Reading QTR 21 step wedge with Densitometer

[R Shaffer]

So I'm probably being really dense here, but how do I convert the reflective density readings from my X-Rite 810 to the %K steps on the printed step wedge. I have been using the scanner to create my curves so far and that's working pretty good. But I would like to compare the results and fine tune.

I saw the "Ideal Densities for various Dmax" table in the user guide, but it only goes down to a Dmax of 1.6

TIA

 

[Ron-san]

So I'm probably being really dense here, but how do I convert the reflective density readings from my X-Rite 810 to the %K steps on the printed step wedge. I have been using the scanner to create my curves so far and that's working pretty good. But I would like to compare the results and fine tune.

I saw the "Ideal Densities for various Dmax" table in the user guide, but it only goes down to a Dmax of 1.6

TIA

Rob-- I also do not know how to convert reflective density readings into %K steps. If anyone can enlighten me I would also appreciate it. Anyway, that is why I use a flatbed scanner to measure tonal steps in prints. I know how to convert scanner readings into %K and that seems to work just fine in making QTR profiles.

Cheers, Ron Reeder

 

[clay]

There is an excel spreadsheet i have that allows you to put in the dmax of a particular process and spit out the equivalent dot percent. The joker in the deck is the N- factor, which takes into account the matte nature of a particular substrate. I usually use 1.5-2.5 for the matte paper that we alt printer types use in our process. It is based on a academic formula called the Yule-Neilsen relationship.

It is too large to use as an attachment here, but anyone who wants it can send me an email at clay (at) clayharmon (dot) net and I will send it to them.

 

[R Shaffer]

Thanks Ron,

I don't feel quite so dense now. The manuals you have put together for creating digital negatives are great. I used your pure pd curve as a starting point for my curves. It took 12 test strips to dial in, but I made a couple prints last weekend that look great. Thanks for the effort.

Thanks Clay,

PM sent. It will be interesting to see how your spreadsheet compares to my data from the scanner. I did a bit more research on the Yule-Neilsen formula. How do you approximate your value of N?

I'm printing pure pd on Arches plantine. Right now I'm only getting a Dmax of around 1.3, but I've just started with the pd and need to work on my coating process a bit. My pd emulsion is not absorbing into the paper as well as my kallitype emulsion.

 

[pschwart]

Thanks Ron,

I don't feel quite so dense now. The manuals you have put together for creating digital negatives are great. I used your pure pd curve as a starting point for my curves. It took 12 test strips to dial in, but I made a couple prints last weekend that look great. Thanks for the effort.

Thanks Clay,

PM sent. It will be interesting to see how your spreadsheet compares to my data from the scanner. I did a bit more research on the Yule-Neilsen formula. How do you approximate your value of N?

I'm printing pure pd on Arches plantine. Right now I'm only getting a Dmax of around 1.3, but I've just started with the pd and need to work on my coating process a bit. My pd emulsion is not absorbing into the paper as well as my kallitype emulsion.

If you get a dMax of 1.4-1.45 with Pd on Platine you will be doing well. Note that too much absorbtion of the emulsion will lower your dMax.

 

[R Shaffer]

QTR 21 Step Wedge & YN Formula

Did some playing around with Clay's spreadsheet and this is pretty cool. In searching about the net on the topic of the N value, it seems that it is intended as an adjustment to achieve visual 50% dot gain. So you visually pick your 50% dot gain and then read its density value and adjust the N value until the 50% density value matches your density reading. ( please correct me if I am wrong here, quite possible )

So I'm sure there may be some way to scientifically determine N, but for my purposes it is rather irrelevant. And here is the experiment I have started.

I took a kallitype image I had previously printed with a QTR step wedge. The step wedge looked fine ( to me ) through the highlights, but seemed to visually have too big a jump from 100% black to 95% black. This was confirmed on another image that had a lot of shadow values and they appeared to transition harshly.

I have attached a graph of the values I got when I scanned the image ( %K vs Measured %K ). It would indicate that my shadows were fine, but my midtones were too bright.

Then I picked an N value such that it matched my 50% density reading. It turned out to be N = 1.45. The second attached graph is my measured density vs. predicted density for N = 1.45. This graph would indicate that my highlights are good, but my shadow values are not dark enough. What I seem to be finding in my visual evaluation of some of my prints.

I was able to reasonably match the scanned curve by setting my N=1.3

Now it certainly won't be as easy to create the initial curve with density readings. But once I have an .acv curve thats close, I could certainly use these values to fine tune the curve. Hope to have some time this weekend to create a new curve based upon this and see what it looks like.

