# [argyllcms] Re: Dumb question

• From: Fabrizio Levati <fabrizio.levati@xxxxxxxxxxxxxx>
• To: <argyllcms@xxxxxxxxxxxxx>
• Date: Wed, 14 Jul 2010 18:10:28 +0200

```Graeme, Klaus,
Putting together all your useful tips, I have been able to edit the profile
the way I needed. It works as needed and my CLUT now looks like this:

CLUT table:
0  0  0  0: 0.0000000000 0.0000000000 0.0000000000 0.0000000000
0  0  0  1: 0.1006637675 0.0711528191 0.0504158083 0.0000000000
0  0  0  2: 0.1814755474 0.1252765698 0.1049973297 0.0000000000
0  0  0  3: 0.2645456626 0.2058747234 0.1734340429 0.0000152590
0  0  0  4: 0.3671168078 0.2923018235 0.2497901884 0.0004882887
0  0  0  5: 0.4479743648 0.3515983825 0.3103379873 0.0965133135
0  0  0  6: 0.5006179904 0.4135042344 0.3550011444 0.2178835737
0  0  0  7: 0.5318379492 0.4592355230 0.3911345083 0.3419546807
0  0  0  8: 0.0000000000 0.0000000000 0.0000000000 1.0000000000

Having to deal with number encodings and offsets, I had to go through the
ICC specifications (too bad I didn't before...) and among others I found
something I am not able to deal with.

In the lut16type description I find:

The domain of each of these tables is defined to consist of all real numbers
between 0.0 and 65535.0, inclusive. The first entry is located at 0.0, the
last entry at 65535.0, and intermediate entries are uniformly spaced using
an increment of 65535.0/(M-1). For the input and output tables, M is the
number of entries in the table. For the CLUT, M is the number of grid points
along each dimension. Note that since the increment of 65535.0/(M-1) is not
necessarily an integer, the domain is specified to be over the real numbers
rather than restricting it to the integers only.

Since 0 0 0 8 corresponds to 100% black channel I simply assumed that all
the steps in between must have been equally spaced by 12.5% but that doesn't
seem to be the case.
Running a 50% black patch through the profile I obtain values (42C 33M 29Y
6K) completely different from what I expected (36.7C 29.2M 25Y 0K) and I am
not able to understand why.

Would you mind sharing some of your knowledge about the mechanics of ICC
profiles?

Fabrizio

```