Kodak grey card values

Discussion in 'Photoshop Tutorials' started by a, Feb 28, 2004.

  1. a

    a Guest

    If I take a photograph of the 18% Reflectance side of a Kodak grey
    card and open the image in Photoshop. What values should I expect for
    RGB in the Info palette?
    a, Feb 28, 2004
    1. Advertisements

  2. a

    Greg Guest

    Assuming the image is in the sRGB color space, and the image is white
    balanced accurately,
    the RGB values would be about 119,119,119, according to this CIE Color
    I put in a Lab value of 50,0,0 (which I think corresponds to 50%
    reflectance). If you select "scale XYZ",
    you can see that the resulting value for Y (which is Luminance) is close to
    18, but not exactly 18.
    (I'm not sure why there is this slight discrepancy).

    You also need to select "scale RGB" in order for the resulting values to be
    scaled to the range 0-255.

    The values will change according to the gamma of the color space the image
    is in.

    Greg, Feb 29, 2004
    1. Advertisements

  3. If you want to cross post a message, do so by filling in all the
    newsgroups. That way, people who read the same groups, won't see your
    message twice. Don't post the same message separately in many
    newsgroups. Also, people who answer your message, will only need to
    write the answer once. I posted the answer in alt.graphics.photoshop:

    18% reflectance is 18% L in the LAB mode. If you fill in 18,0,0 in LAB,
    you'll get RGB 48,48,48.
    Johan W. Elzenga, Feb 29, 2004
  4. a

    Greg Guest

    I don't think that's right. I think 18% reflectance is 18% *absolute*
    reflectance, which
    corresponds to middle grey, which is a Lab luminance of 50.

    Greg, Feb 29, 2004
  5. You are right. It's 1.41 hours AM here, so I wasn't thinking too
    Johan W. Elzenga, Feb 29, 2004
  6. a

    Greg Guest

    I guess the reason for this slight discrepancy is that middle grey really is
    reflectance, but it's rounded down to 18% for convenience?

    Greg, Feb 29, 2004
  7. a

    Chris Cox Guest

    No - 18% reflectance is close to 50 % in L*.
    Reflectance is not the same thing as perceived gray values.
    (because of the logarithm/power law in human vision)

    Chris Cox, Feb 29, 2004
  8. a

    a Guest

    My apologies for the crossposting, thats the first (and last) time
    I'll do that.

    And, if anyone else needs a reason NOT to crosspost - It's because you
    end up with intelligent, knowledgeable people NOT being able to engage
    in a structured conversation and NOT being able to develop their
    arguments with each other.

    The idea behind the original question was this:

    If I expose a roll of film and load the images into Photoshop with one
    exposure a picture of an 18% reflectance grey card, what should the
    RGB values be in the Info pallete? The consensus seems to be approx
    R=119, G=119 and B=119. If I then correct any discrepancies - can I
    apply the same correction to the rest of the roll of film?
    a, Feb 29, 2004
  9. These are the formulas:

    Y1=(L*+16)/116 = 0.569
    If Y1>0.207 Then
    Y=Y1^3 = 0.184
    Aha, the physical luminance or reflectance is 0.184 of 0...1.
    For sRGB:
    If Y>0.00304 Then
    These are the numbers in the file for the working space sRGB.
    An arbitrary working space with G=2.2 would deliver

    Visualization of 100 gray patches with CIELab and sRGB numbers
    on page 6 (200kB doc):

    It´s up to everybody which gray should be considered as "medium".

    Best regards --Gernot Hoffmann
    Gernot Hoffmann, Feb 29, 2004
  10. There is nothing wrong with crossposting to two groups, as long as you
    do it properly. Write ONE message, and set both groups in the newsgroup
    header of that message. That way, your message will appear in both
    groups, and all answers will also automatically appear in both groups.
    Johan W. Elzenga, Feb 29, 2004
  11. a

    nikki Guest


    There is a discussion on 18% grey card values at


    and they come up with RGB values of 209,209,209 using some very sound
    arguements and pictures(especially from Rantin Al)
    What is going on.
    nikki, Mar 1, 2004
  12. a

    Greg Guest

    I haven't tried to follow the whole thread in detail yet, but some of the
    reasoning just doesn't seem right.

