Skin Tone. Neutral Target Sample Image.

Discussion in 'Digital Darkroom' started by eddie g, Dec 26, 2005.

  1. Other than the image most folks know about...the one with the four
    children at the bottom and the hand holding the martini shaker,
    where can I download an image file that can be used as a target for
    neutral skin tones. If you find yourself wanting to talk about color
    profiles, color space, digital, computers, monitors, printers,
    etcetera, then I didn't do a good job at asking my question....and I
    appologize for that. I'm trying to adjust the colors in photoshop
    CS2 such that the ratio of RGB values are close to what happens in
    the real world naturally, rather than in the wonderful world of
    photography. Computers and digital equipment have nothing to do
    with the real thing. This has nothing to do with personal taste.
    If you took ideal balanced white light (eg. from the sun) and
    measured the ratio of RGB values reflected from the face of 50
    average white skin caucasians, you would obtain the numbers I'm
    looking for. For example, it may be something like; R/G=1.28,
    R/B=1.55, G/B=1.20. Remember, this has nothing to do with
    photography, cameras, computers, digital, color space, nothing of
    the sort. These are ratios that existed two thousand years ago.
    While every caucasian in that group of fifty has diferent ratios,
    the average number will not vary by anything to speak of. If you
    took two diferent groups of 50 you might get a tiny variance where
    one group might have a R/G ratio of 1.28 and the other might be
    1.27. The measurement is taken from a reading of how much
    red/Green/Blue is reflected off of the face/skin of fifty caucasians
    in balanced white light.
    Sorry to get off track...back to the main question of this post. Do
    you know of a target image of skin-tones that is recognized by the
    photographer comunity or can be regarded as a standard that other
    images can be compared against when it comes to skin-tones being
    neutral.
     
  2. Sorry but this all has to do with color spaces and numbers and the fact that the human
    visional system doesn?t work anything like the way a digital camera or computer encoding
    system works.

    Until you define the color space, we can?t even start to talk about numeric values.
     
  3. You can probably come close in this quest using one of several Color Checker charts by Gretag-MacBeth. The charts come with an RGB reflectance scale, and have from two to a dozen so-called "standard" flesh tints.

    From your description, you seem to want the numbers, not necessarily a chart or image file, so that you can adjust an image "by the numbers" to achieve a perfect balance around flesh tones.

    Press operators have used "standard" ratios for flesh tones for years, in lieu of an objective method. Of course, different inks, papers, humidity, process temperatures (on and on) have an effect. Then again, which face or part of a face would you sample to set the "perfect" color balance. Any one face might have a dozen "shades", multiplied by the number of faces to balance. I've seen the numbers, but I don't remember what or when. The practicality of their use is too low to commit to memory, for obvious reasons.

    The ultimate "average" color is neutral grey. That's not such a bad way to set the mid-range color balance, if you have something in an image, either a test chart or by serendipity, to measure. Invent any RGB number you want, just make them all the same.

    The practical solution (other than a grey card) is to use one of the Color Checker charts to create a profile. That gives you linearity over a range of colors, with an emphasis on flesh tones. But you won't do that - what do I know?
     
  4. Eddie,

    There was a holy grail sort of target depicting the perfect
    fleshtones and its corresponding print made by Kodak for
    wedding photographers I came across a while back in another
    forum. But when I saw the linked digital image of what the poster
    considered perfect fleshtones, I gave up on the pursuit. It
    depicted a wedding couple with way too maroonish/magenta-ish
    looking fleshtones for my tastes. And this target/print was
    expensive as well.

    There is no such thing as a neutral fleshtone because of all the
    varying changing conditions involved with human perception of
    color and neutrality. The four children PDI target you mentioned, I
    use only to check white point calibration for the appearance of
    varying levels of yellow which can be seen in all fleshtones no
    matter the lighting conditions, gender or race unless the subject
    is covered in pancaked makeup. I'm a pasty caucasion and I still
    see some yellow in bright daylight in my own fleshtone.

    The gases in some artificial lights bouncing off at certain angles
    as it enters the lens like in flash units at close range can in
    some instances throw off camera sensors into reducing this
    yellow (by adding too much blue) where you end up with the
    magenta-ish fleshtone. I've seen this on some Nikon sample
    images on the web. Check the tutorial on skintone correction on
    LonestarDigital.com to see what I mean.

    This site shows how even when using Auto white balance and
    applying neutral RGB numbers to white can throw off skintone:

    http://www.sphoto.com/techinfo/rawconverters/pages/whitebalan
    ce2.htm.......Take the space out between n & c in the word
    balance.

    In short just go for what looks good on a properly calibrated
    neutral looking monitor.
     
  5. My appologies again, Andrew. Two thousand years ago, before color space was invented, light had a property that made it reflectable by the skin on a white mans face. Even back then, the light which was being reflected off of the skin contained a specific ratio of Red to Green as well as other colors. This ratio does not depend on the subjective eye of the human. These ratios varied from person to person and even from forehead to nose to cheek of the same person. Another thing that existed long before computers and color spaces is an average of this ratio found in white faces. So if a tree falls in a forest and no one is there to hear it, does it make a sound? Well, if no one is looking at you, does your face reflect light? And if no one is looking at you and no one takes a picture, does this light which is being reflected by your face have a specific ratio of colors?
     
  6. Maybe I should not use the word neutral because maybe the color of skin is not neutral. If I understand the word neutral, it would be equal amounts of R/G/B, right?
     
  7. "...If I understand the word neutral, it would be equal amounts of R/G/B, right?..."

