New opportunities in offset color separation: iccGPU


Active member
Spectralcalc iccGPU™ introduces new possibilities for offset color separation.
Published for commercial use at:
The first year of use is usually free.

iccGPU uses a graphics accelerator whenever possible for the first time in the history of computer color management, so the GPU is in the name.

The clear step-by-step design of profiles allows the color technologist to control the process with the help of tables as deeply as no program can intervene at any intermediate stage. In fact, the colorist becomes a co-author of the programmer when creating the separation profile.

I made three color separation profiles for Fogra 39 in Spectralcalc iccGPU:

ISO Coated v2 Perceptual Skintones iccGPU - for maximum black width, a fundamental exception is made in the Perceptual table to achieve traditional skintones with a minimum of black ink. The Relative colorimetric table does not exclude the maximum black width for skin like it does in the Perceptual table.

ISO Coated v2 Heavy GCR iccGPU - profile with maximum black width and Heavy GCR generation for monochrome and achromatic images.

ISO Coated v2 BTAC 240 iccGPU is a profile with maximum width and blended black generation with the lowest possible offset lightness of a blend of achromatic super black (~ 60-42-38-100), but TIL=300 in chromatic colors. For any photographic image, except for skintones and achromatic, for them - the first and second profiles are better suited.

Idealliance specifications are similar to ISO.

All three profiles use magenta and cyan inks UCA and yellow anti-UCA to achieve the best possible print contrast with separations from the profiles shown. All profiles in the Relative tables use a very special maximum black ink width, which allows you to extract the highest possible pure saturated color from process printing. Wherever possible, the third contaminating color ink is replaced by black, which has less effect on the reduction of saturation and only adjusts the lightness. Thus, all 4 colors are found in color separation almost only in balance neutral and weakly saturated near-neutral colors close to it.

In the creation of non-colorimetric tables Perceptual and Saturation, the unique capabilities of the chromatic adaptation algorithm CIECAM02 are used. The Perceptual Skintones profile makes skin a little purer when separating with Perceptual Rendering Intent. RGB colorants are usually lighter than CMY colorants, so the Perceptual and Saturation tables, to preserve the saturation of the original, slightly lower the lightness of the colorants to maintain their maximum saturation in darker CMYK process colors. This perceptual gamut mapping is shown schematically in the screenshot.

Product support:

IMG_9136_iccGPU_at_iPhone8.PNG GrayBalance_3_new_profiles_iccGPU_2021_03_17.png


Well-known member
Thank You very much, mihas!

So, that is a tooI which is to use for creating own Profiles by using the charactarisation-Data-sets from FOGRA/ECI??
Just for a donation? Awesome for everybody who always struggled by trying Argyll - like me... :)

I would like to try similary Profiles for FOGRA51 and 52 sometimes as you have made for FOGRA39 thankfully sharing with us (skintones, heavy GCR...)




Active member
Thanks for the kind comments!
Indeed, iccGPU separations are more like Argyll separations than any others. Apparently Graeme and I came to similar conclusions about the need for an inverted calculation from the inks we need to the color.
And special tables Perceptual and Saturation based on CIECAM02 algorithms add additional useful functionality to iccGPU profiles.


Active member
ISO Coated v2 MaxK MaxWidth iccGPU - profile with Max Width and skeletal black MaxK. Colored paints in the balance are mixed into black to achieve maximum contrast only below a lightness of 25; up to this threshold, a single black ink draws neutral.
Thus, all four inks are not found anywhere in color separation, except in deep neutral shades. I decided not to leave a bit of color in super black completely without any admixture, otherwise the lightness of the blackest point with one black paint will be too high and the contrast of the photo image in print will suffer. But for the drawings, I'll make another Pure K.
In demand in flexography.

ISO Coated v2 Less GCR in Skintones iccGPU - profile with Heavy generation of black GCR and maximum black width, while for skintones an exception is made in all tables, and black ink does not fall into the color of the skin. It is convenient to use for color separation of artistic portraits on a gray background. The picture below shows a black channel as a result of color separation by this profile, in skintones there is no black ink, and in a gray background - many.


  • ISOcoated_v2_Less_GCR_in_Skintones_iccGPU.jpg
    279.9 KB · Views: 114
  • MaxK_MaxWidth.png
    41.6 KB · Views: 112


Active member
I would like to try similary Profiles for FOGRA51 and 52 (skintones, heavy GCR...)
I am afraid to work with FOGRA 51 and 52 because of the M1 filter and I am waiting for this filter to be deleted from the standard. The programs apply Lab D50 math head-on to the Lab M1 data and balance violations occur because of this, the programs simply do not know that the blue Lab M1 is not actually blue.

Heavier than the originals, the GCR 75's black generation further accentuates the balance differences due to the change of the M0 filter to the M


  • f39_GCR75_grayBalance.png
    153.7 KB · Views: 112
  • f51_GCR75_grayBalance.png
    149.2 KB · Views: 127
  • GCR75_CMY_neutral_f51_f39.png
    779.2 KB · Views: 117


Active member
I want to show how two-dimensional gamut mapping based on CIECAM02 works in the profile perceptual table from iccGPU. For comparison, I took a perceptual table of a well-known profile from ECI - ISO Coated v2, for clarity. I tried to show it on photographs, but there is not so clear a comparison than when you move the results close to each other and you can clearly distinguish the difference at the border, so let there be just squares with gradients.

Perceptual gamut mapping from iccGPU and from ColorToolbox

The gradient in the center of the composition - the original in sRGB with a sufficiently high lightness of saturated colorants, which does not exactly fit into offset printing, CMYK offset inks are darker than sRGB colorants and cannot be both light and highly saturated at the same time, one of the two, either we lose in lightness, or we lose in saturation. Actually, the two-dimensionality of gamut mapping in the simultaneous control of lightness and saturation consists of.

On the gradient on the left - the result of the work of the perceptual table from the first domestic iccGPU profiler. The gradient on the right is the result of the ECI profile perception from ColorToolbox. Below, for clarity, the results of two different perceptual tables of different profilers meet together, forming a border where the difference is clearly visible.

What we see: both perceptual tables dropped the lightness L of the original, which is unattainable in a triad offset, but at the same time the two-dimensional gamut mapping of the iccGPU profile tried to preserve the saturation C as much as possible, and the perception of the ISO Coated v2 profile coped with maintaining saturation worse, both in the upper and lower parts of the gradient with a delta E of about 8 are worse in Chroma. The upper and lower edges of the CIE LCh color gradients are signed, we look at Chroma (saturation) C in numbers and visually along the comparison border and draw conclusions.

When creating the iccGPU profiler program, I tried to pay maximum attention to the unprecedented purity and saturation of colors, both in the relative colorimetric table, and of course in the Perceptual and Saturation tables. A comparative example of very good profile separations from ECI shows that I was able to achieve exactly what I intended.


Active member
The spectral calculator has a significant update: now it is possible to build, in addition to CMYK, RGB profiles for printing devices with a low-cost annual subscription.
In the RGB section of iccGPU profiles, the ability to turn on suppression of blue turns purple by Hung & Berns or equalize tone to Hue Linearized CIELab from Fairchild & Ebner has been implemented.
This Rochester Institute publication was used to create the algorithms:

Test profiles in typical color gamuts Adobe RGB, Display P3 and sRGB here.

Below is an example of offset color separation with Perceptual gamut mapping and effect suppress "blue turns purple" for coated paper using standard tools on the left, and on the right using profiles from iccGPU.


Canon Research

Advancing Productivity and Service Delivery
Enhancing the Print Customer Connection (Part 4)

Read All About It