1

What does it mean when both RGB histogram and luminance histogram look similar?

Please explain in detali.

mattdm
  • 143,140
  • 52
  • 417
  • 741
wow
  • 89
  • 1
  • 5

2 Answers2

4

The human eye is more sensitive to green light than red or blue. For that reason, digital sensors have twice as many green photosites than red or blue. The overall luminance of an image, then, is more dependent on the green channel than red or blue. So the luminance histogram will look most like the green histogram. So if the dominant colours in the image are more green, and less red and blue, the overall RGB histogram will I guess look more like the luminance histogram.

Also if the image is very neutral (greys, rather than red, green or blue) then all the histograms ought to be about identical.

MikeW
  • 33,408
  • 10
  • 88
  • 123
  • 1
    Some of the newer sensor designs are leaving the strict 2X2 Bayer color filter array in favor of a more "randomized" distribution of Red, Green, and Blue pixels. The new Fuji X-100s uses a 6X6 non-repeating pattern that includes 20 Green, 8 Red, and 8 Blue filters. This minimizes moire caused by fixed patterns without the use of an anti-aliasing filter. – Michael C Apr 02 '13 at 07:58
  • 1
    @MichaelClark Is the point there just that some newer arrays aren't exactly twice as many green as blue or red? There's still more; the basic point stands. – mattdm Apr 02 '13 at 14:32
  • 1
    @MichaelClark that's interesting, clever design! – MikeW Apr 02 '13 at 17:38
  • @mattdm: I wasn't disputing the basic point, I was supporting it. In fact the new Fuji sensor has about 10% more green than a classic Bayer filter array. – Michael C Apr 02 '13 at 22:03
1

A luminance histogram is a summation of the various components of an RGB histogram. Luminance is the total amount of "light" in the scene where are Red, Green and Blue are the 3 colors that make up that "light". When you have the RGB histogram using an overlay (where they overlap each other), then the shape is going to be pretty similar. If it is displayed additively (where the colors are stacked one on top of the other to show their portion of the histogram), they are going to be identical. When they are shown separately, there is going to be potentially very little resemblance unless the image is largely grey-scale or largely one particular color (or made of two particular colors equally).

AJ Henderson
  • 34,864
  • 5
  • 54
  • 91