A PRACTICAL GUIDE TO INTERPRETING RGB HISTOGRAMS

 Learn how to interpret your digital camera's thumbnail image histogram to help determine proper in camera exposure
Learn how to adjust tonal range in your scans or existing digital images using the levels-histogram tool
PAGE 2

HISTOGRAM AND RGB COLOR AND TONAL RANGE EXPLANATION - Part 2
There are three different kinds of histograms. The most commonly used is the RGB histogram. The RGB histogram is a composite graph of the tonal values for each color channel RED, GREEN and BLUE. Another type of histogram is a Luminance histogram. The luminance histogram is basically an RGB composite histogram that takes into consideration the human eye's greater sensitivity to green, then red and lastly blue. This type of histogram is also sometimes called a 'brightness' histogram because it displays a 'compensated brightness' range. The luminance histogram gives you the best graphical representation of the visual brightness and contrast of your image. You can also view each individual color channel's tonality by viewing an individual channel histogram for each color RED, GREEN and BLUE.

Some imaging programs allow you to see only the RGB or only the luminance histogram. Photoshop's Levels tool will allow you to view and adjust a master RGB histogram or adjust each color channel individually while viewing the histogram for the selected color channel. Photoshop 7's 'image / histogram' shows a histogram palette with choices to view RGB, R, G, or B and luminance histograms for the open image. Photoshop CS has a floating histogram palette that lets you select to view each type of histogram and an option to view the histogram for each color channel and a luminance or RGB histogram all on the same palette simultaneously. Photoshop CS also has an overlay histogram of the separate color channels that Adobe calls the 'colors' histogram. The Photoshop CS histogram palette also shows the effects of your image adjustments with changes to the histogram graph in 'real time'.

Your digital camera may show either a luminance or an RGB histogram if it has a histogram view option available during image review. Either type of histogram will work to help you determine proper 'in camera' exposure. However, it would be beneficial to know which type of histogram your camera is using. Sometimes your camera's documentation will tell you which type of histogram is used in the image playback/review. If your camera's documentation doesn't tell you which type of histogram it uses, it is fairly easy to determine which type of histogram your camera is displaying. Take a picture with your camera. Download the image to your computer and open it in your image editing program. Compare the RGB and luminance histogram of that photo in your imaging program to the same photo's 'in camera' histogram.

Since a luminance histogram has had some perceptual compensation (unbalanced averaging) applied, it may not always show you if you have pixels in the 0 or 255 tonal area. Not having the full 3 channel RGB histogram can make interpreting 'in camera' histograms a bit harder. We will discuss methods for interpreting digital camera thumbnail image histograms a bit later.

An RGB histogram is probably the best histogram to use for adjusting scans or existing digital images because it lets you know how many, if any, pixels are in the 0 and 255 range. The "bookends" (0 and 255) of the histogram are very important. Even with high key and low key images you usually don't want to see a lot of pixels climbing up either the left or right margin of the graph. There really isn't very much noticeable texture (color) in an image when all three color numbers are above 250. On the other end of the graph there really isn't much detail or texture in the shadow areas when all the color numbers are under 20. See the examples below.

If your monitor is adjusted correctly, you should see a slight difference between 95 and 100 and 0 and 5 in the grayscale pattern below my color number samples. If this is not the case, click here to go to a very simple monitor calibration adjustment routine. Your monitor must be running true color at either 24 or 32 bit depth to view photographs properly. On my monitor I can see a very slight difference between 20R,G&B and the pure black background on my 'black' color number example image. My Epson 2200 printer will also resolve this difference to a barely noticeable degree too. The high key color number samples are more discernable with a good differentiation between the sample disks and the pure white background all the way up to 250R,G&B. I see the same results on prints from my Epson 2200 for the high key samples.

There may be instances where your darkest shadows have to go pure black to maintain proper exposure in the rest of your image. In cases like this it would be acceptable to have the pixels representing the darkest shadows climbing up the 'black' or left side of the graph. It would be a very rare photographic situation where the image content would justify a lot of pixels climbing up the 'white' or right side of the graph. Studio lighting where you have strong background lights to 'washout' your white background would be one example.

