Today’s Question: What does the histogram vertical axis indicate? Is it number of pixels in a certain range or “bucket” of luminance values? What is the peak value; what number of what?
Tim’s Quick Answer: The vertical axis of the histogram shows the relative (not absolute) number of pixels in an image for each tonal value represented by the horizontal axis.
More Detail: While the general concept of the histogram is relatively straightforward, there are a couple of factors that can cause it to be a little confusing.
First off, the horizontal axis represents tonal values in the image, with black at the far left and white at the far right. However, not all tonal values are actually represented. In most cases, the horizontal axis for the histogram is calculated based on only 256 tonal values from black to white. This represents 8-bits per channel for a single channel, effectively meaning that the histogram is calculated as though the image was in black and white and 8-bits per channel even if it is a full-color 16-bit per channel image.
Once the image data is simplified in this way to calculate the shape of the histogram, the pixels are tallied to determine how many pixels in the image represent each tonal value. That number determines the height of the histogram chart on the vertical axis for each tonal value of the horizontal axis. You can think of the histogram in this context as being a bar chart where there are 256 vertical bars along the horizontal axis, with the height of each of those bars representing the number of pixels at each tonal value.
However, the data is manipulated a bit further to improve the overall display of the histogram chart. The chart is generally scaled so that whatever the highest point on the histogram chart is will go to the top of the vertical scale. In other words, the top of the vertical axis can represent a significantly different number of pixels in the image depending on the overall distribution of tonal values in the image.
For example, if you view the histogram for an image that is mostly comprised of very bright pixels, the right end of the histogram will be very tall, and the rest of the histogram will be very low. For an image with a relatively even distribution of tonal values throughout the full range, the entire histogram will be relatively tall. In both cases, the tallest portion of the histogram will extend to the top of the vertical axis.
Of course, different software will use different methods for calculating the histogram display, but the basic concepts outlined above are relatively universal. And of course, if you’re displaying a full-color histogram, the same concepts apply with the difference being that three individual histogram charts are generated for each of the three color channels and then overlaid on top of each other.
