Tools to use Isis (and other spectros) to verify a printer profile

[Doug Gray]

This is an image and set of tools to check the print side (the normal printing part of a profile) of the 3D LUTs known as BtoA1 tables. Many tools that print dE values just compare target patch colors with the reverse lookup tables (AtoB1) but these are not the ones used to print an image.

The attached zip file contains 5 files. The only essential file is a 16 bit tif in Adobe RGB containing the colorchecker patches as well as a set of neutral colors from L*=4 to L*=90 in steps of 2. This is also attached separately to this post for clarity.

'ColorChecker Plus Neutrals.txt'
CGATs file of LAB values. The Colorchecker(r)  ones, which are the first 24 colors, are derived from BableColor's CC spectrum which was run through Patchtool to create Lab values. The remaining colors are the sequential, neutral LAB values described above.

'ColorChecker Plus Neutrals 10x10 Cnt68 DPI720 12-17-17  9-32AM.tif'
This is a 16 bit tif file in Adobe RGB(1998). Note that high accuracy Lab values in tif files can be read in Photoshop's info panel if you set the resolution to 32 bit and select LAB as the display format.

INSTRUCTIONS FOR PRINTING:
Please note that though it looks like a target file it should not be printed with ACPU or other target file printing utilities. The file should be printed by selecting the paper and its profile and using Photoshop Manages Color. It must be printed using Absolute Colorimetric Intent.

'ColorChecker Plus Neutrals 10x10 68 12-17-17  9-32AM 9500 283a_M2.txt'
This is the output of an Isis scan with a printer/paper profile made using the smallest possible target, the TC2.83 target provided by I1Profiler. It has an average DeltaE2000 of 1.9 and many of the patches are much further off.

'ColorChecker Plus Neutrals 10x10 68 12-17-17  9-32AM 9500 957a_M2.txt'
This is the output of an Isis scan with a printer/paper profile made using the default, single letter page, Isis target with 957 patches It has an average DeltaE2000 of .7 and a few patches are a bit over 1.0.

'ColorChecker Plus Neutrals 10x10 68 12-17-17  9-32AM.txf'
This is the target file for those with an Isis. It will set the patch size (10mm by 10mm) and other defaults. Do not use this top print! It's for measurement only.

These patches are made large enough to be easily read in spot mode with an I1Pro but they are formatted to automatically scan with an Isis for people with that instrument.

[GWGill]

This is an image and set of tools to check the print side (the normal printing part of a profile) of the 3D LUTs known as BtoA1 tables. Many tools that print dE values just compare target patch colors with the reverse lookup tables (AtoB1) but these are not the ones used to print an image.

Hmm. I've always referred to the A2B tables as the forward tables, since they reflect the natural behaviour of an output device - i.e. you send device values to the device, and you get visual color out of it (CIE), and the A2B table directly mimics this measured behaviour. In contrast, I refer to B2A tables as backward/reverse/inverse tables, since they invert the actual behaviour of the device, and the construction of the B2A table involves inverting the A2B table.

The B2A/inverse table is the one used in printing, because you are trying to null out the natural behaviour of the device so that you can specify the desired color in terms of device independent values.

[Doug Gray]

Hmm. I've always referred to the A2B tables as the forward tables, since they reflect the natural behaviour of an output device - i.e. you send device values to the device, and you get visual color out of it (CIE), and the A2B table directly mimics this measured behaviour. In contrast, I refer to B2A tables as backward/reverse/inverse tables, since they invert the actual behaviour of the device, and the construction of the B2A table involves inverting the A2B table.

Well, A2B is the more accurate table as well since the B2A table is derived from it. But "forward" is a term that depends on one's POV. For most printers, I suspect the tables  of interest are the B2A ones. Those determine the RGB values that will be sent to the native driver. Most would think of them as the "forward" tables as in "forward" to the printer. It's probably best to just say that "A" represents the device and "B" the profile connection space (or LAB).

The B2A/inverse table is the one used in printing, because you are trying to null out the natural behaviour of the device so that you can specify the desired color in terms of device independent values.

Yup.