Relinearize Quad Curves

In the post "How to Edit .Quad Files without Burning Down the House" I said that re-linearizing the QTR .quad files used in printing was not possible and you shouldn't try it at home. I then set out to prove myself wrong.

Lab_L measurements from an Eboni 6 profile made with the QTR curves creation tools.

At first this started off as just trying to map the quad values made with the QTR curve creation workflow to a different shape—take the straight line and put a slight curve in the shadow end of it. It seemed like it would be easier to push around linearized quad values, but when I got into it deeper I found it really didn’t matter if they were linear to begin with or not.

Piezography Curves, and the occasional need to re-linearize them

There is no arguing that what you are getting with Piezography is a turnkey product that is much better than (or at least equal to, with a whole lot of work) what anyone could do themselves with the same printers and inks. The only problem is that when it doesn’t work there is really no way to know what exactly is wrong without flushing ink or buying new equipment. Then there are the inevitable support emails, desperate forum postings, and then maybe a printer ends up in the bin and another user opts for HP or Canon the next time around.

Lab_L Measurements from an existing K6 profile for the Epson 1430 that indicates a problem with the printer.

I have had three different printers and three different ink sets that have not been able to make a linear print—they either printed much darker than intended or made prints with distinct bands at different parts of the gray scale (and this is something I have tested every possible way). The one with banding was a brand new Epson 1430, and the others were an old 3800, which is still going strong, and a 4800 that died a few years ago.

At first I thought my problems were just a matter of user error, so I double checked my methods with a different ink set (with the same shade inks) and had the same problem. Buying a new printer was not an option, so it was a matter of sorting out how to get the thing to work for me. Doing all that testing and refining the QTR curve creation method started to inform parts of the upcoming QuadToneRIP book, but it also lead me to “hacking” the quad file and really understanding how some of the things under the hood work. I’m glad I had a couple of wonky printers because otherwise there would be a whole lot I wouldn’t know now.

I have spoken to other people who have absolutely no problems with their printers loaded with Cone inks, but there are others who have similar problems to what I am describing. From what I have seen and experienced, a printer that is only slightly out of spec can cause problems with the Piezography curves. That isn’t meant to diminish Jon’s work or his product—I still think it is the best out there, but that the product needs a printer that is in perfect working order or else the shape of the curves can introduce problems. The Piezography style curves might even be less prone to error than QTR style curves, but the QTR problems can be solved by the user.

About my method for re-linearizing the Quad Curves

I have previously written about having a way of somewhat automatically filling in the ink limit and overlaps for partitioning k6/k7 inks using the standard QTR Curve Creation tools and workflow (as well as some additional gray gamma gray overlap settings). This is information that can be found and pieced together from various sources online. I just did it in a more logical, repeatable, and refined manner, and this is what will make up some of the upcoming book. The original Quad curves created with this workflow produce a smooth, linear density distribution, but with my Quad Curve relinearization method I can introduce a slight compression in the shadows and a different shaped gray curve the new quad curves are mapped to, but that does not block up like using the color managed workflow for “re-linearizing”. This is done by reading in the original printed densities, defining target densities with either linear or slightly compressed shadows, automatically creating a correction curve, and then mapping the original quad values to corrected ones that will print the target densities. All of this is done outside of the QTR Curve Creation tools so the "lab values out of order" or "not constantly increasing" linearization errors are never a problem. 

Since the my method works directly from an existing .quad file it means that I can also relinearize existing Piezography curves for new papers, aging or out of spec printers, or to map existing profiles to a different shaped gray curve. The new linearized quad values are then pasted into a text file template with the correct header information for the printer and number of inks being used and is then saved with the .quad extension. That new quad file can then be placed in the correct folder for the quad printer (depending on the mac/pc requirements) and installed like any other .quad profile.

It is also possible for me to create custom Quad Profiles from the existing Piezography master curves (although the higher limit of ~60% in the K channel is a problem and can cause reversals from 96%-100%.) I have solved this issue by developing a way to parse Piezography master curves to 21 control points so ACV curves can be created and edited in Photoshop and then assigned to each channel in a QTR ink descriptor file (doing this with the ink_curve= input bypasses the standard QTR gray ink partitioning functions and creates new master quad curves based on each of the ACV curves). You can then control the amount of ink for each channel with the ink limit settings in the ink descriptor file. The 51-step grayscale is printed, measured, and then the original quad file and measurement file are imported and automatically linearized to the ideal densities. 

