PS curves and QTR

[andy138]

does anyone know how to embed a PS derived curve into a QTR profile on a mac (no QTRgui)? I've finally gotten a half-way decent curve in PS, but I would like to embed it into a profile in QTR to avoid applying it in photoshop. Is this even beneficial, or should I just save the trouble, apply it in PS? I'd imagine that letting the rip take care of the curve avoids interpolation in PS, and therefore the possibility of smoother tones, but I am probably wrong. Anyway, any info would be a big help

cheers
andy

 

[donbga]

does anyone know how to embed a PS derived curve into a QTR profile on a mac (no QTRgui)? I've finally gotten a half-way decent curve in PS, but I would like to embed it into a profile in QTR to avoid applying it in photoshop. Is this even beneficial, or should I just save the trouble, apply it in PS? I'd imagine that letting the rip take care of the curve avoids interpolation in PS, and therefore the possibility of smoother tones, but I am probably wrong. Anyway, any info would be a big help

cheers
andy

Andy,

You may have seen this post on the QTR Yahoo list and it doesn't address your question directly but perhaps it would be helpful to you for using QTR to make digital negatives.

Dead Link Removed

Clay Harmon and Michael Mutmansky share some great tips about their appoach to curve building with QTR.

My guess is that you could take your PS input/output data points and plug them in to the GRAY CURVE setting for starters. Something like this:

GRAY_CURVE="0;2 4;7 11;13 20;19 30;24 40;29 50;36 60;42 70;50 80;60 85;68 90;75 96;90 97;93 98;95.5 99;97 100;98"


Thsi example is from one of Clay's QDIF files posted for the 2200 used for long scale alt processes such as pure palladium.

Not sure what your other QTR settings are such as ink limits so putting the PS curve data into the QDIF file might not lead to the optimum results.

Don Bryant

 

[Michael Mutmansky]

It is definitely better to apply the curve in QTR along with the other curve information that is being applied there (which happens mostly behind the scenes).

Well OK, let's say that it will never be worse than applying in PS, and in soma cases, it will be better, and in some cases, there may not be a noticeable difference.

Statistically, it is better to apply all adjustment curves at one time, rather than doing multiple curves at different times.

As for the thread that Don mentions, it's a little discombobulated. I think Clay and I need to do a tutorial on this. It's actually not terribly difficult, but keeping things straight can be a bit of a chore. We've talked about doing a workshop of some kind on this at some point, as we both feel it is by far the most powerful and flexible approach to negative-building that we have worked with. It really is worth the effort to develop a good curve with this approach.


---Michael

 

[Michael Mutmansky]

And to answer your question, I can't tell you right off because I use the PC version, but I think I can look into it a little and tell you. I'll need to see how the GUI applies a file in the PC, and I suspect that you can do the same in PS.

The other option is to collect the exact points you are using in PS for the curve, which isn't too difficult to do, and then use the coding that Don mentions above to put it in manually. That's how I'd do it anyway, because then you can refine it in QTR easily.


---Michael

 

[andy138]

Thanks guys

I think the hardest part about learning digi-negs is finding one clear, concise path to getting an optimal negative. With it being so new everyone has different ideas of what works best and sorting through all the info available getting tiresome and confusing. Michael-if you and clay plan on offering a workshop on making digi-negs with QTR please let me know, I will absolutely be there.

andy

 

[Kees]

QTR takes .acv files too!

It is easier then you think! You can put you saved Photoshop curve as .acv file in the same folder as the inkdescriptor file and mention the filename or path to the file (if it's in a different location) just after GRAY_CURVE and run the Curves command as usual.

From the QTR Getting Started file:

"GRAY_CURVE being the most flexible, allows you to create a Curve using Photoshop. Using a grayscale file you run the Curves command to create whatever curve you like. Save the Curve as a ".acv" file for re-editing. This is the filename assigned to GRAY_CURVE. "

 

[sanking]

It is easier then you think! You can put you saved Photoshop curve as .acv file in the same folder as the inkdescriptor file and mention the filename or path to the file (if it's in a different location) just after GRAY_CURVE and run the Curves command as usual.

From the QTR Getting Started file:

"GRAY_CURVE being the most flexible, allows you to create a Curve using Photoshop. Using a grayscale file you run the Curves command to create whatever curve you like. Save the Curve as a ".acv" file for re-editing. This is the filename assigned to GRAY_CURVE. "

Does this work for MAC as well as PC?

Sandy King

 

[Michael Mutmansky]

Yes.

But I wouldn't do it that way because then you have to deal with PS to generate a new curve, rather than doing it more directly through the QTR program.


