well. This would explain why yours and Peter's print is coming out too dark. If one is going to linearize with L values directly, then don't print a test chart with patches calibrated to the profile spaces of dot grain 20 or gray gamma 2.2. Make a test chart with the values corresponding to L values.  
> 
>  
> 
>  
>  
> Message Two from Calvin
> 
>  
> So let me get this correct, you are printing a test chart in gray gamma 2.2 then reading the patches and calibrating to evenly space or linearize the L values? If this is what you are doing, your print is going to come out dark. L values are not evenly spaced in either dot gain 20 or gray gamma 2.2. If you try to evenly space or linearize the L values for a test chart in either profile, the calibration will be wrong. For example the L value for a 50% patch in gray gamma 2.2 is not 50, if you were to have a perfectly white paper and pure black, but rather 54. The 90% patch is not 20, but rather 6 and so on. The difference in dot gain 20 is even greater. The 50% patch has an L value of 62. Let's look at a real world example of calibration- L values for paper white might be 95 and black might be 10. If you take a test chart in dot gain 20 and read the 50% patch, it should read 61 which is not the midpoint between 95 and 10. The midpoint is 52.5. The difference being 8.5, so your print would come out 8.5% too dark.
> 
>  
>  
>  
> 
>

> On Oct 19, 2017, at 12:21 PM, tubadude06@... [QuadtoneRIP] <QuadtoneRIP@yahoogroups.com> wrote:
> 
> To make sure we are on the same page, let's create a 50% patch in photoshop in gray gamma 2.2, in dot gain 20, and in Lab mode. With the same paper as before, having a dMax of 13 and a Dmin of 97, what should be the printed lab values for each patch?


> I would say the gray gamma 2.2 patch should read L*60,

Let’s assume your linearized your curve with a gray gamma 2.2 target (which is default).

The mid-point between 13 and 97 is (13+97)/2 which is L* 61.5. So L*50 in your 2.2 image would be 61.5 on the paper.

> the dot gain 20 patch should read L*64,

Are you saying that if you had a curve that was linearized for gamma 2.2 and then you printed a dot gain 20 target through that, it would by 64? Why would you want to do this? This is just not needed and will confuse people.

If you linearized a second curve using a Dot Gain 20 target, and printed your dot gain 20 target again through the linearized curve, your Photoshop L*50 would be exactly L* 61.5 on the printed target. Same as the GG2.2 environment. 

> and the Lab patch should read L*55. Correct? 

?? No. LAB is device independent and QTR doesn’t even support it as an output space. So I’m not sure what you mean by “LAB Patch”. In Photoshop? If you printed LAB encoded (essentially gamma 1.0 if QTR supported lab)? Again, not needed for fruitful discussion here.

//

Are you trying to figure out the encoded L* differences between grayscale spaces when an environment is calibrated for one space and you then print a different encoded target through it? If so why and for what purpose?

I think that the continual confusion on this list about grayscale spaces is that somehow they are like “photoshop curves” but they aren’t. They are different. They are just reference points. We’ve all agreed that Gray Gamma 2.2 is our magnetic North Pole and the we’ve aligned all our magnetic devices for that pole. Comparing the degrees from that magnetic North Pole and other poles is not a fruitful discussion. Instead, compare your local environment (a direction of a distant tree e.g.:  the L* in a measured target) to your magnetic North Pole (Gray Gamma 2.2). 

If you want to change your magnetic North Pole (e.g: switching to Gray Gamma 1.0), it’s fine. Just don’t compare it to the old North Pole (gray gamma 2.2) because it gets confusing at that point and does not help you navigate around the world.


Best,
Walker

> On Oct 19, 2017, at 4:52 PM, calvin.grier@... [QuadtoneRIP] <QuadtoneRIP@yahoogroups.com> wrote:
> 
> 13 and 97 is (13+97)/2 is 55, not 61.5 as you said.

You are correct. Sorry about that. My calculations were off. Typed in the wrong number. Also, I never assign, always embed . . . when dealing w/ grayscale output.

//
 
And I apologize again, I did not see your use of the “K” value until your most recent reply thinking instead you were using L* values. K values will indeed change. Non-tagged will be something like K of 64% (at the L*50 point) and converted/embedded with GG22 will be K 54% (at the L*50). K value is different than L*value which changes very slightly when converting (less than 1%). K percentage shows the difference in the gammas nicely (as a number) but I’ve never used it as a reference point as I find it doesn’t have any bearing really on a properly linearized environment. I prefer simple L* values . . . . sorry for my quibbling and confusion. I probably should not have been typing today! . . .

All the best,
-Walker