Unclear behaviour when taking photos of a calibrated monitor

Home Forums General Discussion Unclear behaviour when taking photos of a calibrated monitor

Viewing 5 posts - 1 through 5 (of 5 total)
  • Author
    Posts
  • #2909

    Patek
    Participant
    • Offline

    Hi,

    I do not understand how my camcorder and my calibrated and profiled monitor interact.
    There is some strange behaviour and maybe someone can give me some directions for help.

    The colours of the films/photos of landscape, sky, etc. taken with my Sony HDR 810 camcorder have been not accurate (visually by my eyes).
    There might be two reasons: Maybe the camcorder is wrong, maybe my expectations are wrong.
    The camera is not calibrated.

    So, I made a simple test. In order to better controll the parameters, I took photos not of real life (sky, tree etc.)
    but of my easily controllable monitor.

    I displayed a white picture (rgb=255,255,255) on my monitor (60 Hertz, calibrated and profiled with xrite i1 display pro and displaycal).
    I manually set the white balance adjustment of the camera to this motive.
    I took a photo. F/4, 1/60 s, monitor screen out of focus to avoid Moiré. Cinetone and all other image functions of the camera were switched off.
    I analysed the photo in gimp with the colour histogram function:
    The result was rgb=164.3 164.8 163.3. (Each number is the average of the values of one channel.)
    The colour channels agree (quite closely) which is as expected, since the whitebalance was based on the motive.

    Now I took photos of the following pictures red, green, blue, yellow, magenta, cyan displayed on the monitor:
    255,0,0
    0,255,0
    0,0,255
    255,255,0
    255,0,255
    0,255,255.

    I expected that in the gimp analysis the “255” is simply replaced by “164” in the histograms. But it was not.
    Here are the results:

    255,0,0 became 184 0 4
    0,255,0 became 4 187 0
    0,0,255 became 17 0 241
    255,255,0 became 147 180 0
    255,0,255 became 163 0 239
    0,255,255 became 53 169 179.

    Apparently, the yellow motive looked very greenish in the photo.
    And the magenta was a lot on the blue side.

    These are massive deviations from the correct colour.
    So, maybe something was wrong with my setup. Or is the camera wrong?

    I am aware that the above motive is unusual in the sense that its spectrum has three rather
    narrow bands (red, green, blue) in contrast to real-world motives. But since white (255,255,255)
    is measured correctly by the Sony camera, why would the subset of colours (255,0,255) etc. be measured incorrectly?
    My understanding is that (255,255,255) on my calibrated monitor is equal to the combined output of (255,0,0), (0,255,0) and (0,0,255). So, even if the camera measures incorrectly, the output of the camera for (255,255,255) should be the sum of the outputs of the three colours r, g, b. At the very least, the blue component of the picture of (255,255,255) should not be less than the blue component of the picture of (0,0,255). But this does not hold as indicated above. So, I am puzzled.

    Is this somehow related to the fact that the profiling of LED screens with colorimeters requires special correction files?
    After all, the camera is similar in function to a colorimeter. Maybe the camera needs special corrections when taking photos of a monitor? In this case I would have to get a calibrated colour chart as a motive for my camera.

    Another thought I had was that the white balance of the camera does something wrong. It might be ok for (the chosen) white,
    but then it could be off for all other colours.

    Thanks for any suggestions.

    i1Display Pro on Amazon  
    Disclosure: As an Amazon Associate I earn from qualifying purchases.

    #2922

    Florian Höch
    Administrator
    • Offline

    Hi,

    Apparently, the yellow motive looked very greenish in the photo.
    And the magenta was a lot on the blue side.

    RGB values alone do not allow to draw any conclusions about the actual color unless the color space is know (e.g. sRGB, AdobeRGB, …). And if the camera is not profiled, the result is not defined anyway (a deterministic relationship of RGB to CIE values is missing).

    #2923

    Patek
    Participant
    • Offline

    And if the camera is not profiled, the result is not defined anyway (a deterministic relationship of RGB to CIE values is missing).

    Good point, true, thanks.

    I have now checked the documentation of the camera and the output is specified as: PAL color, CCIR standards.
    After a short search, I assume that this is the same as PAL (ITU-R BT.470-6). Which agrees with sRGB in red and blue but differs very slightly on green. (There are other PAL color standards which agree with sRGB exactly).

    So, theoretically, the camera outputs PAL color and the pictures should display correctly on my sRGB monitor.

    But I realise that the color balance setting has only three options: “indoor”, “outdoor” and “manual”. There is no “camera makes no changes to color” or “PAL” setting. Whatever setting I choose, the camera does something unspecified with the colors. This sounds crazy, but it looks to be true.

    My understanding is that the only option for me to get correct colors is to choose a fixed color balance setting (either “indoor” or “outdoor”) and then to profile the camera.

    What do you think about profiling a camera by a calibrated and profiled monitor? There is probably something wrong with this idea, otherwise someone would have come up with this. Instead, printed color target are always used.

    Thank you for taking the time for a reply!

    #2949

    Florian Höch
    Administrator
    • Offline

    What do you think about profiling a camera by a calibrated and profiled monitor?

    The (usually) peaky spectra of a monitor make it less suitable for use as a camera profiling target. A ColorChecker or similar reflective chart is usually a far better choice imo.

    #2963

    Patek
    Participant
    • Offline

    Thank you for your help!

    Kind regards

Viewing 5 posts - 1 through 5 (of 5 total)

You must be logged in to reply to this topic.

Log in or Register

Display Calibration and Characterization powered by ArgyllCMS