Thursday, October 03, 2013

Making a Granger Rainbow for Debugging Color Management

The Granger Rainbow is an image which presents all saturated rgb or lab colors, possible or impossible, in a smooth palette.  When printed such an image allows one to assess both the gamut of a device and the smoothness of a rendering, from one image.

To create a Granger Rainbow in Photoshop, in Lab space, we will cross two gradients. The gradient tool can be selected by typing "g". Starting from a filled white canvas, we first deploy a  "Spectrum" gradient preset that cycles through all hues from left to right.  



Then we apply a "black and white" gradient to the L channel only, from bottom to top, to get the light tones in the sky where we expect them. Tip: Hold the shift key down while drawing the gradients in order to force a square orientation.



Edmund Ronald


Sunday, July 07, 2013

Iliah Borg's MakeInputICC: A camera profiler wrapper for the Argyll engine

Nutshell summary: Iliah Borg has assimilated Argyll into a user-friendly wrapper for making input profiles; this Borged wrapper assists quality assessment by graphing the profile's curves for each of r, g and b.

Download links at the bottom of this post!

Argyll is Graeme Gill's highly respected open source profiling engine. As we all know, open source like Linux is often quite useful, but a bit —errm, awkward. Well, in the case of Argyll, very useful and very awkward. Camera profiling is already sufficiently frustrating without the need for long command-line incantations.

Sometimes, geeks do realize the failings of other geeks and write graphical interfaces. And we regularly see wrappers emerge which combine the know-how of a specialist's use of profiling workflows with Argyll's engine. Such is the case of MakeInputICC which wraps Argyll's nightmarish command line options into a menu system that a mere Ph.D might understand, while providing some feedback on profile quality.



One question which comes to mind of course is: "But where does the input data come from"? Traditionally, input profilers are interpolation engines which need two input files: A reference file for the target, which is provided here in spectral form for some industry standards; and a very clean test shot corresponding of said target. 

However, MakeInputICC expects to ingest a pre-digested CGATS file of rgb values extracted from the target. The necessary file can be supplied by Alex Tutubalin and Iliah Borg's Rawdigger which is not open source, or by any other method known to ColorGeeks;  Danny Pascale's PatchTool is one very convenient commercial utility which can help do this job, generating a CGATS file after reading in a target image Tiff generated by one's favorite Raw converter. 

I am not going to comment on profile quality, as all the authors of camera profiling software are by now personal acquaintances. However the beginner should be warned that camera profiling is something of a black art and is highly dependent on the abilities of the photographer.

But even if camera profiling is a black art, it remains a necessary evil: ICC profiles are essential for the use of any of the open-source dcraw descendants. They are also a requisite for any serious art reproduction work with an ICC-compliant commercial converter such as Capture One or Iridient's Raw Developer. In fact some color geeks consider it very unfortunate that Adobe's ACR is not compatible with ICC profiles.

Among profiling software, MakeInputICC has the advantage of creating license-free profiles; it  should find an audience among specialist photographers who may have been frightened of Argyll's steep learning curve. Said exigent experts will no doubt run their own tests in their own lighting and working conditions, with their own Raw converters, and compare the output of Argyll to those profiles created by boutique products such as Basicolor Input. 

I like Iliah's software because it is simple to use and provides instant feedback on whether a profile is clean or not.  On the other hand, the tabular output is less useful to humans than a target image would be; it seems to me that  the author might as well integrate a patch reader and target display into his software. Or maybe some other geek will revise this geek's work too — accretion seems to be the constructive process by which the open source bazaar extends its labyrinthine cathedral.

Saturday, May 11, 2013

Debugging Profiled Inkjet Printing Issues.

Printing should just work, but often it doesn't, and you have doubts about the colors. I'm going to show you a quick way to figure out whether you have a hardware problem, or whether software issues are corrupting your profiled printing workflow. The key step is to get some original vendor paper and make a print with the vendor workflow aka. "Printer Manages Colors". This print then acts as a reference baseline for debugging, and finding issues becomes a matter of comparison, not absolute judgement.

The debugging process I detail below leads you through three steps. First you get some of the printer manufacturer's paper, and make a print with vendor settings. Then you make a print with the manufacturer's own profile; finally you make your own profile and print for the same vendor paper. The condensed set of steps, is followed by some explanations about vendor color and canned profiles.

