The histogram is an awkward visual metaphor invented by software Mathematician/Engineers to describe the tonal values of each color channel in a digital image. It is a noble attempt to visualize the dynamics of an image’s tones, even if it is a bit confusing. And, the ubiquitous histogram can actually be quite misleading.
We have been warned over-and-over again in Photoshop and Lightroom tutorials about the perils of “gappy” histograms produced by editing images in 8-bit depth. O, the pixel carnage! Think of all the data loss represented by histograms containing uneven tones and horrid gaps between them. We’re told how irresponsible it is to produce images files with gaps in their histograms. As if a smooth histogram is a somehow a healthy histogram, and the gappy histos provide evidence of fractured tonal ranges.
Here’s something to keep in mind about images. Detail is a product of contrast alone. Contrast is the dynamic difference between tonal values, with higher contrast delivering more dramatic detail. The more contrast between adjacent pixels, the more visual is the detail. And that, my friends, creates the gaps in a histogram. A “gap” in a histogram could mean that two side-by-side pixels differ in value by as little as one-third of one percentage point- virtually indistinguishable to the human eye.
Of course the point is always made that smooth gradients in photos (gradual changes in tones) will posterize and show banding. While this is absolutely true, the vast majority of photos will be unaffected by this slight irregularity. The two images seen above look pretty much identical except that the one on the left shows less detail than the one on the right. The other major difference though is that the image on the right displays less tones, shows gaps in its histogram, AND shows more detail.
When the contrast difference between tones reaches just 1%, a small gap shows in the (8-bit) histogram. Remember, histograms only give us crude feedback as to the relationship between adjacent tones. In English… there are 256 vertical lines in an image’s histogram, each occupying 1 pixel in width and representing one tone (1/3 of each 1% in value). That’s why a Photoshop histogram is exactly 256 pixels wide.
In a typical 8-bit image, histograms only display 256 levels of tone per color. And that means that those 12-bit image histograms are only showing you a small portion of the tone level status of the image. To actually show you a 12-bit histogram, the dialog box would have to be 55.5 inches wide! Naturally, we opt for the abridged version.
This is why you don’t see any fractures in the histogram when you operate in hi-bit depth. The fractures are there, they are just so insignificant that you can’t see them. The bottom line is this. Without contrast there is no detail, just mellow mush. Images that contain no gaps and spikes in their histograms probably lack both drama and detail in their appearance.
This is neither my 95 Thesis on histograms, nor is it a hill I’m willing to die on; it’s just a self reminder that an occasional gap between vertical bars in a histogram doesn’t mean the image has lost any data; it has simply reassigned those tones to another value for the sake of contrast in an attempt to deliver visual detail.
Thanks for joining me. See you next time. Herb
Click the Follow button at the top of the page so you don’t miss any future posts.
PS. if you have an iPad and are interested in learning more about the fundamentals of digital photography, I suggest that you take a look at my Accurate Color iBook in the iTunes Store