An in-depth ICC color profile

0
893

Throughout this article we will delve a little deeper into understanding and interacting with the parameters that we can obtain from an ICC color profile once it is inspected, with programs such as ColorSync, for example.

To inspect an ICC profile on Mac OS X it is enough to open it from ColorSync. However, as we will observe later, some data from it are not displayed particularly clearly, and it is better to resort to the command line tool of ArgyllCMS: iccdump.

Profiles are organized by tags, with associated values and descriptions. The most interesting are:

Header

sRGB CabeceraThe header of the profile contains interesting descriptive data about the construction and operation of the profile. Within this tag we can observe several parameters:

Device Class (Class):

It indicates the type of profile depending on how it will be used, the values will be:

  • Input: For input profiles, which are the ones we usually create with RoughProfiler
  • Output: For profiles intended for the Soft Proofing technique
  • Display: They can be profiles intended for use on monitors, as well as profiles intended for workspaces and viewing through electronic media.

Color Space (Space):

It is the color space in which our profile works, possible values for this parameter can be: RGB, CMYK, Gray or LAB

Conn. Space (PCS):

This is the Profile Connection Space, or connection space from which we will link the two color spaces with which the profile will work. Essentially this space will be XYZ or LAB

Dev. Attrbts (Device Attributes):

This is one of the parameters that is not very readable at first since we will find a long string of zeros with some value at its end. Although there is a correspondence between these patterns and their final values, it is best to inspect our profile with iccdump so we can observe attributes with the following possible combinations:

Dev. Attrbts = Reflective, Glossy, Positive, BlackAndWhite
Dev. Attrbts = Transparency, Glossy, Positive, Color
Dev. Attrbts = Reflective, Matte, Positive, Color
Dev. Attrbts = Reflective, Glossy, Negative, Color
Dev. Attrbts = Reflective, Glossy, Positive, BlackAndWhite

These attributes can be controlled from RoughProfiler from the Attributes field.

Rndrng Intnt (Rendering Intent):

This tag indicates the type of intent used in the profile when making equivalences between spaces. The types of intents or Render Intents are:

  • Perceptual
  • Relative Colorimetric
  • Saturation
  • Absolute Colorimetric

intentJust like the previous tag, from our ColorSync when it comes to Relative or Absolute only the term Colorimetric is shown, so it will be necessary to inspect with iccdump to be clear about which type of intent it is.

Illuminant (PCS Illumination):

This tag indicates the type of illuminant, or illumination used in the connection space or PCS. Whenever a transformation matrix is generated it is necessary to provide the illuminant under which it works. Usually, and by ICC convention, the PCS illuminant will be D50 or what is the same in XYZ coordinates: 0.964203, 1.000000, 0.824905

If we observe other coordinates that are not those described, then we must suspect that the PCS illuminant is not the standard D50.

White Point Tristimulus (wtpt)

triestimulos White D50When we talk about tristimulus, we refer to the three stimulus quantities provided by the XYZ primaries of the CIE XYZ color space. Therefore, the tristimulus of the white point will be the XYZ coordinates where the white point is located. The location of this point varies depending on the illuminant under which we observe our color chart. The concept of white point is closely linked to the perception of color temperature.

The wtpt is calculated from the illumination under which we record our color chart. If we photograph or correct our color load for a D50, the coordinates shown here will match those indicated in the previous point. However, often these values will correspond to the color temperature under which we exposed our chart.

whitePointCareIf despite having made a color chart with a certain dominance, or we have not managed to calibrate the dominances correctly, if we use the RoughProfiler parameter: White Point Care, we will disregard the possible dominances on our white patch, and create a profile with the standard D50.

RoughProfiler White PointConversely, from the White Point slider in Rough Profiler, we can introduce modifications to this point, which help us warm up or cool down the profile results.

Black Point Tristimulus (bkpt)

The black point tristimulus is, similarly to the white point, the location of black for a specific illuminant from the point of view of a particular observer.

Red, Green, and Blue Colorant Tristimulus (rXYZ,gXYZ,bXYZ)

These tags only exist if we are dealing with a matrix-type profile, which as we said in another article, use a 3×3 transformation matrix to perform the transformation calculations.

