Photography -- Being Comfortable With Your Histogram

BEING COMFORTABLE WITH YOUR HIST OGRAM: Not a Gynecological Concept

If you’ve used a digital camera at all, you’ve already discovered that the LCD on the back of the camera is fairly useless as an aid in judging exposure. Its brightness varies considerably with ambient light, and in some cases, user-controlled settings. Its optical quality is minimal, and it lacks tonal detail, displaying a very contrasty image.

Fortunately, camera manufacturers are aware of this (yes, sometimes they do pay attention…) and they provide you with an alternative, which actually does a great job of helping you judge exposure— The Histogram. Highlights, shadows, mid-tone, tonalities, dark, light, high key, low key, black, white–all are terms that describe in words what the histogram shows pictorially. It is essentially a simple graph that charts the tonal values in your image.

The position of tones along the horizontal X-axis tells you about the distribution of tonalities, with pure black on the left and pure white on the right. The absolute numbers don’t matter as much as their relative position along the line. (Originally, the X-axis was divided into 256 segments because the original black and white histogram was made for 8-bit images that contain 256 shades of gray). For our purpose, however, you can divide the X-axis into thirds, with the left third representing shadow detail, the right third representing highlight detail, and the middle third representing mid-tones. This is a crude approximation, but it will help you visualize the tonal values and where they fit into the graph.

Photography and Text © 2009 Michael Lustbader

A mid-toned image will have most of its tones in the middle of the histogram.

A light image will show the bulk of its tones grouped towards the right side of the histogram.

Dark tones are found on the left side of the graph.

2

BEING COMFORTABLE WITH YOUR HISTOGRAM

The position of values along the vertical Y-axis simply tells you how many pixels are present at that particular tonal value.

This luna moth on wild geranium shows mostly middle tones, from light to darker green. Your meter rejoices when it sees an image like this and will nail the exposure with little, if any, help from you. A mid-toned image will have most of its tones, as demonstrated above, in the middle of the histogram.

This crab spider on a Datura blossom would be considered a “high-key” image, with its tonality mostly towards the right side of the histogram. I used fill-flash for this image, and paid careful attention to the histogram so I did not “clip”, or over-expose the highlights. Several exposures were required to “fine tune” the flash so none of the highlight values were lost. This photograph of a marine iguana in the Galapagos Islands (referred to by Charles Darwin as “imps from hell”) contains mostly darker tones which are grouped towards the left side of the histogram. The shadows under the claws and chin did need a bit of a boost during the RAW conversion but paying attention to the capture histogram ensured that the correction needed was minimal.

Personally, I find blown-out highlights more objectionable than blocked-up shadows. If I have to compromise and make a choice between the two, I will usually choose to sacrifice shadow detail rather than highlight.

It is important to remember that there is no “correct” histogram, just as there is no “correct” image. The shape of the histogram will vary according to the distribution of tonal values (light, dark, mid-toned) within the image. It is however, more than a graphic representation included to impress you with another flashing light–the histogram is a valuable tool that helps you create

3

BEING COMFORTABLE WITH YOUR HISTOGRAM

More examples:

When the image has a preponderance of light tones, the histogram is shifted to the right. The fact that the peak of the histogram runs off the graph is not important–it just means that there are more light tones than the graph has room to show. These tones are not lost, just not visible on the graph.

This, however, is an accurate representation of the image, and does not require adjustment incamera. That doesn’t mean that I may not still do some dodging and/or burning in Photoshop. As we will see later, an image may still need to be optimized for maximum aesthetic impact; the histogram just assures that all of the tonalities are present in the original capture.

histogram will vary according to the distribution of tonal values When the(light, image contains mostly dark tones, its histogram is shifted to the left. dark, and mid-toned) In this image of two oystercatchers, there is slight within the “clipping” (which we will discuss shortly) on each end, image. It is, to me, since it represents areas that but it is acceptable however, truly contain no detail. Analysis of the image shows that the areas where detail is lost are located in the black more than shadows in the lava and in the white pebbles. No detail is just a graphic

lost in the black or white feathers of the birds. Attempts at correction might lead to excessive lightening of shadows, burning out of highlights, or both, so I chose to expose “as is”. Again, some burning and dodging took place incomputer, but all the important tonalities were preserved in the original exposure..

