i1Profiler - 7 Color Profile Contamination

[mikestr]

First of all, if you want anyone to comment in a specific way about your results you need to provide all your data, including especially the settings you used in i1Profiler. (Did you tell us how you made this conversion--in what application? Simply insisting you're right will not advance any solution for you beyond possibly drawing the wrong conclusions. I will nonetheless tell you that I have before me a similar 7-color profile, also made with i1Profiler, which yields the following build from the Lab values you've supplied: C28 M20 Y0 K78 R0 G0 B53. While we can't directly compare exact percentages for these different output devices (this was an offset press, not that it matters), it's clear that what you say about i1P--that it produces almost no black separation in a multicolor profile--can't be an inherent feature of the profiling engine or it would happen every time that color was converted by it.

 

[Schnitzel]

Mike,

I'm not insisting I am right - I'm just pointing something that in my intuition and logic seems wrong. The color scientist in my team also agrees that something looks strange in the profile. I'm simply trying to find out if anyone else had seen this behavior.

I did the conversions using a CMM that I've written, and validated the results against Argyll CMS.

...can't be an inherent feature of the profiling engine or it would happen every time that color was converted by it.

I disagree - there may be edge cases that cause this behavior. Also, every computer software has bugs (not that I'm implying that's the case here).

Regarding the data, as I wrote before, I'm not permitted to share it since it concerns an area that is under R&D.


In any case, I really appreciate your contribution and everyone else's. You've given me some directions to follow from here.

Thanks again,
Shahar

 

[mikestr]

Well, clearly something went wrong for you--there is no dispute over that. The question is how and why. I should be able to produce a similar result from your Lab value using i1P and I can't, so I would simply say that you've introduced a source of error somewhere. My main point is that it is a mistake to reflexively attribute a particular result to the design of the profiling engine. There are cases where a color may be mapped a bit differently and produce a somewhat better or worse separation, but even this can't really be evaluated unless one sees the underlying measurement data and what decisions were made in setting up the profile--ink limits, black generation, etc.--and the conversion settings. Because you won't share any of this AND have introduced a proprietary and apparently secret conversion engine you couldn't have expected much help here in resolving this.