Skip navigation
elihorn
Currently Being Moderated

Legalities of color—and why are my RGB values inconsistent?

Apr 24, 2010 4:41 PM

Hi there,

I have a client who is trying to trademark their logo and the RGB values. I have sent them the files and corresponding RGB values, but when they check the files in their mac Digital Color Meter they come up with different values. On my own computer, the values are consistent in the DCM and Photoshop, although I opened it on another computer and found that while the values were still correct when I open the files in Photoshop, the DCM values were incorrect there too.

 

Does anybody have any history with this? Should I just tell them to ignore their DCM and that Photoshop is correct, or am I doing something wrong with profiles or something? It wouldn't be such a big deal but for the legal implications behind it...

 

Thanks for any help!!

 

Eli

 
Replies
  • Currently Being Moderated
    Apr 25, 2010 1:47 PM   in reply to elihorn

    Apple's Digital Color Meter has been broken for years and years.

     
    |
    Mark as:
  • Currently Being Moderated
    Apr 25, 2010 1:49 PM   in reply to elihorn

    But also remember that Photoshop uses the monitor profile to send the colors to the monitor.  If the monitor profile is bad…

     
    |
    Mark as:
  • Currently Being Moderated
    Apr 25, 2010 6:04 PM   in reply to elihorn

    Photoshop is accurate but I think what's missing is that ANY RGB values are dependent on a color space, such as sRGB. You should really use the LAB values which are not dependent on a color space.

     
    |
    Mark as:
  • Currently Being Moderated
    Apr 26, 2010 3:28 PM   in reply to bret_linford

    I will back what Bret says, you have to use a non-dependent colour space such as Lab or sRGB but if you want to use a device space you must identify it such as "23C, 45M using USSWOPV2"  where you link the colour values to the profile so that it is a complete value.  Lab is a complete value on its own, and easy to convert into other spaces.

     

    When they view a job produced with Lab root colour, they will see another value on screen but that will be correct behaviour as they will be seeing the intended colour as near as their device can show it and the readings in RGB are equivalent for their specific environment which is as much as anyone can do!  You can't view things in Lab (arguably ) it is a theoretical colour space only.

     
    |
    Mark as:
  • Currently Being Moderated
    Apr 26, 2010 4:16 PM   in reply to elihorn

    P taz is right, and to elaborate you can think of Pantone colors (often used as corporate colors)

     

    In Photoshop, make a new Lab document. In Color Picker go to Color Libraries. Pantone Solid Coated. 293. Note the Lab values: L29, a7, b-66.

     

    Fill the canvas with this color. You do not actually see L29, a7, b-66. You see the closest color your monitor has to offer.

     

    Now go to view: Proof Setup, Custom. Device to Simulate, choose sRGB.

     

    Now View: Gamut Warning. The screen goes gray. sRGB cannot reproduce L29, a7, b-66. Not perfectly, anyway.

     

    Now Edit: Convert to Profile. Select sRGB.

     

    The values are R0, G70, B173.

     

    R0, G70, B173 is not a color in and of itself. You have to specify sRGB along with the numbers. Now you have the official sRGB Pantone 293 C values (it's true, check a new Pantone book if you have one). It is not a perfect match to the real color, but it's as close as you can get in sRGB.

     

    CMYK is even sadder. It falls very short of L29, a7, b-66, no matter what CMYK color space you use. It's important to understand this if you ever try to recreate spot colors and not spend extra money for the special ink.

     

    Anyway, Lab is the common denominator. But when you print on a press, using a can of Pantone 293, you aren't printing L29, a7, b-66 – not exactly, anyway.

     

    Lab values are dependent on lots of things – the ink (is it mixed properly?) the white point (is the paper the correct color?), and the light you use to proof (soft white? bright white? daylight?)

     

    But if the ink is correct, the paper is correct, and the viewing conditions are consistent, someone in Argentina can agree with someone in Russia – yes, we see the same thing, this is what Pantone 293 C is supposed to look like. And if you have a spectrophometer you can measure the printed result to get Lab values, and it should be very close to L29, a7, b-66.

     
    |
    Mark as:
  • Currently Being Moderated
    May 5, 2010 4:41 AM   in reply to Printer_Rick

    A slight correction on the use of the DigitalColor Meter tool on a Mac.

     

    It's true DCM's RGB readouts will not be consistent from one display to the next because it's reading off the corrected/adjusted video LUTs from the system display calibration profile (different for each display) when sampling from a color managed app like Preview, Safari or Photoshop. You can see this correction especially in the 0-20 RGB shadow regions comparing to actual Photoshop RGB readouts.

