trouble creating good printer profile - help!

gkoulomz · Nov 11, 2009

I have been trying to profile my monitor and printer, but cannot achieve anything near an accurate match. For example, near-whites on the monitor have a significant yellow cast on the printer..

I have the Eye-One Photo LT running on Windows XP Sp3, with a MultiSync LCD 2690WUXi monitor and a Canon i9100 printer.

(I have D50 lighting for examining prints. I have performed the "advanced" monitor profiling process, using gamma of 2.2, luminance of 120, D65 white point per X-rite recommendations. I have turned off printer based color management. I have cleaned the printer nozzles. I have profiled the printer following x-rite's recommended settings).

To help me track down the source of my problem, I want to isolate the problem to the display or printer profiling (I suspect the printer, but how can I tell?)

Your answers to the following questions would help me very much.

1) using Eye-One Share Create/Measure, I calibrate the i1 instrument, and then immediately, leaving the instrument on the calibration tile, take a sample of the tile, and get the following values: L*: 95.9 a*:-.05 b*: 2.5.
IS THIS CORRECT? I would have expected BOTH a* and b* to be close to 0.

2) if I measure (using Eye-One Share Create/Measure) an image containing a known (in-gamut) LAB color value on a properly profiled monitor, how close (delta-E) should the displayed LAB value be to the value specified in the image file in a properly color managed environment?

3) If I print an image containing a known (in-gamut) LAB color value on a properly profiled Printer, how close (delta-E) should the printed LAB value be to the value specified in the image file in a properly color managed environment?

4) The file GretagMacbeth\i1\Eye-One Match 3\Reference Files\Monitor\"Monitor Testchart.txt" contains RGB values. What is the implicit color space / profile associated with this file (e.g. sRGB, aRGB, ...).

Any answers would be greatly appreciated.

Thanks in advance,

George.

Garbz · Nov 12, 2009

White on the monitor always matches the white on the paper. It all depends what light you use. You won't ever be able to properly match white on white without a dedicated light box. In that case matching it is usually a case of setting the Lightbox to D55 or D50, and the monitor to 6500K (this will compensate slightly for the fact that photo paper actually has a slight blue tinge).

The problem can be either. In the case of calibration you're usually calibrating something against something else and ruling out all sources of error, and taking into account eyes will adjust. This is why the environment is so very important when you're trying to match something self illuminating, like a display to something that is illuminated by a specific source.

Anyway:

1) Do you get a deltaE value for this too? The numbers are some what meaningless. This may represent a delta-E less than 1.

2) However close you can get it. DeltaE is a measure of at what point people can detect a colour has changed when the two colours are displayed directly next to each other. A deltaE value of less than 1 I think means that 50% of the people won't detect the colour difference. A value less than 1, and even greater than 1 is easily sufficient for photography, but may not be for things like medical imaging. See my note at the bottom.*

3) Exactly the same as above. If you have unlimited money and the best equipment it'll be zero. Since we don't it won't be.

4) None. That's the point of a monitor test file. When it displays those values it ends up displaying the monitor profile.

*Note: All of this is somewhat academic. Sure we'd all like 100% perfect colour reproduction, however a lot of this hangs on the fact that we can compare images side by side, however our clients aren't. The idea of perfect colour management is more critical for industries where the print needs to be 100% perfect first go with no proof print and no room for second chances. Photography is not one of those industries. Our clients can't tell if an image is colour managed or not because they don't compare their prints to our working screens, and if we are not happy with the first proof of our prints, we can usually for next to no cost run another proof.

Also one more note. Does the X-Rite software recognise your screen's internal Lookup Table? I couldn't get iOne Match to recognise the fact that my 2690WUXi has a 12bit lookup table, so I use NEC SpectraView II to calibrate.

gkoulomz · Nov 13, 2009

Thanks for your answer....
My goal here is to figure out what is off, the printer profile or the monitor profile. So having a ballpark value for acceptable dE values in a good workflow would help. I.e. I can see if my dE's are in the ballpark. I'm not looking for perfection, just objectively "good enough". Being new at this, I don't have a feel for what the numbers of a good set up look like.

Your last comment is VERY interesting. I have the same monitor, and I have been working under the assumption that Eye-One will operate on the software profiles (i.e. the ones PS will use) rather than the 12-bit ones implemented in the monitor hardware. This may make the monitor seem like a bit of overkill, but until I can spring for the extra money for Spectraview software, it will have to do.

BTW, did Spectraview accept the Eye-One Pro puck?

Did you using Spectraview significantly improve your results?

Eye-One support has been rather opaque in their answers....

Thanks,
George

Garbz · Nov 16, 2009

Well as a general ballpark, dE under 1 is excellent for the purposes of photography. Is either your monitor or your printer registering that with your eye-one hardware?

Spectraview supports the Eye-One Display2, and that's about all I can commit to without going and finding the manual. The Display2 is actually what they ship as rebadged hardware with the NEC spectraview kits. I can suggest download the 30day trial and help.

Those software profiles for photoshop don't calibrate the monitor. They tell the operating system or rather the applications what shape the monitor's gamut is, what the native white point is, and any information needed to convert one colour to another so it displays well. However the actual deviation from the tone curve, i.e. the bit that says push the blue slightly at this level so the screen is still grey here, is written into the lookup tables. In the absence of a hardware lookup table a software lookup table in the video card is used.

This however has the downside of lost fidelity. As soon as you change the lookup table of the video card you eliminate any chance of it displaying all possible colour values since for instance to make the white point a bit bluer the video card won't ever put out a red value of 255 when instructed to make "white". If you can understand what I mean. Compared to the lookup table in the monitor which is 12bit -> 8bit. Basically every 8bit colour value is mapped to it's correct shade in the screen. So the video card still has it's full 8 bit output to work with, and the monitor interprets and displays the correct value with no loss of fidelity.

I hope you understand this, my explanation is crap!

But basically the software colour profile tells the app what the screen should look like.
The software lookup table tells your video card how to fudge it's output so the screen looks like it says in the profile.
And the hardware lookup table tells your screen how it should behave for any given input which is by far the best option.

I am unsure if the eye-one software supports this, basiccolor does so I know it's not necessarily propriety but it's definitely worth looking into.

As something to compare to, I have calibrated my screen white to 5504k (D55). My calibration results are 5522k with a dE of 0.33 (which means that most if not nearly all people can't tell that my screen has a slight green / purple cast compared to a pure 5522k value). And the average dE value across 64 sample points from my screen is 0.35. I would use this as an indication that my screen is doing pretty good.

Also the more you push your screen from it's native white point and native gamma, the worse it will perform. At 6500k white point I measure about 0.5 average dE value. I calibrate to D55 since it's closest to the 5700k or so native white point of this screen, however if you're comparing to a print you may not be so flexible. Remember your eyes adjust white point to the most obvious thing they see, so it makes no difference to you what white point your screen has until you compare the screen to something.