I know. Its been done. Beaten to death, in fact. But remember my “mission” for my tutorials. I have some friends who have asked me to try to translate this stuff into plain English. My tutorials will never be a substitue for the in-depth, detailed and knowledgeable work already out there. And, maybe I will even add a point of view or insights that some writers have missed or even in some instances, mislead about the usefulness of the histogram tool.
Now that I have you reading at least this far, I’ll admit, there is no such thing. I read many comments and texts extolling the virtues of the histogram tool on the DSLR camera back; as well as the “blinkies” (blinking white highlights indicator). Often, the commentator warns that if there are “any” blinkies, you must dial exposure back until they disappear, and if there is any part of the histogram which spikes and pushes up against either end of the graph, dial exposure up or down until the histogram stays tamely between the goalposts.
To which I say, not so fast! The histogram and the blinking highlights indicator are tools. As such there is no “do-all tool.” Nor is there a set of hard and fast rules which govern the absolute use of the tools.
So What Does a Histogram Do For Me?
It might surprise some digital photographers that the word histogram is not some new digital photography concept. A histogram is a frequency distribution chart. The histogram we use on the back of modern DSLR cameras and in software programs like PhotoShop, is a graphical depiction of the distribution of pixels from black to white, with pure blacks represented at the far left side of the graph and pure whites at the far right. This might lead one to conclude that the “perfect histogram” would always be represented by a symmetrical shaped, such as the histogram shown here.
But in many cases, they would be wrong.
The histogram measures the relative number of pixels at each level of brightness. So it stands to reason that a properly exposed image that has lots of bright tones in it (for example a snow scene, a water scene with lots of specular highlights, or a “high key” fashion image) will naturally display more of the histogram shape to the right. Likewise, a night shot, or shot with a lot of dark tones would shift the bulk of the histogram to the left. A neutral toned image should come closer to the bell curve. The Histogram for the photo of Auke Bay looks very different, and it should, as the vast majority of pixels are very light tones:
Until recently, the histogram on the back of most cameras only showed us its measurement of the B&W pixels in an image. That was/is o.k. In exposure, what we are trying to measure is the relative brightness in an image (or its luminance). Newer cameras have overlays of red, green and blue. For our purposes here, we will ignore them (though you should know that the same principles apply, so that if any of them “spike” at either extreme, the histogram is suggesting that the pixels in that color have been “clipped”).
I use the word “suggest” for a reason.
The histogram and the blinking highlights indicator are imprecise measuring tools. There are at least two important reasons for that.
First, there area times when the image should have portions that these tools will indicate are overexposed. For example, there are, at times, images that have specular highlights an/or very bright whites that should be pure white. In this case, the properly exposed image may actually have some small amount of “clipping” (blinking highlights and a very narrow spike at the right extreme of the graph).
In my earlier tutorial, “Getting Exposure Right” I said that photographer should not be fooled into thinking that the light meter in a camera can “think” for her. The histogram is similar in that regard. While it differs in that it can measure the overall image exposure before, during and after the fact, it is the same principle. The photographer must know the overall characteristics of the image and exposure he seeks. Only by accomplishing that, can the value of the histogram (and the “blinkies”) be appreciated as the very powerful tool that it is.
The second reason applies to raw images (and I hope you are shooting raw images). The histogram on current DSLR cameras is created, on-the-fly, by the same in-camera software that renders a jpeg image (if you shoot jpeg). As I opined in “Why You Should Shoot Raw” it doesn’t make sense to go to the expense of a multi-thousand dollar camera, an expensive computer and sophisticated rendering software such as Photoshop, and then let rather unsophisticated in-camera software render your images (“cook” them) in such a way that you have very little ability to “develop” them yourself. But that is just what the in-camera jpg algorithm does.
It is the in-camera created jpg that the histogram (and the blinking highlights indicator) is measuring; not the raw image.
And, notably, this jpeg-based histogram is not the same histogram you get in Photoshop, ACR, Lightroom, or any other post processing software’s adjustments menu! So it is very important to understand that your particular histogram my not completely accurately measure what your sensor is able to capture in a raw image. Like the exposure meter, it is worth doing some testing in a controlled lighting and image environment, taking note of it and making the appropriate adjustment or interpretation in the field.
Your Mileage May Vary.
I have found that on some cameras (true of my D200 –seemingly less so of my D700, which is perhaps a function of the camera companies just getting better and better at interpreting, measuring and presenting what the end result will be), I can push the image beyond the blinkies and spike and still get a very acceptable “default” histogram when I open it in ACR. As we like to say on forums, YMMV.
So What Am I supposed To Do With The Histogram?
Within these limitations it is important to understand the guidelines. The left and right extremes represent the limits between which the histogram should normally reside. If spikes upward at either end (left or right sides), the histogram is suggesting that the respective black or white points are being clipped. Another way of saying this is that the spike and beyond represents areas in the photo in which all detail has been lost in the dark areas, and/or all detail has been lost in the whites (“blocked up shadows” or “blown highlights”). Importantly, these are guidelines; not rules to be slavishly adhered to.
Now that we know the limitations of the tool and how it works, we can use the histogram to make arguably better judgments about exposure. But remember, they are just that; judgments. Once you understand how the tool works and what the information it gives us means, it just makes sense that the histogram for a shot with mostly darker tones would show substantially more information to the left of the histogram:
Prior to the advent of this tool most of us only had the light meter and our knowledge of it and of exposure techniques as tools to be used prior to making the exposure. While in most cases, accomplished photographers knew how to meter and judge correctly and rarely worried about the end result. In cases where tricky lighting or other conditions prevailed, photographers often either bracketed a series of images to cover any overlap from measurement to reality, or in some cases, took Polaroid shots that could be “instantly” developed and analyzed. And in the end, you still didn’t really know until you got the developed images back from the processor.
In my view, the time and effort required to get to that point was clearly greater than it is with the modern DSLR and the histogram. Today (especially when capturing raw images), we can let the camera’s automatic metering make the exposure and check the histogram, adjust and then take another exposure. And newer cameras with so-called “live view” lcd panels can actually display a “real-time” histogram, which can be used as an additional measuring tool.
I am not suggesting that the histogram be a substitute for knowing how to properly expose, and pre-thinking your shots.
You still need to understand the principles of good exposure before you can use any tool to make well-exposed images, in my view. And, my “shoot first and ask questions (well, check the histogram) later” approach will not work if you are shooting action or in rapidly changing light conditions (although in an action scenario, such as sports, or in a situation where you are anticipating future action, you can take a test exposure and get the adjusting done before the action starts in many cases). But for my routine landscape imagery, it is a method that works just fine. Except in rare or unusual cases, I rarely use the spot metering function on my camera, and never have used (in my recent memory) the weighted average function. I set it to the “matrix” (honeycomb pattern) meter, take the exposure, and check the histogram. For good measure I also check the blinking highlights indicator, just to get more information. In certain lighting situations, I know my particular camera may meter at a certain exposure solution, but I need to adjust the exposure compensation by 1/3 or 2/3 stop.
Raw Shooters: Expose To The Right
One last point; relevant if you shoot raw images, but not so much if you shoot jpeg. See my previous tutorial, “Expose Right To Expose Correctly,” for an explanation about why it is important to favor, or bias the right side of the histogram in exposing raw images. In short, the vast majority of information is captured in the highest EV of capture (twice the amount of the entire rest of the perhaps 5 EV range). Thus, the sensor can capture more important information in the highlights. It is important not to leave any gaps on the right side of the histogram, and if the photographer must choose or make judgments, the right side should be favored.
For a really in-depth, illustrated tutorial on Histograms, see The Luminous Landscape here.