Primary Color Correction: Tonal Correction Tools - The Art and Technique of Digital Color Correction - page 47

Histograms
A histogram is a very simple graph. Horizontally across the x axis (redundant, I know, but I'm trying to be clear)
indicates the image's tonal range with black to the left and white to the right. The vertical axis, or y axis, shows the
number of pixels at each tonal value. Most histograms of a well-exposed image will look somewhat like a bell curve,
with some pixels at absolute black and some at absolute white, and then most of the pixels will be in the midrange
of the picture. There are some beautiful images that may not look like a bell curve, but if you have a histogram that
looks like a bell curve, it should be a pretty well-exposed image.
When analyzing an image on a histogram, though, the bell curve shape is less important than identifying the
danger signs of a histogram: sharp peaks or “cliffs” at either end of the histogram. These indicate that clipping is
occurring. So you can use this “cliff” effect to help you judge when your correction starts to clip.
Look at the following example, which uses the “Steve dark at the park” clip from the DVD.
In the Master tab, at default, the numbers under each histogram are the same (16, 128, and 235); see Figure 2.15 .
This indicates that black (16 in NTSC levels for a 7.5IRE setup) on the source is mapped to black on the output side.
Fig. 2.15
Now, on the input side, slowly move the black triangle to the right, releasing when the source black side reads
25 ( Figure 2.16 ).
Fig. 2.16
Now the slope on the output side has a nice clean incline, from the black triangle to the peak, but if you slowly
move the input slider up from 25 through 30 and 40, you will see that the output side starts to have a significant
spike of pixels above the black triangle.