Diagnosing Profiled Printing Issues: The Procedure


  1. Set up the vendor workflow with "Printer Manages Colors" and make a nice colorful test print on vendor paper, making sure to choose the right paper type in the print driver dialog. If the print made with the vendor workflow looks ok, then your hardware is working well. If the print is  bad then  you might have a hardware issue eg. a clog, or you might have inappropriately chosen, fiddled with or otherwise corrupted the paper and color settings, so you need to clean the printheads and need to reset the default values for the print system and then make another test print.
  2. If the print from (1) looks ok, then set up your usual profiled workflow, and make a test print using the same vendor paper, the paper settings validated in (1), and the vendor's canned profile. If this print does not match the  print made in (1) then your profiled printing workflow is compromised by inappropriate settings or some other software issue —vendor profiles should always match vendor color.
  3. If the print made in (2) with the vendor's profile looks ok, then your hardware and profiled printing workflow is good. Now it's time to focus on your own profiling technique. Make a profile of the vendor paper yourself to compare with the vendor's profile. To print the profiling testchart the safest thing to do now is to make use of Adobe's ACPU utility, using the same paper settings as in (2), and then scan the print with your favorite spectro and software to create a new profile. Now make a new print with the profiled workflow. If this matches the vendor color print in (1) you are done, just go ahead and profile your own media. However if your own profiled print on vendor paper does not roughly match the print in (1) then you probably have an issue with the profiling process itself —hardware or software— and should call the profile tool vendor for support.  

Some explanations. 

Making profiles and printing with them is a fussy business, and many of us would prefer that "it just works". The good news is that Epson, Canon etc realize users want an easy solution, and they provide one: If you have  decent modern printer you will obtain decent  color from any app —and Photoshop —by just selecting "Printer Manages Colors" in the print dialog, and specifying the paper you have loaded. This is sometimes called Vendor Color mode. 

The nice thing about Vendor Color is that it is almost foolproof. To make it work correctly you just need to reset the printing system, make sure you have the right paper type selected, and make sure that there are no leftover color bias settings hidden away in your print dialog.

Of course there is a price for the simplicity of this workflow: you must use the manufacturer's paper, and for neutrals to stay neutral and color prints to be really spot on, you should have one of the more expensive printers which have several gray inks, and whose manufacturing tolerance is so good that it closely matches the canned profile. Incidentally, the printer driver's Black and White Mode works great on these models. 

Vendor mode is also a great debugging tool, because on their own papers Epson delivers very closely matched results by using vendor mode and by using their own canned profiles.  So to see whether your color-management printing setup is working, you just need to locate the canned profile for your paper and print with it using color management - this is called printing in Application Color Mode. 

And how do you find the canned profiles for your printer? Well, on Macs at least, the profiles should  already be there in your color management dialogs. Otherwise, if you open the Colorsync utility, and select the printer, the installed printer's profile paths will be displayed for you. 

Printing an image in Vendor mode should show you whether your printer is set up correctly and working ok. Printing in Application mode and comparing with the vendor mode should tell you whether color management is working ok. Your last problem if you do wish to profile your printer is to print a testchart with no color management - to do this it might be best to download Adobe's dedicated ACPU utility which excels at printing charts. 

Edmund

Tuesday, May 07, 2013

Sanity checking the i1Pro2 vs iSIS


I own this very nice 18th century antique card-table, and on the green baize sit the little i1Pro2 and an iSIS XL. The iSIS XL is Xrite's top of the line desktop scanner which can read 5000 patch single-page A3 charts. The i1Pro2 is the recognized swiss army knife of instruments, a universal handheld spectrophotometer which can calibrate screens and projectors, profile printers, and take ambient light measurements. The two instruments have been getting friendly and comparing measurements.

Checking instruments against each other is something ever color geek does occasionally to detect hardware issues. I also wanted a sanity check so as to know whether profiles which I make in the field or which others make with the i1Pro2 will match those I make myself at home with the larger spectro. I will be running some more rudimentary checks with the DTP70 and the Barbieri LFP within the next few weeks. 





By using the compare feature in the i1Profiler software we can read in data from two charts and compute some comparative statistics. Here, on the i1Profiler 400 patch RGB chart, we have a max deltaE of 2 and an average of 0.56, which is quite good, considering that the two charts didn't have a full dry-down, and that my Isis is an old warhorse which has seen better days, and should probably be recalibrated.

  


As you can see the iPro2 handheld device does not have any tendency to operator-induced read errors, which were sometimes an issue with the old i1Pro. Also, using the new reading ruler to measure a chart is lengthy but it is not at all stressful. This is big progress compared to the old i1 which required quite a bit more patience, practice and elbow-grease. I've had some very pleasant experiences of late with the i1Pro when calibrating screens and am starting to appreciate this upgraded instrument and the new software. 

Edmund Ronald