We will proceed by having a look at some example images and their corresponding histograms. In the first example below we have a photo of Font's Point in Anza Borrego Desert State Park. We also have included the different histograms for this image along with the extra 'colors' histogram that Photoshop CS makes available. The colors histogram is an exact overlay of the three separate color channels and shows a colored graphical distribution of all of the colors RGB and the CMYK colors where the tonal range for the RGB colors overlap.

The histograms below this desert photograph and all the rest of the histograms shown on this page are screen captures of the Photoshop CS histograms from the original same resolution 16 bit TIFF images.

I used the color picker or 'eyedropper' tool to read the pixel values for a highlight, midtone and shadow area and made an arrow to the corresponding area of the RGB histogram graph. The highlight cloud area color numbers indicate that there is no unwanted color cast in this image. The white clouds have a very slight blue cast but that seems very reasonable for thin clouds against a blue sky.

The luminance histogram is very close in shape to the green histogram. This reflects the heavy bias toward green referred to in the luminance histogram explanation in the first paragraph of this page.

The original tonal range in this desert scene is mostly in the midtone area. There are very few pixels in the 0-30 range and very few in the 240-255 range. While this image looks decent on a computer monitor, it would probably look a little flat when printed. We will adjust this image a bit later using a 'levels' tool.
This next example is an atypical image but serves well to demonstrate how RGB image histograms are constructed. It is a shot of a full moon at twilight. I have sampled pixels from two places on the moon with one sample from the lightest area and one sample from the darkest area. I have also sampled one area of sky.

You can see the individual colors on the histogram by following the arrows from the sky sample B. The blue channel is further up the brightness scale on the histogram due to blue's slight predominance (higher color number) in the sky color. Note the area between D & E on the RGB histogram corresponds to the tonality of the moon area.

As plain as this image is the pixel samples tell us all we need to know about this image. For instance in the C sample moon highlight area the pixel values tell us that we have close to neutral gray but a very slight yellow cast due to the blue channel's lower tonal number. This may be because of a little smoke in the air from recent fires or it may be that the RAW converter or digital camera just missed the white balance by a little bit. Or it may be, as the old nursery rhyme goes, that the moon really is made out of cheese, thus the slight yellow tint...:^)

The B sample sky area makes a decent dark sky tone with its predominance of blue pixels with slight cyan influence from the Green channel's 64 tonal number (blue + green = cyan). Since red is the complement of cyan the red channel's 55 tonal number controls amount of cyan influence. As a demonstration we'll make some 'what if' scenarios. First, lets move the red to 64. Now we have yellow (R64+G64) as the second highest dominance. Since yellow is the complement of blue (B84). Our sky turns to a pure dark blue. If we reduced the red influence, the sky would turn more cyan. If we increased the red color number beyond 64, the sky would have a slight magenta cast since red would then be our second dominant color and red + blue makes magenta.
In this example of windfall apples in Julian California I've sampled the colors in a warmish 'yellow' area on one of the apples in the foreground. The color coded arrows point to the tonal area represented on the appropriate color channel histogram.

With the sampled color numbers of 240R, 210G,105B we should expect to see a light reddish yellow since red is the predominant color at 240 with green not too far behind. In RGB color red + green makes yellow. Since blue is the complement of yellow, the 105B influences the overall brightness and hue of the dominant reddish yellow color of this apple. A lower blue number would darken and intensify the yellow color of this apple. A higher blue number, up to 210B, would reduce the influence of the green channel and at 210B cyan would be our second dominant color since 210G+210B = cyan, which is the complement of red. With red as dominant and cyan as the second influential color the yellow cast would disappear and the color of the apple would be a very light red. 240R would still be the dominant color but the new cyan influence would reduce the intensity of  red.
This photograph was shot late in the day on 4X5 Velvia, thus the warm colors. You can see a verification of this by looking at the blue histogram. The blue histogram is quite to the left of center, which confirms our visual of the dominance of warm tones in this image.

In the next section we will have a look at the 'levels' tool in Photoshop. The levels tool is a very good tool for adjusting tonality in digital images. This tool is available in almost every image editing program.

NEXT

LINKS:
Histogram and RGB Tonal Range Explanation - Page 1
Digital Image Tonal Range Adjustment Using a Levels Tool
Interpreting Your Digital Camera's Thumbnail Image and Histogram

Web Site Links