---Michael

 

[Ben Altman]

does anyone know how to embed a PS derived curve into a QTR profile on a mac (no QTRgui)? I've finally gotten a half-way decent curve in PS, but I would like to embed it into a profile in QTR to avoid applying it in photoshop. Is this even beneficial, or should I just save the trouble, apply it in PS? I'd imagine that letting the rip take care of the curve avoids interpolation in PS, and therefore the possibility of smoother tones, but I am probably wrong. Anyway, any info would be a big help

cheers
andy

Yes, it's worth doing it in QTR, because one can create a negative that fits the characteristics of the print material by controlling ink densities, rather than by distorting the levels in the PS file. That leaves you with more room to edit the PS file for density and contrast etc.

It helps to understand what standard QTR profiles or other print engines are designed to do. The more ink a printer puts down, the more the dots tend to overlap. This is dot gain. So to make a good, normal, positive print, the print engine is non-linear, backing off the amount of ink it puts down in the darker tones so they separate and are not muddy.

If you print a standard, optical, Stouffer wedge on Pt/Pd, you'll see that the dark tones on that print are also compressed. So to get a good rendering with a digital negative we want a QTR curve with large tonal separations between the thin steps of the negative and smaller tonal separations between the dense steps. In other words a standard ink profile for positives is doing work that has to undone for digital negatives.

To my mind, this is the weakness of colored-negative approaches that use the printer's built-in print engine - you have to distort the PS file just to undo the complicated work that the printer is doing to make a good positive inkjet print (which is its intended function, after all). Just my take - others may disagree.

I have been building profiles by using a curve for each ink in QTR, removing all the other adjustments that come in a standard profile. This takes some programming in Excel to blend the curves, including an estimate of the total ink so I don't flood the OHP... I won't bore you with the details. The method Michael mentions may make things simpler, or at least help to get one in the ballpark.

Ben

 

[donbga]

I have been building profiles by using a curve for each ink in QTR, removing all the other adjustments that come in a standard profile. This takes some programming in Excel to blend the curves, including an estimate of the total ink so I don't flood the OHP... I won't bore you with the details. Ben

Ben,

Please bore us with the details.

Thanks,

Don BRyant

 

[Ben Altman]

Ben,

Please bore us with the details.

Thanks,

Don BRyant

Unfortunately the computer on which I do this work is in a box as I've just moved. The plus side is that I'm getting to put together a new work-space the way I want it. But I'll try to remember the boring details...

I found that using standard inks with my methods came up grainy, so I switched to LLK, LK and Y inks in all cartridges - have been experimenting with some mixed 50/50 LK/Y also.

The Excel file has several sheets. They are all based on the 21-step wedge (0%, 5%, 10%, 15% etc.) lifted from the QTR calibration routine. I print the individual inks (8 channels on my R1800) on OHP with the QTR calibration routine, measure them with the X-Rite 361 in UV mode, and enter the values to the spreadsheet. Then using some Excel tricks and simple math, the spreadsheet calculates the contribution of each ink to the total ink density, for any input value (between 0 and 100), and plots that contribution on a graph. This plot also shows the sum of the contributions of all the inks, which is the key result. A separate routine adds up the straight input values for all the inks and plots them - this is the total ink volume. On OHP that needs to be pretty close to 100 on each step - any more causes puddles.

Another sheet takes the input values and converts them to a data string that I can cut and paste directly into a QTR text file.

The ink density graph is smoothed by hand trial-and-error: printing negative test-wedges, reading them with the X-Rite and/or printing them on my chosen Pt/Pd mix. (This part needs some more thought, as the trial-and error-part takes too long). The densities predicted by the spreadsheet so far have been close to the values QTR prints, but not right on. I hope to improve this.

The QTR file contains little but the lines that name the inks and then eight lines that look like:
CURVE_X=0,0;5,1.2;10,1.6;15,2.1; ... etc. where X is the name of the ink. (I'm making this up from memory, but I think that's the right syntax).

Hope that's not too confusing. It's a work in progress, but it gave me some very encouraging prints before I had to shut up shop for summer travels and the move.

Ben

 

[Kees]

Does this work for MAC as well as PC?

Sandy King

Hi Sandy, I'm not shure about PC, but it works on a Mac

 

[Michael Mutmansky]

Ben,

I use almost the same approach, with a few modifications. I have adjustments in the curves for effective UV density per ink channel based on tests, a process characterization curve input, an iterative testing step that linearizes based on the Yuill Nielsen formula, and a few other widgets as well, but in the end, it produces a very consistent and grainless result.

It's not rocket science, but it is pretty heavy on math, especialy when considering the polynomial interpolation of the curves, etc., and is much more than most people want to do. However, working this way is without a doubt the most predictable method to date, and it produces a result that it without peer.