My Quad File Relinearization Service

I have this relinearization method more or less perfected with a spreadsheet template that I am using as a prototype and now I am working on developing it into a possible standalone application. Until that is ready, I am offering a re-linearization service for people who can email or upload the 51x3 measurement file and the .quad file used to print the 51 step target (the single 51-step target is measured three times to average out measurement errors). You will receive a graph of the original measurements and a re-linearized quad file.

How is this different than what INk Jet Mall offers with custom Piezography Profiles?

I've chosen to offer this service as an affordable alternative for people who can print and measure their own 51-step target, and is not limited to Piezography curves. Since this works directly with existing .quad files I can re-linearize any QTR Curves based on the 51-step measurement file. This can work with UC K3 profiles, as well as Eboni-6 profiles made with the QTR curve creation tools**. 

I am not claiming that profiles created with my method are as good as what you would receive with a custom profile from Ink Jet Mall. My method doesn't require printing and mailing a full 256 step grayscale file that their PiezoProfiler requires, and while my correction curve and quad value interpolation method is very good and extensively tested, it does not work with the same kind of direct lookup table for determining the final quad values. This is meant as a way to affordably test new papers, correct for printer problems, or just map the existing linear profile to one with a slightly richer shadow curve that doesn't create the blocked up shadows that occur when printing with an ICC profile and Black Point Compensation. 

Why a 51x3 measurement file? 

I've tested this with the 21x4 random target included with QuadToneRIP and found that 21 data points is not enough to correct for the small bumps that often occur when using so many overlapping inks. I've determined that 51 control points allows enough precision for the functions to work accurately while not being too arduous to print and measure. However, there are always small errors in measurements of the same patch and/or target and I've found that averaging three different measurements of the same printed target is sufficient to smooth out any measurement errors. I've found that it is also better to measure the same target three times than to print three targets and measure each target once.

You can download the 51x3 reference file for X-Rite's ProfileMaker5 MeasureTool and measure the 51-step target included with QuadToneRIP. For those with Mac OSX Lion and later, I've created an i1 Profiler workflow and updated 51-step target that works with the i1Pro photospectrometer. You can download these at the link below. Those with the Spyder Print photospectrometer can print the current 51-step target and use the reference file included with QTR to measure and save three separate measurement files.

SpectroWhat?

Alternatively, for those without a way to measure the printed target, for an small additional charge you can upload your quad curves and then mail in the target for me to measure and then receive back a relinearized set of QTR curves. The page to upload you quad curves and instructions for mailing the target can be found here: http://www.bwmastery.com/quadlin-mail-target

INSTRUCTIONS

SPYDERPRINT OR X-RITE PROFILEMAKER5 MEASURETOOL

  1. Print the Step-51-gray.tif found in the EyeOne folder in the QuadToneRIP application folder using the QTR Curves you want to relinearize.
  2. Dry the print overnight or with a hairdryer for 3-5 minutes
  3. Download the special reference file to measure the QTR-51-step grayscale three times. 
  4. Measure the 51-step target three times and save the measurement file with the same name as the quad curve and include 51x3 at the end of the file name before the .txt extension.
  5. Upload the Original Quad Curve File and the Measurement File of the 51-step target using the form below.

I1PROFILER

  1. Download and extract the i1Profiler 51x3 workflow folder.
  2. Follow the instructions here for printing and measuring the updated 51-step target for i1 Profiler. 
  3. Upload the Original Quad Curve File and the Measurement File of the 51-step target using the form below.

CAN'T FIND YOUR .QUAD FILES?

These QuadToneRIP curves are buried in the system folders and can be found by navigating to the file path below:

  • Mac: Macintosh HD/Library/Printer/QTR/quadtone/
  • Windows: c:\Program Files\QuadToneRIP\QuadTone
  • Then choose the folder for you printer model and ink set
  • Then you can drag the quad file you want to relinearize to the field in the upload form below

** I have not yet tested this with digital negatives for alternative processes, so for now this service is limited to positive inkjet prints, but I hope to be able to do the required testing for digital negatives within the next few months.