Split RGB histograms

Discussion in 'General Technical Discussion' started by Chiba, Nov 5, 2005.

  1. Chiba

    Chiba

    396
    Nov 3, 2005
    Ontario
    Can anyone help me understand how split RGB histograms can help a guy like me? :confused:
    Thanks in advance
    Chiba
     
  2. TOF guy

    TOF guy

    208
    Mar 11, 2005
    The composite histogram shows luminosity (on the Nikon D70 and I think on all Nikon dSLRs currently being sold). This is a weighted average of the RGB numbers which gives a higher weight to green than blue, and more weight to blue than red, reflecting the way we see brightness.

    So potentially, one could overflow one channel - in particular the red channel - and yet the histogram woud show no overflow: because the weighted average stays under 4096 (in raw).

    Separate RGB histograms allow to avoid this.

    There is another separate but related issue. Incandescent light, for instance, is weak on the red part of the spectrum. The red values after applying white balance coefficients may look like they fill the histogram as in a good "exposure to the right" , but in fact the channel is poorly exposed, resulting in poor details and noise. Iliah has proposed that the problem can be detected by setting WB so that all channels are multiplied by 1. Then one can evaluate how poorly exposed is the red channel and apply corrective measures, like using a filter (a grey card must be also shot to evaluate the true WB coefficients and apply them in post-processing). Another situation would be one where the lighting corresponds to WB coefficients which multiply one channel by a value higher than one and it looks like this channel (but not the others) is "exposed to the right". In fact there is some additional "room" to increase exposure without overblowing that channel. Again using a fictive WB with all WB coefficients equal to 1 helps diagnosing the problem.

    But to be able to do the analysis as described above you first need separate RGB histograms.

    Thierry
     
  3. TOF guy

    TOF guy

    208
    Mar 11, 2005
    Off topic. But from what I've read Canon dSLRs show composite histograms calculated from a different image than the one that is displayed, one that is obtained by always applying a specific and constant set of WB coefficients to all pictures (either daylight or incandescent depending on the model). Can somebody confirm this (just out of curiosity ?).

    Thierry
     
  4. Chiba

    Chiba

    396
    Nov 3, 2005
    Ontario
    interesting.

    So I guess I should leave the normal histogram review turned off and only refer to the separate rgb histograms.. Have to admit I'm a bit confused by the last bit but for important shots I'll usually use an Expodisc to get the proper white balance. I'm big on using the blown highlights mode to make sure I push the exposure as far as I can.. I think I need to do a bit more research to understand histograms further.
    My current understanding is that the left side is black the right side is white and middle is grey the height of the lines determines how many pixels in that particular shade so if you have a peak on the far right you have over exposed and left vise versa..

    Thx for the response
    Chiba
     
  5. cwilt

    cwilt

    Apr 24, 2005
    Denver, CO
    The 3 channel histogram display shows exposure with white balance applied. White balance is the multiplication or division of the red and blue channels. So you are not seeing the true exposure of all 3 channels and that could lead to over or under exposure of the red and blue channels. This leads to loss of detail in highlights due to clipping or increased noise from under exposure.
     
  6. Chris101

    Chris101

    Feb 2, 2005
    Arizona
    oops, actually incandescent light is nearly all red light and is extremely blue deficient. The idea is the same though. E^R with a mixed histogram will result in a blown red channel under incandescent light.

    That is why Iliah's idea of always exposing at ... I believe he said 5300K, will result in all channels getting the same amount of amplification. Then adjust the white balance to get the correct colors after the fact during RAW conversion.
     
  7. TOF guy

    TOF guy

    208
    Mar 11, 2005
    Indeed.

    Iliah has also posted a file that allows to set all WB coefficients to 1.

    Thierry
     
  8. cwilt

    cwilt

    Apr 24, 2005
    Denver, CO
    I think Iliah was suggesting this method should be used to adjust exposure then shoot with a correct white balance. Perhaps he will respond.:smile:
     
Loading...
Similar Threads Forum Date
Split focusing screens for DSLRs (Katz eye) General Technical Discussion Mar 22, 2012
Histograms: RGB vs. Luminance General Technical Discussion Apr 24, 2011
sRBG or Adobe RGB General Technical Discussion Sep 25, 2009
Totally stupid question RGB vs sRGB General Technical Discussion Sep 25, 2008
D-80 - sRGB -Adobe RGB issues General Technical Discussion Dec 1, 2007