representation included to impress you with another flashing

The black areas of this Amazonian poison dart frog’s skin are mucoid and truly without detail. Even though the histogram shows clipping of the dark tones, it would be fruitless to attempt a salvage of tones that are not present in the original capture. I bracketed this image in the field, and even in the over-exposed images, there was no detail in most of the black areas. The remainder of the image is quite middle-toned, and I was able to use the camera’s recommended exposure. A reflector provided the catch light in the eye. I was afraid that fill-flash, even at low intensity, would cause too many distracting reflections.

OK, so that’s a lot of theory. How does the histogram work for you in the real world? a final image containing all the information camera is exposures. It helps you choose a Essentially, it significantly decreases the needyour for bracketing capable of capturing. combination of f-stop and shutter speed which will yield an optimal exposure and identifies exposure problems at the time that you are best able to correct them--in the field. BEING COMFORTABLE WITH YOUR HISTOGRAM

4

What is Clipping?

Each end of the histogram should taper off and approach the baseline BEFORE it “hits the wall”. A graph that runs into the wall BEFORE it hits the X-Axis on either side indicates “clipping” and subsequent loss of available information.

If clipping occurs on the left, detail is lost in the shadow area of the image. If you simply increase the brightness after exposure (with Levels or Curves in Photoshop, for instance), all you will accomplish is an increase in NOISE in the affected areas. (Noise being the digital equivalent of grain). You can lighten the area in question, but you cannot recreate detail that has been lost. This is a problem you must try to improve before exposure, although you do have more leeway if you are shooting RAW. Shadow Clipping

Shadow clipping occurs when the histogram “hits the wall” on the left side. Detail is irretrievably lost if this is not recognized in the field, although some salvage is possible if you are shooting in the RAW format. Change your exposure parameters to shift the histogram to the right.

There is no “correct”

You can correct clipping of shadows by using a slower shutter speed or a larger aperture--in other words--MORE LIGHT. To this end, you can also use fill-flash or a reflector. You can usually judge the result of using the reflector through the viewfinder, but if you use flash, be extra careful not to burn out your highlights. You will learn to rely on your histogram. Fill-flash was used to photograph this crab spider. I used a two-headed Nikon R1C1 ring flash, with one flash straight ahead at 2/3 power and the other from the side at 1/3 power. Just enough light to open the shadows, and not enough to burn out the white highlights.

histogram, just as there is no “correct” image; there is a unique histogram for each unique image. The shape of the

5

BEING COMFORTABLE WITH YOUR HISTOGRAM Highlight Clipping

If the clipping takes place on the right side of the histogram, highlight detail is lost. Change your exposure to shift the histogram to the left. Common wisdom states that, to minimize noise (the electronic equivalent of grain), the histogram should be shifted as far to the right as possible without clipping the highlights.

OK, so that’s a lot of theory. IfHow highlight or shadow detail is for compromised the original exposure, you cannot restore it in does the histogram work you in the in real world? your editing software. Remember the saying that you only get one chance to make a good first Essentially, it significantly decreases the need for bracketing impression? same true of capturing an optimally exposures. ItThe helps youholds choose a combination of f-stop and exposed image. The only chance you get to capture a full range tonal values is in theone original exposure. If you are capturing shutter speed which will yieldofan optimal exposure, RAW files, you do have a second chance at salvage during your RAW conversion, but even the which contains all the tonalities of the scene that your camera best RAW conversion cannot replace detail that has been lost. is capable of capturing. Each end of the

In this photograph of a milkweed bug, the initial histogram showed two problems. First, the milkweed silk was overexposed (highlight clipping was evident on the initial histogram). In addition, the blacks were also blocked-up and without detail.