    No, not always. As previous posters have discussed, it's not that simple.

    Just as a test, create a Photoshop image with equal amounts of RGB, and compare how it looks on different monitors (side by side), printed on different inkjet and laser printers, as well as offset.

    Something tells me all those "neutrals" will look different.
     
  8. What's neutral? Color scientists have been grappling with this
    for years as I've found out from this wikipedia page:


    http://en.wikipedia.org/wiki/White_point


    I just got hooked on wikipedia yesterday. This site is like a
    magnet of info surprisingly quite a bit plumbed from the web.
    How reliable that is anyone's guess.
     
  9. > My appologies again, Andrew. Two thousand years ago, before color space was
    > invented, light had a property that made it reflectable by the skin on a white
    > mans face.

    Color perception is a phenomena that happens in our brain.

    I'm not sure what your point is and I'm not sure what you're looking for here.

    Computers can only handle 1 and zeros. If you are asking for some numeric values that
    define a color, you are going to have to provide the color space you want those numbers
    in. Numbers along do not provide enough information to define a color appearance. That
    is why R255 is not the same in Adobe RGB (1998) as sRGB or any other color space.

    I am not sure how to provide you a color ratio without numbers! And I certainly can not
    provide you anything that is defined on a computer without numbers. That is all they
    understand. Lastly, the human visual system as I said does not operate in a linear fashion
    thankfully or our heads would explode when we went from a dimly Lite room to full
    sunlight.
     
  10. The H7 patch on a Color Checker SG chart looks like a "normal" Caucasian skin tone to me. Your mileage may vary - and there are 13 other "skin tone" patches to choose from. The RGB reflectivity for the H7 patch is 201/145/126 respectively on a scale of 0 to 256. The source of illumination is D50 (5000K) and the acceptance angle is 2 degrees. On the same chart and same illumination, neutral grey is 111/111/111 (within a few tenths).

    This begs the question - what is red/green/blue? Certainly not discrete wavelengths, but each is a band of wavelengths with a certain distribution. One would expect peaks and valleys in both the source and reflectivity, especially reflectivity of a biological nature. My Eye One Pro takes 41 measurements from 380nm to 730nm in 10nm increments.

    Forget the Empiricist kaka about forests, trees and perception.
     
  11. -->The RGB reflectivity for the H7 patch is 201/145/126
    respectively on a scale of 0 to 256.

    In WHAT color space? Again, the numbers are meaningless alone.
     
  12. The color space is not specified in the reference file, nor is there there an embedded tag in the TIF file. It's all meaningless, just like this question....aaaagh!

    AFIK, the values are absolute reflectance on a 0 to 255 scale. The white patches are assigned RGB values of about 243/245/242, black patchs are about 024/025/025. I don't know what reflectance standards are used, nor the shape of the bandpass for RGB. When I open these files for printing, I select "not color managed", or something to that effect, and print without CMS. Profiles are independent of color space, being instead curves which give the best fit of observed measurements to the reference data. Everything is relative, I guess (take that Descartes, I'm with Einstein). Gretag-MacBeth seems to know what to do and how to do it - that's why we pay them big bucks to get something that works.
     
  13. And besides if you went with some number formula, all your
    fleshtones would look the same. Check these pro
    photographer's sites and how they capture fleshtones. You
    couldn't keep up with a formula.


    http://www.onemodelplace.com/member.cfm/P_ID/89799

    http://www.melvinsokolsky.com/vogue2002_gallery.html

    http://www.neilsnape.com/
     
  14. Here is a scan of GretagMacbeth's Digital ColorChecker SG card, that I have neutralized to daylight viewing and then placed in the Adobe RGB color space. Open it in Photoshop and leave it in Adobe RGB so that you can use it as a reference. The skintones can be found in the patches D7/8 to J7/8. Patch E2 is the average of all dark skintones and and patch F2 is the average of all light skintones. Hope this helps. Otherwise never mind and paint your folks whatever color suits them.
    00Edc0-27156284.jpg
     
  15. Here's a skin tone chart.
    00Edcr-27156384.jpg
     
  16. Here's a hair color chart.
    00Edct-27156484.jpg
     
  17. What is 'caucasian'?

    As I understand the term (as a European) it would include the peoples of Greece, Spain, Italy,
    Britain, Sweden, Finland etc... Quite a range of skin-tones.

    What is the point of averaging them out? What would this colour really represent?
     
  18. Chris,

    I recognize that skintone chart. I forgot I had on my hard drive.

    Just a question about that linked GMB CC chart, why are the
    midgrays on the green side. The reason I ask is I'm noticing a
    trend toward a green neutrality in daylight captures even in
    digicam shots of everyday white objects. Is this a color space
    transform interpretation I wonder or is that gray on the chart really
    greenish?
     
  19. well, even though I lacked the knowledge to ask the right question, I did learn a lot. Thanks to all of you !
     
  20. You did ask the right question. Viewers tend to be more sensitive to and critical about the accuracy of skin tones than with other colors in an image, especially if they know the people in the images.

    Books by Eismann and Margulis do provide color numbers in different color spaces for the range of ethnic skin tones. But these numbers (and the charts in this thread) are only accurate when the skin is under direct and natural lighting. They should only be used as the starting references if a person is not under direct and natural lighting. For example, a face under outdoor bright daylight will have a set of numbers that look right. But these numbers will be totally wrong for the same face under indoor dim candle light.

    It is much easier to get the skin tones right in a studio, as those in the referred sites in this thread. The lighting is under control and known, and a gray card can be included in the shot as a reference.
     

Share This Page