6 Replies Latest reply on Jan 13, 2009 4:39 AM by Gernot Hoffmann

    More ink = lighter image ?

    Gernot Hoffmann Level 3
      A question to those who are familiar with
      ICC Profile Inspector:

      Open for instance EuropeISOCoatedFOGRA27.icc
      Choose A2B1, which means A-to-B1:
      Input 4 channels CMYK and output Lab.
      Rendering Intent Media-Relative Colorimetric
      (RelCol).
      Choose CLUT
      All sliders at the top
      Read L*=65280
      This corresponds to L*=100
      Mapping to paper white has to be done separately
      Move sliders CMYK to the bottom, C=M=Y=K=max(100)
      Read L*=8418
      This should be the darkest rich black
      Move the third slider (Y) to the top, Y=0
      Read L*=7566
      Without yellow ink the surface becomes DARKER.

      1. I expect: more ink = less reflected light.
      2. Where is the ink limit ?

      Any explanations ?

      The effect cannot be simulated by Photoshop.
      Here we have L*=0 for C=M=Y=K=100, which is
      wrong but in a certain sense understandable.

      Best regards --Gernot Hoffmann
        • 1. Re: More ink = lighter image ?
          Lou Dina Level 3
          Hello Gernot. I hope all is well with you and that you had a good New Year.

          I have ICC Profile Inspector but haven't used it much. I tried your experiment, but used ISO Coated, since I don't have the Fogra 27 profile on my system. I got similar results.

          The following is speculation on my part, and is not backed up mathematically. It makes sense to me that adding yellow could result in reduced density, just like adding a little bit of white paint to a liter of black paint would give a slightly lighter color. I know...inks are more or less translucent, but I think the effect may still be similar. In fact, with the ISO Coated profile, I moved the Yellow slider to 0, then I moved the Magenta slider from 15 to 14 and got a darker value, which suggests that with this profile, 100C/100M/100K doesn't give Dmax. You need to back off just a little bit on the Magenta ink to obtain the Dmax for this profile. And adding ANY yellow reduces Dmax.

          Also, I have built a lot of profiles in the past for B&W printing in QTR and IJC/OPM (using the Epson UC color inkset). To get a neutral profile, I would use combinations of light gray, light cyan, light magenta and sometimes yellow in the highlights and midtones. But, I never used yellow in the shadow areas. I used mostly Black (either MK or PK inks), plus sometimes a little dark cyan or dark magenta. The added dark inks sometimes increased Dmax, especially if black ink alone wouldn't do it, but I used them mainly to neutralize the warm tone of the Epson black inks. Adding too much of even the dark inks could result in a reduced Dmax by exceeding the ink limit of the paper. Adding yellow definitely reduced the Dmax. I never used any light inks at the shadow end. Perhaps this is the same thing you are seeing here.

          I don't know how to obtain Dmax info from this program (or even if it is available).

          Those are my thoughts after a quick look.

          Lou
          • 2. Re: More ink = lighter image ?
            Gernot Hoffmann Level 3
            Lou,

            thanks for taking the time. Wish you all the best
            for the new year ! Especially no color in your B/W
            prints.
            By the way, this profile Europe...FOGRA27.icc came
            with CS2. It's not the best choice - better is
            ISOcoated-v2-eci.icc.
            Your much appreciated explanations are clearly in
            contradiction to text book knowledge: offset inks
            are transparent filters with thickness 1 or 2 micro-
            meters. Therefore a new and fresh interpretation is
            welcome.
            There are two open questions:
            1) Which black point is used for BPC - the darkest
            black or the darkest neutral black ?
            2) These CLUTs are using n points on each axis C-M-Y-K,
            n=11 for Europe and n=16 for ISO.
            All combinations are used, which results in 16^4 =
            65536 sets L-a-b. Where is the ink limit ?

            In the moment I'm programming graphics for gamut
            boundaries in 3D and 2D, similar to ProfileMakers
            GamutView.

