Need help interpreting color test results

Discussion in 'Photoshop Tutorials' started by ronviers, Aug 16, 2006.

  1. ronviers

    ronviers Guest

    I have a titanium white backdrop - perfectly white.
    I painted my main light reflector, by mixing acrylics, so it lights the
    backdrop neutral, that is, when I photograph the backdrop, in raw, and
    load the image into RSE or ACR in Adobe RGB at 16 bit, I get a readout
    of R=197,G=197,B=192 a tiny bit yellow I admit but I thought I could
    live with that being tired of mixing paint. I have never used a D50
    but I bet they are not much better than that.
    Anyway, next I convert the raw image, to Tiff with RSE and PSD with
    ACR, with no modifications in either converter, and open them with CS2.
    I do this on two computers, my fast one and my slow one. My working
    space is Adobe RGB and mode is 16 bit/channel. Both computers CS2
    environments are identical and both use CRTs.
    After the conversion I have both the Tiff and the PSD loaded on each
    computer. I check the readouts at the several coordinates and get the
    exact same reading on both the output from RSE and ACR - e.g., at
    5000x3500, I get R=196,G=197,B=191. This is a surprise but also a
    relief. I kinda thought they would convert them slightly differently.
    Then I set CS2 to display the image of the converted raw file, the PSD,
    in full screen preview. So now I have a Photoshop image of my raw file
    displayed on the CRT. Then I set the original raw file still in the
    camera as the source for the custom white balance setting for the
    Next I photograph the Photoshop image at a quarter second to get
    several screens of refreshed image.
    Then I load that image back into RSE and ACR. So what do you think I
    get? I thought I would get a nice neutral image possibly a little
    yellow. But no! I get a very blue image. What could be going on? A
    typical reading after the photo is R=56,G=87,B=143. Look how blue that
    is. Can someone help me interpret these results? Shouldn't this be
    a good test for checking the color moving through the system?

    ronviers, Aug 16, 2006
    1. Advertisements

  2. ronviers

    Mike Russell Guest

    Hi Ron,

    Not many people have systematically calibrated their system, from start to
    finish, as you have, so congratulations. My guess is that your results will
    be of interest to others.

    Now for your questions. The missing piece of information is monitor
    calibration. More than likely you have the color temp of your monitor set
    to something like 6500 - a rather blue appearance. Try dropping the color
    temp to about 5500 or so, using Adobe Gamma or some other means such as
    adjusting using your video driver, and you should get closer to neutral.

    As for whether it's a good test for color moving through your system: I
    suggest you ignore what others say and continue your experiments to their
    logical conclusion. I think you have concocted something original and even
    revolutionary that may form the basis for others to calibrate their systems
    without expensive equipment. Keep us posted.
    Mike Russell, Aug 16, 2006
    1. Advertisements

  3. SNIP
    It indicates that your display is emitting light with a higher color
    temperature than your camera is assuming. Many default to something
    like 9500 Kelvin. You could try, if possible with your camera, to
    manually set the Whitebalance to something like that, and see how it
    looks then.

    The kind of neutral (but slightly yellow) backdrop looks that way to
    your camera when captured with the lightsource and colorbalance you
    used. Your display will emit those colors with its own whitepoint /
    color-temperature setting. The eyedropper measurement is just the
    signal that gets sent to your display, not what is emitted.

    Besides, your camera may respond differently to the display's color
    temperature than to your original capture's color temperature. That
    has to do with things like metamerism and color (in-)consistency.
    Bart van der Wolf, Aug 16, 2006
  4. ronviers

    ronviers Guest

    Mike that was an amazing bit of diagnostics; I had my monitor
    temperature set to 6500. I reset it and reshot and got a
    R=100,G=101,B=98- just what I was hoping for. Notice it maintained
    that bit of yellow all the way through.

    ronviers, Aug 16, 2006
  5. ronviers

    ronviers Guest

    You are exactly right. It amazes me how well you see the big picture.

    Thanks as always,
    ronviers, Aug 16, 2006
  6. ronviers

    ronviers Guest

    I thought of one more step and it should be the first one. First set
    the 'Windows Desktop' to white. Then clear all icons and put the
    theme to Windows Classic to get rid of the blue line at the bottom.
    This turns the monitor into a white light. Then point the monitor at
    the titanium white backdrop and photograph the backdrop using raw.
    Then load that image into ACR or RSE to get a balance then adjust the
    monitor's temp according to that and reshoot as needed. This should
    complete the loop. Does that sound reasonable?

    ronviers, Aug 16, 2006
  7. ronviers

    Kingdom Guest

    We just enjoyed the round trip on this one.
    Kingdom, Aug 16, 2006
  8. ronviers

    ronviers Guest

    I got some sleep and then reread your reply. Are you saying the test
    is bogus?

    ronviers, Aug 16, 2006
  9. ronviers

    ronviers Guest

    The kind of neutral (but slightly yellow) backdrop looks that way to
    If I create a jpg image using CS2 with the following options:

    No color management
    8 bit/Gray scale
    Size = 3072x2048
    Filled with R=170,G=170,B=170

    And saved with the following options:

    No ICC profile
    Quality maximum
    Baseline standard

    Then upload the image to my camera (300d) and use the image for the
    custom wb source - will this enable me to get more consistent results
    from my cameras internal metering and software and enable me to use the
    camera for things like mixing paints for correcting color balance and
    balancing the RGB guns of my CRT?

    ronviers, Aug 18, 2006
  10. ronviers

    Mike Russell Guest

    Hi Ron,

    Possibly, but the details that Bart mentioned will prevent perfect color
    reproduction . The reason has to do with the fact that color is the result
    of a combination of physical, perceptual, and cognitive processes, none of
    which are represented completely in the RGB definition of a color, or the
    representation of color on the printed page using cyan, magenta, and yellow
    (and other) inks.

    In the case of some colors, such as deep blues and purples, this
    representation falls flat.

    Talk is cheap, and what counts is results. What no one can answer is
    whether you are on the track of something new that will help us calibrate
    things reasonably well without the use of expensive instruments. This is
    why I'd encourage you to complete your experiments and let the rest of us
    know how well it works. If, for example, you can get a good neutral gray
    this way, that's a big step in the right direction.
    Mike Russell, Aug 18, 2006
  11. ronviers

    ronviers Guest

    Thanks Mike,
    This is a nice project. When I am out of ideas for shooting and tired
    of Photoshop tutorials I can occupy myself with understanding my
    equipment. Even if the idea ultimately goes nowhere I will not be
    It's all about control.

    ronviers, Aug 18, 2006
  12. ....
    If it was that easy ;) I think you did not undestand the concept of
    white balancing yet, and, there is no need for white balancing in the
    camera if you set up a raw workflow.

    I'd recommend you go to the Adope page on CS2 camera raw file support:

    and read the two papers on "Understanding digital raw capture" and
    "Linear gamma". If you like Bruce Fraser's writing, consider to buy
    his book "Real World Camera Raw with Adobe Photoshop CS2"

    URL at amazon:

    The book is rather cheap, but you will also need a Gretag Macbeth
    Munsell Color Checker with 24 swatches. Read the book first since
    there are several flavors of the Color Checker and you shold buy the
    right one.
    In the book you will find a detailed description on how to calibrate
    the Adobe raw converter for your camera. A link is provided to a
    (free) action, which will perform the calibration for you, if you
    supply a shot of the color checker taken at reasonable lighting

    Peter Wollenberg, Aug 18, 2006
  13. Jpeg compression does not preserve colours; you'll need to check if jpeg
    greyscale compression keeps the same values before going any further with
    that idea.

    Andrew Morton, Aug 18, 2006
  14. ronviers

    ronviers Guest

    Hi Andrew,
    Thanks for the feedback. I realize now that the jpeg will not work. I
    am looking into replacing the data in a raw image and copying it back
    to the camera.

    ronviers, Aug 18, 2006
  15. ronviers

    ronviers Guest

    Hi Peter,
    I realize the jpeg will not work. But let's say I replace the data in
    a raw file with perfectly neutral data then copy that back to the
    camera then set that as my custom wb source then shoot jpeg to balance
    my colors. Won't that work?

    ronviers, Aug 18, 2006
  16. ronviers

    ronviers Guest

    I forgot to say that I appreciate the links and other information and
    that I have downloaded the pdfs and will read them later today.
    Unfortunately buying books and charts are out of the question for now.

    ronviers, Aug 18, 2006
  17. I'm not sure what exactly you want to achieve, but a better white
    balance is not what you will achieve if that's your aim. The principle
    of 'custom' white balancing is that you shoot a neutral grey card in a
    certain light, so you can compensate for the color cast that this light
    source would cause by comparing this shot against neutral grey.
    Consequently, the reference shot is normally *NOT* neutral grey, that is
    exactly why you use it for white balancing in the first place. Replacing
    the reference shot by an artificially made neutral shot will make white
    balancing go wrong, not be more accurate. Also, your camera does not use
    this reference shot for anything other than to set the white balance to
    the light source in question. It is not 'calibrated' against this shot,
    nor will it calibrate itself against that shot.
    Johan W. Elzenga, Aug 18, 2006
  18. ronviers

    ronviers Guest

    Hi Johan,
    I am not going for accuracy just consistancy.

    ronviers, Aug 18, 2006
  19. But why would this give more consistant results than just setting the
    white balance at a fixed value, say 5500 K? That is exactly what you
    would be doing with this method, except you'd be using a very
    complicated way of achieving that result.
    Johan W. Elzenga, Aug 18, 2006
  20. ronviers

    ronviers Guest

    It may not but I want out of the auto modes in hopes of minimizing the
    magic the Canon OS performs.

    ronviers, Aug 18, 2006
    1. Advertisements

Ask a Question

Want to reply to this thread or ask your own question?

You'll need to choose a username for the site, which only take a couple of moments (here). After that, you can post your question and our members will help you out.