 

[R Shaffer]

If you get a dMax of 1.4-1.45 with Pd on Platine you will be doing well. Note that too much absorbtion of the emulsion will lower your dMax.

Well that may be my problem. My kallitype sensitizer soaks right in quickly and I rarely can get above 1.35 My early pd tests, I did not use enough sensitizer and it really seemed to just want to pool on the surface. So I then tried a bit of paper humidification and it went on more like my kallitype sensitizer. But I had too much sensitizer. So I need to be more careful in controlling my process as I am making multiple changes each try.

 

[carioca]

Another approach using only your densitometer, could be by implementing you logD readings straight into the 'Linearization'-line of QTR. This will linearize your 21 steps.
Make a new test print, and the steps should now be nicely distributed.
Depending on the Dmax of your process, you will have to evaluate the logD value of a 50% mid grey by eye, there is no absolute fixed value or calculation to get it. In my case, a pleasing 50% grey has a logD of around 0.65 (with Dmax 1.8-2.0)
I then measure the step on the new test print, that has this .65 value and note the %-reading of the stepwedge, (for ex. 45% on my wedge corresponds to my visually pleasing 50% grey).
Back in QTR, I then make one grey curve correction, shifting the 50 value to 45. The reason I only make one curve correction is the fact, that this 50% shifting (between anchor point 0 and 100), works as a gamma point shift.
This works pretty well for me.

Sidney

So I'm probably being really dense here, but how do I convert the reflective density readings from my X-Rite 810 to the %K steps on the printed step wedge. I have been using the scanner to create my curves so far and that's working pretty good. But I would like to compare the results and fine tune.

I saw the "Ideal Densities for various Dmax" table in the user guide, but it only goes down to a Dmax of 1.6

TIA

 

[R Shaffer]

Sidney,

I like the simplicity of your approach. What are you printing with a dmax of 1.8-2.0? Silver Gelatin ?

My steps don't seem to be linear enough to have a single point correcting the curve. I need a couple of pushes and pulls here and there along the way to get the boys nicely lined up.

 

[carioca]

Sidney,

I like the simplicity of your approach. What are you printing with a dmax of 1.8-2.0? Silver Gelatin ?

My steps don't seem to be linear enough to have a single point correcting the curve. I need a couple of pushes and pulls here and there along the way to get the boys nicely lined up.

Hi Rob,
I work with the carbon transfer process.

This is exactly what the linearization process of QTR is about, 'it lines up the boys', the 'one' point shift then adjsts your mid-grey.
Have you actually tried to input your x-rite values in QTR? You should try and see what happens to your curve.
(start with the densest value, then down to lowest)

Sidney

 

[R Shaffer]

Cool. I had not realized carbon had such a high dmax.

I'll give it a try. I just did a few tests last night trying to get my dmax up with pure pd. I had been just letting the paper air dry. So I did one test with a hair dryer after 3min and one test adding a drop diluted LFN as a surfactant. No improvement over air drying. The single drop of LFN coated very nicely, but looked a bit salty in the shadows. So I was going to dilute the LFN further and try a 10min air dry prior to hitting it with the hair dryer.

I could certainly throw in trying out your method for curving as I am mostly looking at dmax.

 

[Ron-san]

This is exactly what the linearization process of QTR is about, 'it lines up the boys', the 'one' point shift then adjsts your mid-grey.
Have you actually tried to input your x-rite values in QTR? You should try and see what happens to your curve.
(start with the densest value, then down to lowest)

Sidney

Hi Sidney--

Just got back from teaching a QTR workshop at Project Basho in Philadelphia. Nice place and a great bunch of students. Today I tried to use your method of linearizing a QTR profile. In my hands it does not work (maybe I am missing something essential?).

I took an old profile that makes very linear print tones on a palladium print. I pulled the gray curve out of the profile and used the resulting profile (minus gray curve) to print out a negative of a 21 step wedge. I then printed this negative on palladium. I then read the reflective density of each of the 21 steps with my Xrite 810 densitometer. The resulting densities were plugged into a LINEARIZE= function in the above (minus graycurve) profile. I quickly learned that the function wants at least a 0 before the decimal point of each reading. And the values must always change in the same direction from one value to the next. Which was a problem since several of the lightest values had the same reading as did a couple of the darkest values. To fix that I just deleted repeats of the same value (so I ended up with less than 21 values inserted).