    I've just taken a photo of my white wall with my digicam after calibrating
    the white balance,
    and the result is a lot closer to that predicted in our thread here. The
    wall was not evenly
    lit (but not terribly uneven, either). Centre pixel values are about
    corner pixel values 100,101,102, and if the entire image is averaged, the
    result is 122,123,122.
    (a simple average like this isn't totally valid, because the camera was set
    for centre weighted
    average - I didn't apply any weighting)

    The values predicted in that thread, at least for a typical gamma RGB space,
    just seem way too high.

    Greg, Mar 1, 2004
  13. a

    Greg Guest

    Forgot to specify the colour space of the digicam image - the camera outputs
    in sRGB, and I left
    the image in sRGB when importing into Photoshop.

    Greg, Mar 1, 2004
  14. a

    nikki Guest


    I thought so too , at first, but the tests I have made indicate the
    RGB value is higher than the the RGB 122,122,122 you indicate. You can
    make a wall into any value by just changing any of a dozen parameters.
    I don't think empirical testing, without controls, is valid. I did a
    test and the histogram data seems to indicate a higher RGB value than
    the one you predicted.(Above the RGB 150,150,150 value). I think the
    key here is refelectance, not grey value(18% grey to be exact).
    I hope there is a Color scientist out there who is also a PS user.


    nikki, Mar 1, 2004
  15. a

    Warren Sarle Guest

    There is no single answer to this question because it all depends on
    the brightness range of the subject and what you want the image to
    look like. Pixel values around 140 are reasonable for typical
    low-contrast scenes, and in fact my 3 digital cameras also yield
    values around 140 for a subject with constant brightness.

    The gurus are very confused.
    Warren Sarle, Mar 1, 2004
  16. a

    Greg Guest

    Are you willing to provide details of your tests?
    Well, Chris Cox has already replied in our thread here, and his name is on
    the Photoshop
    author's list. I don't know whether he's a color scientist, though. ;^)

    Greg, Mar 1, 2004
  17. a

    LLutton Guest

    I've always figured 127, 127, 127 as the colors for a gray card. Maybe it
    varies with a different gamma setting but I use 220 which is the Windows
    default. If you're using a Mac, set it to 180.

    Try this:
    1. Select a rectangular shape
    2. Set the default foreground/background
    colors by typing a "D".
    3. Click on the Gradient tool and drag a
    horizontal line across the selection.
    4. Image > Adjustments > Posterize
    (use an uneven number, like 5).
    5. With the info box open, notice the
    numbers 127 appear in triplicate. If not,
    click on the eyedropper in the Info box
    and select "Actual Color".
    LLutton, Mar 2, 2004
  18. a

    Greg Guest

    The folks over in that web forum are very confused, IMHO. Not the gurus in
    this thread. ;^)

    Greg, Mar 2, 2004
  19. a

    Greg Guest

    There's no need to do all that. If you can accept that the L* value for
    middle grey is 50 (on a scale of 0 to 100),
    just use the CIE Color calculator (as I said earlier). Punch in 50,0,0 in
    the L*a*b* fields and you can see that the resulting pixel values in sRGB
    are 119,119,119. The L* scale is perceptually uniform - that's why half
    scale (50) represents middle grey, exactly. The fact that the resulting
    pixel values (in sRGB) are not *exactly*
    half scale just means that sRGB is not *exactly* perceptually uniform. In
    any case, 127,127,127, is less
    than a fifth of a stop difference from 119,119,119, so yes, 127,127,127 is
    almost middle grey.

    Greg, Mar 2, 2004
  20. a

    Greg Guest

    I repeated this test, using spot metering instead, to reduce the effect of
    the uneven lighting.
    The resulting pixel values in the centre of the image are now about:
    132,133,131. Camera is a
    consumer digicam (Nikon Coolpix 4300).

    Greg, Mar 2, 2004
    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.