CHROMiX

Known issues with profile graphing?

I’m trying to graph profiles with ColorThink 2.2’s 3D Grapher and running into awful looking graphs. The profiles are from ProfileMaker 5 and Agryll. Both are CMYK profiles with severe ink limits and no black generation. The problem on the graph is with the bottom end, so I’m wondering if ColorThink has trouble with small TILs or CMYK profiles with no black generation?

I’m running ColorThink 2.2 on Windows2000. I’ve posted screenshots here:
PM5 (with source data as points)
Argyll (with source data as points)

Thanks,
jeff

I’m seeing the same issues with Apple’s ColorSync Utility. PM5 can graph the profiles properly – not a surprise for the PM5 profile, but I was happy to see that they handled the Argyll profile properly as well.

If anyone has any thoughts on why ColorThink and ColorSync Utility might have trouble, I’m still curious.

Thanks,
jeff

At 9:51 AM -0800 1/6/06, jefrey_c wrote:

I’m seeing the same issues with Apple’s ColorSync Utility. PM5 can graph the profiles properly – not a surprise for the PM5 profile, but I was happy to see that they handled the Argyll profile properly as well.

If anyone has any thoughts on why ColorThink and ColorSync Utility might have trouble, I’m still curious.

I would like to take a look at the profiles themselves but I suspect that the profiles are having problems with ink limiting - I suspect the target itself was ink limited?.

ColorThink graphs gamuts in a very different way from most other graphing applications.

We graph the gamut of the device itself, which means we get the numbers from the CMYK->Lab (proofing) transform. This is the gamut of the actual device behavior, not necessarily what you will get when you print to the device through the profile.

Most other graphing applications graph what I call the “rendered gamut”. Which is the gamut the profile will give you when printing to the device. This may seem like splitting hairs but it can be a huge difference (as you are seeing in the profile you graphed)

Truth is, I think ColorThink’s method is the most absolute and verifiable method of graphing a gamut. We do a simple transformation through the profile to obtain the gamut of the device as described by the profile. When you graph the measurements that created the profile you will find that (for correct profiles) the measurement points will correspond well with the shape of the gamut.

In other method some arbitrary set of colors (Lab values) is sent through the profile to RGB/CMYK/etc and the back through the profile to Lab. This has issues including:

  • which set of data was used? all Lab values? (many of which are not visible colors) visible Lab values? Lab values inside another gamut such as Adobe RGB or sRGB? They don’t say so we don’t know if the graphed gamut is a good indication of real-world performance of the profile.

  • the resolution of the data doesn’t seem to be fine enough to actually hit the 100% values of the output device. This means that the yellow point of the profile’s gamut always seems to be stunted in some way and not the sharp 100% yellow point that the device can actually produce (and you’ll see in ColorThink). This may be a good indication that the use of the profile will actually never give 100% yellow but unfortunately we don’t know that for sure (see previous point).

  • In only graphing the data the profile will produce when printing, you are not testing the behavior of the profile with all data. Your profiles are probably a good case-in-point. They may never produce inking higher than, say 280% but that doesn’t mean that you won’t give them inking higher than that. Many printers receive CMYK from customers and don’t know how it was produced and can ink all the way up to 400%. Soft or hard proofing with your profiles will give erroneous results (I expect) and the graph is an indication of that.

To test my theory about your profiles, open a full-inking CMYK profiling target in Photoshop (like the IT8 or ECI target) and ASSIGN your profile to it (no conversion). Then a) see how the heavier-inked areas on the target appear on screen and b) what Lab values you get for those areas. I suspect you will see very dark Lab values just like what is shown in the graph.

Please let us know what you find and send the profiles to me directly if you want, to support (at) chromix.com

One of these days I want to right a white paper on gamut graphing but in the meantime I guess I’ll stick to long-winded list responses…

Regards,

Steve


o Steve Upton CHROMiX www.chromix.com
o (hueman) 866.CHROMiX


Post generated from email list