The values we can observe at this point vary for each profile and are deduced from the process of creating our profile from a color chart or spectral measurements made on other media such as a monitor.

For the example I will use the standard color profile: sRGB IEC61966-2.1 since the values of its matrix are defined by the International Color Consortium.

The values of this matrix for this profile should be the following according to the ICC:

R: X=0.4360, Y=0.2225, Z=0.0139;
G: X=0.3851, Y=0.7169, Z=0.09710;
B: X=0.1431, Y=0.0606, Z=0.7139;

With these values we can obtain our own 3×3 transformation matrix and perform conversion calculations from the RGB space to XYZ:

Our matrix would therefore be:

  
     [0.436    0.2225    0.0139]
  M= [0.3851    0.7169    0.0971]
     [0.1431    0.0606    0.7139]

Our operation:

 [RGB] = M [XYZ]

If for example we take the RGB values of white: 255,255,255, which we must scale previously:

 
  R = 255/255  = 1
  G = 255/255  = 1
  B = 255/255  = 1

If for example we want to use other values that are not white it would be: R= 6/255, G= 10/255, etc.,…

The mathematical calculations on our matrix would therefore be:

 
  x =  R*0.436 + R*0.3851 + R*0.1431 = 0.9642
  Y =  G*0.2225 + G*0.7169 + G*0.0606 = 1
  z =  B*0.0139 + B*0.0971 + G*0.7139 = 0.8249

If we notice we have obtained the coordinates of the D50 illuminant used in the White Tristimulus and in the PCS. However, this is a coincidence, since being a standard profile the White is located exactly at those coordinates, in other profiles, we will obtain variations in such situation.

Red, Green, and Blue Response Curves (rTRC, gTRC, bTRC)

curva gamma matrixJust like matrix-based profiles, we can have curves or not in a profile, although in general we will almost always have the curve type tags, if for example these are linear or of the gamma= 1 type, this indicates that the curves do not impose modifications on the transformation process.

As we said in the article “types of ICC profiles” there are several types of curves:

On one hand we can have the same curve for the three colorants, or we can have one curve per colorant. On the other hand, we can have two essential types of curves:

  • Gamma type, which is the characteristic logarithmic curve with a smooth start to end in a steep slope.
  • Shaper, molding or adaptive type, which we said is an arbitrary curve, much more precise than the previous one since it allows adapting to particular regions of our spectrum.

In profiles for grayscale we can find the KTRC tag which is a gamma type curve for tone response.

Tables

perfil luts matriz adaptacion cromaticaIn the previous article “Types of ICC profiles” I already addressed a bit the table structure, also known as LUT or Lookup Table. As a summary, indicate that LUTs are structures that solve the use of very complex calculation routines by using indices, greatly simplifying certain computational calculations.

In our profiles, we can find several types of tables: direct tables or series that start with the letter A, and inverse tables that start with the letter B. As can be deduced from the name, some perform calculations in one direction and the others in the opposite, understood therefore as “A to B” and “B to A” where the English article “to” is replaced by its phonetic analogue “2″.
A profile can have one or several tables, depending on the intents they implement for their operation. Do not confuse the intent used to build the profile with the intent used for the conversion between LUT spaces.

There are as many tables as intents so that:

  • A2B0 / B2A0 – Perceptual
  • A2B1 / B2A1 – Relative Colorimetric
  • A2B2 / B2A2 – Saturation
  • A2B3 / B2A3 – Absolute Colorimetric

Once inside each table, various controls may be offered to evaluate the behavior of that table, it should be noted that we should not confuse the Matrix, which usually appears inside the LUTs with the matrices indicated above, in this case it is a chromatic adaptation matrix, used to fine-tune the performance of the calculations. There are three essential types of chromatic adaptation matrices: the so-called XYZ Scaling (characterized by a series of ones on its diagonal), the Bradford and the Von Kries. Chromatic adaptation matrices are used to compensate for spectrum variations across different light sources.

Others

The rest of the tags relate to text tags where text strings can be stored to help describe our prof