Gamut comparisons with the HLC Colour Atlas

[:en]

Gamut comparisons with the HLC Colour Atlas - which colours can be produced in which process?

The free PDF version of the HLC Colour Atlas shows the gamuts of different output situations. The CIELAB colour space is systematically run through in H/L/C with a constant step size. This method is exact, practical and meaningful. This article shows a few interesting facts about gamut comparison, which can be explored much further using the pdf version.

CIELAB coordinate space and CIELAB colour body

With the freieFarbe HLC Colour Atlas XL, the CIELAB colour space is divided into its polar coordinates HLC. If H (interval 0..360), L (interval 0..100) and C (interval 0..125) are divided in steps of 5, the result is 32851 coordinates.

CIELAB HLC Koordinatenraum
View from above of the CIELAB HLC coordinate space used in the HLC Colour Atlas.

However, approx. 40% of this coordinate space are not colour tones! The range of real existing colour tones, the CIELAB colour body, is limited by the so-called optimal or rectangular colours. These are spectral curves that have a reflection of 100% in one wavelength interval and 0% in all others. In the article "Limits of the CIELAB colour space", the facts of the case are explained in more detail.

CIELAB Rechteckfarben
The optimum colours (here: 10nm intervals) form the outer shell of the CIELAB colour space. The CIELAB colour space is not only a part of the coordinate space, but it also extends beyond it in some areas.
CIELAB Farbraum
CIELAB colour space in finer resolution (1nm optimal colours) with additional division into 20 brightness levels (L=5..100). (Image: Prof. Bernhard Hill, Aachen).

The CIELAB colour body is very irregularly shaped. In the green range it reaches a maximum a-value of approx. -160, which goes beyond the above-mentioned usual a-interval in software products, so this colour range is not adjustable in the usual products (Adobe CC and others). However, this circumstance is usually not serious, since the highly intensive colour ranges in question, close to the rainbow colours, cannot be represented in almost any output process.

Gamuts of important output variants

"Gamuts" answer the question: what colours can be represented in a particular output situation - where are the limits? It is clear that e.g. in printing on newsprint far less vivid colour tones can be produced than on a good monitor. The gamut creation of the HLC Colour Atlas was done for most output variants in Photoshop: from a bitmap with the entire CIELAB coordinate space, the non-gamut areas were removed via "Select colour range, colours outside colour range". This is quite simple and can be understood by anyone with the Photoshop file mentioned below, but the underlying gamut algorithm remains unknown. In the Epson proof print version, which is also available in printed form, the gamut was determined using the colour management tool ZePrA from Colorlogic, with which a finer adjustment of parameters is possible.

The HLC Colour Atlas compares the following output methods:

CIELAB-Farbraum und sRGB-Gamut
CIELAB colour space and sRGB gamut contained therein (Image: RWTH-Aachen University, Colour Science Research Group).
  • sRGB: Standard RGB according to HP/Microsoft: the international standard IEC 61966-2-1 is also adopted by the ICC and W3C and has therefore become established in the RGB sector.
  • ISO coatedV2: the profile "ISO Coated v2 (ECI) 300%" is published by the European Colour Initiative (ECI) to the characterisation data FOGRA39 developed by FOGRA. It is advantageous compared to the Adobe profile for FOGRA39 (see below).
  • FOGRA39: the profile "Coated FOGRA39 (ISO 12647-2:2004)" is supplied by Adobe in the CreativeSuite/Cloud and is therefore common.
  • FOGRA51: the "PSO Coated v3 (FOGRA51)" profile is the offset printing standard for coated papers as proposed by FOGRA and ECI since 2016. It takes better account of the usual optical brighteners in papers.
  • FOGRA52: the profile "PSO Uncoated v3 (FOGRA52)" is the offset printing standard for uncoated papers as proposed by FOGRA and ECI since 2016.

Scope of the colour spaces

The table shows the calculated share of the HLC coordinates contained in the respective gamut in the total CIELAB coordinate space. The CIELAB colour space has an estimated share of approx. 60% in the CIELAB coordinate space, so that one has to multiply the values by 1.67 to obtain the share in the CIELAB colour space.

Variant Relatively colourimetric Absolutely colourimetric
sRGB 39,1% 39,1%
Coated FOGRA39 (ISO 12647-2:2004) 34,4% 26,9%
ISO coatedV2 37,3% 27,2%
Epson_SCP7000_720x1440dpi10c_GMGsemimatte250_V1_Photoshop 44,2% 40,4%
PSOcoated_FOGRA51 39,0% 27,7%
PSOuncoated_v3_FOGRA52 37,2% 17,0%
Which orange shades can be produced in sRGB (left) and in offset printing on uncoated papers (right)?

Both in proof printing and in sRGB, approx. 64% of the CIELAB colour space, i.e. almost two thirds of all colours, can be produced. Nevertheless, there are big differences in both gamuts:

The Epson proof print (left) can represent the colour ranges more or less consistently well, but sRGB cannot.

In offset printing on coated papers, about another third of the colour tones cannot be reproduced cleanly. Offset printing on uncoated papers cannot even reproduce half of the colours feasible on the proof printer.

The use of the relatively colourimetric rendering intent results in larger gamuts in each case, but this software distortion of the output data does not correspond to the reality of our above-mentioned question. The HLC Colour Atlas therefore shows the gamuts with absolute colourimetric rendering intent.

Further results

In which colour ranges can a process represent more or where can it represent fewer colour tones? Here are some important results:

Even though sRGB has a good colour gamut, it can only represent a few colour tones in the blue range. Here it is even superior to offset printing and even more so to proof printing, whose gamut is almost twice as large in this range.

sRGB can reproduce light colour tones, especially in the yellow and orange range, somewhat better than the proof print, but the latter brings out even more vivid nuances in darker colours.

There are many other differences in detail, using the data from the HLC Colour Atlas XL, these differences can be examined in detail.

HLC-Colour Atlas XL Gamuts (Photoshop PSD)
The best way to do this is with the pdf version of the atlas, which you can download here shortly.

Notes and open questions

More uniformity and transparency would be desirable in gamut algorithms. In contrast to Photoshop, in the colour management programme ZePrA from ColorLogic, for example, one can freely define the desired maximum DeltaE00 deviation from the nominal value; this definition then has a decisive influence on the gamut limits. A brief comparison between the Photoshop and ZEPRA-generated gamut (the latter with maximum DeltaE00=1) shows that the Photoshop gamut is somewhat smaller.

Comparison of the gamut of the colour H50 created in Photoshop and in ZePrA.

The question of a suitable algorithm for the limits of the CIELAB colour space based on the surface points defined by the rectangular spectra is open. If these limits were fixed, the question could finally be answered whether a CIELAB coordinate represents a colour at all (=lies within the body spanned by the rectangular spectra) or not. Non-colour could then be excluded from the representation. A suitable approach might be the wireframe surface generation by means of "meshing", whereby polygon surfaces are placed on these points. Only points in the inner region of the body defined by the polygonal surfaces are colours, points lying outside are excluded from the colour calculation or reverted to a surface point.

Author: Holger Everding