Inkjet Primaries, GRACoL primaries, ICC profiles.

Steve(or anyone else),

Using a RIP I linearize a printer, using appropriate ink restrictions for a GRACoL target proof. I then create a profile and make a print. Upon verification of the resulting proof we notice that the ∆E between 100% Magenta of the Proof and the GRACoL spec are around 5 ∆E76. After a bit of optimization in the RIP the ∆E for the primaries falls below 4, and everything is happy…but here’s the question.

If I look at a plot of the magenta patches I see that the hue of the the two 100% patches are different. Because of the dramatic hue hooks in the inkjet, similar chroma values produce different hues. Looking at the total gamut volume however the entire GRACoL gamut is contained by my inkjet profile. From this perspective I could assume that color management will allow me to match any color contained within my output gamut. Where that logic seems to break down though is that while I can produce the a GRACoL 100% magenta using only magenta ink on press because my inkjet’s primaries differ from those of my theoretical press I have to combine two or more colors to produce that 100% magenta patch. Because of the particular properties of my inkjet primaries is it correct to assume that I can exactly match that 100% GRACoL magenta?

Is the high ∆E the result of basic “noise” in the color management system, or a hard limitation of the results I can achieve using the specific primaries of the printer. Or put simply is there a combination of inkjet CMYK that will perfectly match an ISO Magenta? Can I really reproduce every color contained within the gamut boundaries of my printer’s ICC profile?

So this is one of those questions that I feel I should have an answer to, but I fall back on speculation rather than facts. Can you provide a bit more depth?

There are many reasons why the ∆E may be high in a situation like this. When you start stacking the various elements that can contribute to this, you’ve got: measurement device repeatability, printer stability (of each printer), mathematical rounding errors, the accuracy of the profiles, the instability of color around that hue hooking area. Maybe that last one being the greater.

This is a fascinating question. Theoretically of course, it is not possible to reproduce every color within another printer’s profile, since every color necessarily includes every combination of color numbers. In much the same way, you can’t define every single place on the edge of a ruler, because there are those little gaps between the 1/32 marks. This is sort of a problem of resolution. If you are talking about “human perceptible” color, then that might be a different story - you should be able to reproduce every color that is humanly perceptible - although I’m sure there are some who would disagree on that also.