I completely agree with you about the shortcomings of using the Epson driver. It's the reason some printers can't be made to work in other methods, but experience has shown them to be 100% suitable for digital negatives using this approach.


---Michael

 

[Ben Altman]

Hi Michael,

Thanks for reassuring me that I'm not the only one thinking this way... You clearly have put more thought into it than I. Perhaps we can compare some notes once I get back into it. I don't yet have an easy way to read the print densities, so I have not added a process curve, but that would clearly speed things up. Also I'm using linear interpolation, as I figure that's going to be pretty close, at least while I get the routines worked out.

Best, Ben

 

[sanking]

It is easier then you think! You can put you saved Photoshop curve as .acv file in the same folder as the inkdescriptor file and mention the filename or path to the file (if it's in a different location) just after GRAY_CURVE and run the Curves command as usual.

From the QTR Getting Started file:

"GRAY_CURVE being the most flexible, allows you to create a Curve using Photoshop. Using a grayscale file you run the Curves command to create whatever curve you like. Save the Curve as a ".acv" file for re-editing. This is the filename assigned to GRAY_CURVE. "

I have a question about this. The background is that several months ago Clay Harmon posted a QTR palladium profile for the Epson 2200. I found that the use of this profile resulted in less banding with this printer than when using the green ink. However, the QTR profile did not give perfectly linear results so I developed an .acv curve which, in combination with the QTR profile, gives perfectly linear results in palladium (and in other processes where I have developed special .acv curves). I figure that the QTR profile is doing most of the heavy lifting so for all practical purposes results are just as good as if I had adjusted the profile.

My question is this. Is there any difference functionally in just loading the .acv in Photoshop, as opposed to doing what your propose, i.e. putting the saved Photoshop curve as .acv file in the same folder as the ink descriptor file? Or is placing the .acv file in the folder with the ink descriptor file merely a convenience that keeps us from having to do so with Adjustments>Curves>Load when printing the file?


Sandy King

 

[Kees]

My question is this. Is there any difference functionally in just loading the .acv in Photoshop, as opposed to doing what your propose, i.e. putting the saved Photoshop curve as .acv file in the same folder as the ink descriptor file? Or is placing the .acv file in the folder with the ink descriptor file merely a convenience that keeps us from having to do so with Adjustments>Curves>Load when printing the file?


Sandy King

Hi Sandy,

Apart from being convenient it might be better to leave the heavy lifting to the driver instead of altering the file with a curve. But one of the photoshop and QTR geeks should join in to explain the underlying math. And maybe tell us that there's no difference at all. From the the viewpoint of keeping things simple I would say that making a process (and user) specific QTR profile, that does all process related corrections, is the best choice. In a weblog post I explained a curve merging method I found recently. You could use this method for merging your .acv curve and clay's curve (the data allready in the QTR profile) to one curve and use that in the driver. If you also invert the curve (i.e. merge an inversion curve in this curve) you can send your on screen positive to the printer and get a correctly curved negative out of the printer. Be carefull though, when merging curves, don't mix 'apply to positive' and 'apply to negative' curves. QTR takes curves that have to be applied after inversion to a negative. Other curves (chartthrob for example) have to be flipped first.

kees

 

[Ben Altman]

My question is this. Is there any difference functionally in just loading the .acv in Photoshop, as opposed to doing what your propose, i.e. putting the saved Photoshop curve as .acv file in the same folder as the ink descriptor file? Or is placing the .acv file in the folder with the ink descriptor file merely a convenience that keeps us from having to do so with Adjustments>Curves>Load when printing the file?


Sandy King

I'd say if the QTR curve is close, there is no practical difference. If the .acv correction needs to be large or unsmooth, then editing the QTR profile is better. The key reason for using a RIP like QTR is so that you don't end up fighting the paper profile and the manufacturer's printer driver (which are designed for positives on paper and are quite different from what is needed for negs on OHP) with the Photoshop curve you apply to your image-data. For negs, a good QTR profile should remove the need for extra image-data manipulation that can lead to problems.
The image-data from a negative scan has usually been pretty heavily worked already (either by the scanner software or Photoshop) just to get it to fill the histogram, so the fewer additional changes in PS the better.

Kees - thanks for finding and sharing the curve-blending workaround. I know I'm going to be using that one!

Ben

 

[Kees]

Kees - thanks for finding and sharing the curve-blending workaround. I know I'm going to be using that one!

Ben

Yes, and it's also very handy to 'flip' a 'before' to an 'after' curve. Just sandwich your curve between two 'invert' adjustmentlayers and apply this package to the raw file, flatten and save as .amp file. And there's your flipped curve.

-k