Anyway, pressing onward, the profile installed and gave smooth, reasonable looking ink curves. I used this density-linearized profile to print a negative of my own trusty step wedge, made a palladium print, and read out its tonal values with my usual flat bed scanner method. The print tones were much too dark in the middle of the scale. Worse, the necessary Gray Curve, needed to force the tones into linearity, could not be described by a simple curve with a single point in the middle. It was much too complex for such a simple adjustment.

So, I learned something. I did not know that density values plugged into a LINEARIZE= function would work at all on a negative. Why they seem to do so, I do not understand. But I do not see how a simple three point gray curve will finish the linearization. And since you still need to derive the usual fairly complex gray curve, I do not see the advantage of measuring and plugging in the densities in LINEARIZE.

As usual, I am probably missing something. Any clue as to what??

Cheers, Ron Reeder

 

[carioca]

Hi Sidney--


I took an old profile that makes very linear print tones on a palladium print. I pulled the gray curve out of the profile and used the resulting profile (minus gray curve) to print out a negative of a 21 step wedge. I then printed this negative on palladium. I then read the reflective density of each of the 21 steps with my Xrite 810 densitometer. The resulting densities were plugged into a LINEARIZE= function in the above (minus graycurve) profile. I quickly learned that the function wants at least a 0 before the decimal point of each reading. And the values must always change in the same direction from one value to the next. Which was a problem since several of the lightest values had the same reading as did a couple of the darkest values. To fix that I just deleted repeats of the same value (so I ended up with less than 21 values inserted).

Anyway, pressing onward, the profile installed and gave smooth, reasonable looking ink curves. I used this density-linearized profile to print a negative of my own trusty step wedge, made a palladium print, and read out its tonal values with my usual flat bed scanner method. The print tones were much too dark in the middle of the scale. Worse, the necessary Gray Curve, needed to force the tones into linearity, could not be described by a simple curve with a single point in the middle. It was much too complex for such a simple adjustment.

So, I learned something. I did not know that density values plugged into a LINEARIZE= function would work at all on a negative. Why they seem to do so, I do not understand. But I do not see how a simple three point gray curve will finish the linearization. And since you still need to derive the usual fairly complex gray curve, I do not see the advantage of measuring and plugging in the densities in LINEARIZE.

As usual, I am probably missing something. Any clue as to what??

Cheers, Ron Reeder

Hi Ron,
for my carbon process, I have a base profile without gray curve that prints fairly even from 0-100% density (contrast can be controlled in carbon by using different sensitizer strengths, which is my first approach prior to applying a correction curve).
I do get increasing density values one way, which then are evened out with the linearizing feature of QTR, and I obtain a quite perfect greyscale in the subsequent test print.
QTR does indeed have this flaw of not accepting non increasing or decreasing density value input. If you leave out some of your density values (because they were repeats), will not make the linearization process work. The input values always have to be evenly spaced, for example with a 21 step wedge, you can use all 21 values, or you could use only every second step (1,3,5,7,etc.), Missing out some of the steps would simply compress your step wedge in that area and QTR will completely falsify the linearization process.
My described 'one point' shift does not make any sense, if the linearization has not been carried out correctly.

I hope this clears things up a bit,
let me know.

Sidney

 

[Ron-san]

Missing out some of the steps would simply compress your step wedge in that area and QTR will completely falsify the linearization process.
My described 'one point' shift does not make any sense, if the linearization has not been carried out correctly.

I hope this clears things up a bit,
let me know.

Sidney

Sidney-- Thanks. I think it is now obvious where I goofed. However I routinely find that I cannot get smooth increases at the extreme endpoints and need someway to guesstimate and work around that situation. In making a gray curve with a flat bed scanner I can usually just drop out those flat points and the resultant curve still works OK. So maybe I will stick to the scanner, which I know how to use, and leave the densitometer alone until I can figure out a better approach.

But thanks for showing us yet another way to work in QTR.

Cheers, Ron Reeder

 

[carioca]

the right starting curve

Sidney-- Thanks. I think it is now obvious where I goofed. However I routinely find that I cannot get smooth increases at the extreme endpoints and need someway to guesstimate and work around that situation. In making a gray curve with a flat bed scanner I can usually just drop out those flat points and the resultant curve still works OK. So maybe I will stick to the scanner, which I know how to use, and leave the densitometer alone until I can figure out a better approach.

But thanks for showing us yet another way to work in QTR.

Cheers, Ron Reeder

Ron,
QTR is a great tool to tweak curves, but it cannot create process curves out of nothing, we do need a good curve to start with.
Logical 'guesstimation' and curve testing is still a main part of the deal!

Sidney