     

    For example you may get a PS reading of 5 RGB for absolute black but DCM will show maybe 10-20 RGB. That's because the embedded display profile calibration curve in the profile is correcting for the nonlinear behavior of the display that was present during the calibration/profiling process. Color RGB numbers may be slightly close but never exact as well.

     

    Other than that DCM has a Lab setting in its drop down menu that does correspond to Photoshop's Lab readouts and is quite accurate. From my experience you can only get DCM Lab readout selection in its menu when using a Mac display. Not certain about this because I've only had non-Mac CRT's in the past that never allowed this selection. With my iMac's integrated LCD display I now get the Lab selection.

     

    I use DCM to get Lab readouts of color managed previews in Adobe's DNG Profile Editor when creating a profile for my DSLR using a color managed DNG Raw file of the X-rite Color Checker Chart. The Lab readouts in DNG Profile Editor are in the sensor's linear space which is quite dark and useless as a reference to the actual CM preview. The DCM Lab numbers are quite close to Photoshop's after opening the converted DNG Raw reference file of the CC chart with the newly applied custom profile.

     

    However, using Lab numbers is more useful for consistently nailing spot colors for logos and graphics, but it doesn't make for very pleasant looking photographs due to what the surround effect of reality does to human perception on contrast and color temperature.

     
    |
    Mark as:
  • Currently Being Moderated
    May 5, 2010 5:03 AM   in reply to Tim Lookingbill

    Tim Lookingbill wrote:

     

    ... Lab ... doesn't make for very pleasant looking photographs due to what the surround effect of reality does to human perception on contrast and color temperature.

    ?

     

    Not following you here. Explain?

     
    |
    Mark as:
  • Currently Being Moderated
    May 5, 2010 1:22 PM   in reply to Rick McCleary

    I was referring to "Scene Referred" CCchart Lab reference numbers derived from measurements taken of this chart from a spectrophotometer. You'll note the black patch is at L20 according to X-rite. The white patch is quite easy to attain L96 without clipping during exposure of the CCchart to build the profile from using DNG Profile Editor.

     

    What's not easy to control during exposure is the contrast ratio recorded by the camera. Adjusting the middle slider in Levels or the Contrast slider in ACR will show just how much contrast affects hue and saturation viewed on a display. Reality doesn't behave this way.

     

    Contrast/Luminance ratio's are disregarded or barely adjusted for in the DNG CCchart Wizard to come up with proper hue/saturation levels for each of the 18 color patches. The contrast adjusts required to get all 24 patches to match exactly (especially black at L20) makes the image not look right or follow a normal human visual response or perception.

     

    You can adjust contrast in the image after applying the new profile to get a reasonably correct look but quite a few of the 24 patches go off the charts according to the X-rite reference Lab numbers. The whole reason for going by the Lab numbers in profiling a device like a DSLR in the first place is to maintain consistency image to image by using measured results. As you can see this is not possible to achieve in a photograph except maybe in a controlled studio environment, but it does work when trying to match single spot colors that aren't influenced by the surround effect that affects human perception inherent in a photographed scene. A pleasing photograph is not about accurate numbers.

     

    A spectrophotometer is a static device that only reads and measures color wave lengths from color patches and converts them to Lab numbers. It's not a constantly adjusting device like our eyes which adapts to hue, saturation, color temp and contrast levels in any given scene.

     

    I'm going to post three images of a CCchart exposed under a high noon sun that I used to build two different DNG profiles. The first is using ACR default DNG converted source image and doing a straight dual illuminant table profile according to instructions. They're screenshots that include Apple's DCM to show how the black patch changes to accommodate exact Lab numbers as opposed to achieving a pleasing look. You'll note the more the image is made to "pop" the less accurate the Lab readouts and appearance.OriginalprofileSource.jpg

     
    |
    Mark as:
  • Currently Being Moderated
    May 5, 2010 1:24 PM   in reply to Tim Lookingbill

    Here's the second attempt modifying the DNG tone curve that affects contrast to get exact Lab numbers.ExactLabDNGEdit.jpg

     
    |
    Mark as:
  • Currently Being Moderated
    May 5, 2010 1:25 PM   in reply to Tim Lookingbill

    Here's a compromise by using the original dual illuminant DNG profile from the first but adjusting contrast in ACR to make the image look better.LookRightFinal.jpg

     
    |
    Mark as:

More Like This

  • Retrieving data ...

Bookmarked By (0)

Answers + Points = Status

  • 10 points awarded for Correct Answers
  • 5 points awarded for Helpful Answers
  • 10,000+ points
  • 1,001-10,000 points
  • 501-1,000 points
  • 5-500 points