            Best regards --Gernot Hoffmann
            • 3. Re: More ink = lighter image ?
              Lou Dina Level 3
              Hi Gernot.

              I looked at the ISO Coated profile that came with CS2 using ColorThink Pro. It shows a Maximum Black of 12L, 1a, 2b (rounded off the the nearest integer). Black ink alone shows a value of 17L, 0a, 0b. Looking at the Relative Colorimetric Curve in ColorThink (for Neutral Rendering), it shows a Dmax of 12.4L with the following CMYK percentages (using an input of 10, which I assume represents L*):

              100C, 91.7M, 52.4Y, 100K, for a total ink coverage of 344%.

              If you follow the curve all the way out to an input level of 0 L*, the CMYK values are:

              99.2C, 99.2M, 51.6Y, 100K, for a total ink coverage of 350.1%. The L* value is still 12.4, so the additional ink does not increase Dmax. In fact, input levels 0-6 are all identical. Input levels 7-10 show CMYK values that change very slightly, but Dmax remains 12.4 L*.

              I assume the yellow is included in the mix to maintain better gray balance in the shadows, since so much cyan and magenta ink is utilized. It still isn't perfectly neutral, but pretty close (12.4L, 1.0a, 0.6b). Unlike Epson black ink (PK or MK), which is warm, the K ink used for presses is much more neutral, so the added yellow makes sense to counterbalance the C & M inks. I don't know if the Dmax would improve with the removal of the yellow ink or not.

              By the way, the profile curve shown in ColorThink has a maximum yellow of 62.7Y at input level 16 (16 L* output). Then it rapidly drops off to about 52% at input level 12 (12.5 L*). So, the ISO profile also reduces the amount of yellow ink as you approach Dmax.

              I cannot answer the two questions in your last post. Perhaps someone else has some answers for you.

              Hope some of this helps.

              Lou
              • 4. Re: More ink = lighter image ?
                Gernot Hoffmann Level 3
                Lou,

                thanks for your examples. They explain nicely the black
                generation.
                I don't have ColorThink, but one can get some insight as
                well by ProfileMaker > ProfileEditor > GrayBalance.

                Inspecting some older inkjet profiles, I found that the
                results are depending very much on the chosen GCR method
                and parameters. Sometimes Y is monotonously increasing,
                sometimes not.
                Now, all this concerns profile CLUTs BtoA1 (1 for Relcol)
                together with the associated onedimensional input and
                output curves. Input is Lab, output CMYK. Ink limit is
                here established.

                The CLUTs AtoB1 (together with IO-curves, which are mostly
                straight lines) reflect the results of target printing
                and measuring. Input is CMYK, output is Lab.
                Instead of printing 65536 patches, about 1000...1500 are
                used, and the other table entries are interpolated.
                Here we don't have a total ink limit (TIL). Both targets
                ECI-2002CMYKi1 and IT8.7-3CMYKi1 contain patches with CMYK
                100% each, 400% ink.

                Nevertheless, at least an inkjet print has to be protected
                against too much ink.
                My workflow, using a RIP, is as follows:
                1. Print linearization target by RIP.
                2. Let dry one day.
                3. Make linearization profile by RIP.
                4. Prink ink limit target, inspect visually and define TIL.
                So far we have a linearized machine with explicitly
                defined 'RIP-TIL'.
                5. Print calibration target by Photoshop.
                Excesssive ink is prevented by RIP-TIL.
                6. Let dry one hour.
                7. Make Profile by ProfileMaker, using a Profile-TIL, which
                should not be larger than the RIP-TIL.
                As a result, we have TIL in fact at two positions: in the RIP
                and in the profile.

                My toner printers (PostScript) don't have linearization, there
                is no external RIP and no RIP-TIL. The machines are perhaps
                internally protected internally against too much toner.

                Best regards --Gernot Hoffmann
                • 5. Re: More ink = lighter image ?
                  Lou Dina Level 3
                  Hi Gernot.

