For More Detail Check the Histogram

by Ellen Anon

One of the most useful advantages of the digital evolution has been the histogram, but surprisingly few people know how to interpret them. Simply put, a histogram is a graph of the tonal values in an image - whatever they are, ranging from black to white. For convenience sake we speak of histograms as having 256 values from 0 on the left for pure black to 255 on the right for pure white. (These numbers were based on the number of tonal values available in each channel in an 8 bit image, which is 2 to the 8th. Even though 16 bit images have considerably more values, it's easier to conceptualize and interpret a histogram while considering it to have 256 potential values.)

A spike on either end, particularly when the graph looks like a mountain that was abruptly sliced off, indicates there is clipping. This means that there is a group of pixels that are pure black or pure white without detail, as long as the histogram is a luminosity histogram.

The catch is that some cameras and software use luminosity histograms, some RGB histograms combined into a single set of data, and others RGB histograms that show each individual channel superimposed upon one another. In Aperture you have the choice of a multi channel RGB histogram, a luminosity histogram, and individual channel histograms.

Choos Histogram.jpg

Which is best? If you are outputting to print and your main concern is to ensure that there are no areas of your image that are pure white (where the paper itself would show through, showing whatever shade of white it is), then the luminosity histogram is the most efficient.


2007-07-12 06:53:00
Important to note is that the luminosity histogram is corrected by using contrast and brightness sliders (exposure, curves, levels...) while the RGB histogram needs correction in the saturation of the specific color. Afterall the RGB histogram doesn't correspond to the luminosity but to the saturation.
Correcting a single RGB channel for brightness (like you'd do for luminosity histogram) will result in color shift / change of hue.
2007-07-12 15:02:56
I'd love to read your article "Improving your photography through the histogram" but it seems that we need to pay to read it :-(
2007-07-12 23:15:46
how do you do the 3D effect on the watermark?
Ellen Anon
2007-07-13 12:44:24
I didn't realize the article was limited to paying members of the AUPN site - sorry.

Hendrik, sometimes adjusting the saturation is helpful when there's clipping in the RGB histogram, but sometimes exposure/tonal adjustments will also help. It's similar to the difference between a perceptual and a relative colorimetric adjustment for out of gamut colors.

As for the watermark, I type it in in Photoshop, then add a basic embossing to it using layer styles, and reduce the fill to zero. I often slightly reduce the opacity of the layer as well. I haven't tried setting it as a watermark in Aperture, but I think it should work.