                  It's been quite a while since I built profiles for a RIP (I use RGB profiles with my inkjet). If I recall correctly, (and I may not), I used to determine the individual and total ink limits first, then I linearized using the RIP and my EyeOne after printing a linearization target with the predetermined ink limits. Once the linearization data was read into the software, I printed a profiling target. It also seems to me that I set the TIL in "either" the RIP or ProfileMaker, but not both. Like I said, it has been awhile.

                  The ISO Coated profile has a white point tag, but no separate black point tag shown in the Tag Table (similar to ProfileMaker). Monaco Profiler has a separate black point tag in the Tag Table.

                  I looked at a custom press profile a friend of mine made using Monaco Profiler. In the overview, it shows Maximum Black as 11L*, 3a*, 5b*, (not very neutral), but the neutral rendering curve shows a Dmax of only 13.6 L* (perhaps due to inclusion of CMY inks to attain neutrality). So, this suggests that, at least with this profile, Maximum Black does NOT coincide with a neutral black. I am not sure if this is representative of most press profiles or not.

                  Perhaps this is not a valid comparison, (since it is not CMYK), but I have an RGB profile created by Monaco Profiler (built by a different color management expert). This is for Crane Museo Max matte fine art paper and an inkjet, so the Dmax is low, as expected. The Overview shows a maximum black of 18L, 1a, 3b, but the neutral rendering curve shows a maximum "neutral" black of 20.6L*.

                  I also have a profile built using ProfileMaker and the identical input data. I was curious how they would compare. (I own ProfileMaker, but do not own Monaco Profiler). It also shows a Max Black of 18L* on the Overview tab of ColorThink, like the Monaco profile, but the neutral rendering curve for RelCol shows 18.8L*.

                  I'm not sure if this information is helpful, or just adds confusion. It appears Dmax is not the same a the darkest "Neutral" black from inspecting these profiles. I'd guess that the profile is using the Max black instead of the Max neutral black as the Dmax, since it is hard to detect any color toning as you approach solid black.

                  You can also open both the profile and the reference data in Gretag's free ColorLab (undocumented) to see the input data and the output data. There is probably an easier way.

                  Lou
                  • 6. Re: More ink = lighter image ?
                    Gernot Hoffmann Level 3
                    Lou,

                    the discussion is helpful, thanks. But it's hardly possible
                    to discuss all the details.
                    I don't think that one needs ink limits for single channels
                    (see below).
                    In my workflows the linearization happens for unlimited
                    C,M,Y and K. Then the TIL target is printed and interpreted
                    by appearance. RIP-TIL is defined in advance to printing the
                    calibration target.
                    So far for ColorGate ProductionServer. Instead of the
                    built-in profiler I'm using ProfileMaker.
                    I think it doesn't hurt if I define here an additional TIL,
                    preferably a little lower than the RIP-TIL.
                    K can be limited below 100, but so far this wasn't necessary
                    or useful.

                    Meanwhile I've finished the gamut volume visualization,
                    chapter 9 here:
                    http://www.fho-emden.de/~hoffmann/labproof15092008.pdf

                    It turns out that one has to compare RGB gamuts and CMYK
                    gamuts in planes L*=const not for the same L* as done in
                    ProfileMaker(GamutView). RGB-white is mapped to paper white
                    (RelCol) and RGB-black to the darkest printable black (BPC).
                    Given RGB-L*, the gamut boundaries for CMYK have to be shown
                    for a corrected CMYK-L* value.
                    The darkest L* is about 3.7 for an inkjet. Rather low, but
                    that's in the CLUT (measuredby I1). PM GamutView tells me
                    something between 7 and 8. Measuring the target print by
                    X-Rite DTP22 delivers values below 5. All that is probably
                    at the instrument's limits.
                    But it's safe to say that the achievable inkjet density is
                    considerably higher than that of offset processes.

                    Isn't the functionality of ColorLab integrated in PM ?

                    Best regards --Gernot Hoffmann