The solution here was to underexpose to restore detail to the seedhead. Then in Photoshop, I was able to manipulate the RAW file to restore some of the detail to the black areas on the bug’s wing. When I have to choose, I will always optimize highlights first--To me, a burnt-out highlight is much more distracting than a blocked shadow. In this image, I could have lived with the overly-dark blacks, but the burnt out highlights would have ruined the image for me. This is a subjective call--everyone has a different tolerance level.

So, youshould see thetaper pattern theapproach graph and Now change your exposure settings graph offof and theinterpret baselineit.BEFORE it (aperture and/or shutter speed) to compensate for the clipping and recheck the histogram after “hits the wall”. your next exposure. A cliff-edge pattern on either side indicates “clipping” and Issubsequent clipping always “bad” thing?IfNot necessarily. depends loss ofa information. it occurs on theItleft, detailupon the content of the image. In some clipping areasIfofyou pure black,increase which contain no shadow detail. In is lostcases, in the shadowmay arearepresent of the image. simply other situations,after it may represent highlights (reflections off water, metal or glass), and the brightness exposure (in specular Photoshop, for instance), all likewise may not contain any highlight detail. The histogram gives you warning of potential you will accomplish is an increase in NOISE in the affected trouble and allows makeequivalent adjustments areas. (Noise beingyou theto digital of before grain). you see on the monitor that you can’t “fix it in Photoshop”. If the clipping takes place on the right, highlight detail is lost. If highlight or shadow detail is compromised in the

Some general rules of thumb:

6

BEING COMFORTABLE WITH YOUR HISTOGRAM

A. Avoid clipping at either end unless you are comfortable that it represents either deep black without detail or specular highlights. B. If the contrast range is too great to avoid some clipping, then clip shadows rather than highlights. There is software available and “workarounds” that can help you minimize shadow noise, but once detail is lost, you cannot replace it. Blown-out highlights are more objectionable than blocked-up shadows. C. Move the body of the histogram as far to the right as you can without clipping highlights.

(No, I have not forgotten about High Dynamic Range techniques or layer masks for contrast adjustment, but those are topics in their own right and deal with issues which may be unfixable with a single exposure in the field). Become comfortable with the histogram. It may appear several times during the life and processing of your digital image: A. In-camera, as just described, you can use it to optimize the original capture. B. In your conversion software (Camera Raw, Nikon Capture, etc.) if you are shooting RAW. C. In some scanner software (if you are still shooting slides or scanning your archive into your editing software). D. In your photo-editing software. Remember, the more optimal your image is BEFORE you bring it into your photo-editing software (Photoshop®, Photo Elements®, etc) the more information will be available for you and the better the final quality of your image will be.

An adjunct to the histogram is the “highlight blinky” warning. This can be helpful in letting you know that you are clipping your highlights, but it’s like the idiot lights on your dashboard. The oil warning light will tell you when your engine is about to melt down because of lack of lubrication, but an earlier warning would be more helpful. Likewise, the blinky warning will tell you that your highlights are blown, but the histogram gives you the details of which tones and by how much your image is compromised. This was a slightly tricky histogram to interpret, because the dark shadow of the grasshopper did contain detail, but the holes in the leaf did not.

The histogram reflects this, with one peak in the middle, representing the greens, another peak on the far right, representing the shadow, and a clipped region off the left side of the histogram, representing the detail-less black hole in the leaf.

Many cameras have outgrown the simple black-and-white (luminance) histogram and now offer light– it is a valuable that helps you interpret thebeing represented by its own curve (along the same graph, but in tool RGB, with each color channel distribution of values for each image and create a final image the same axes). that contains The importance of the three-color histogram was all the demonstrated recently, when I photographed the wings

7

BEING COMFORTABLE WITH YOUR HISTOGRAM

of a morpho butterfly suspended in a spider web in Costa Rica. The luminance histogram looked fine in the field, but when I uploaded the images into Photoshop®, I discovered that the blue channel was burnt out, with NO detail in several areas of the bright blue wings. Ray Klass, a Photoshop®-savvy friend had showed me a technique called “channel blending”, where detail can be optimized in one channel and then re-applied to the image, and I was able to bring out some detail in the burnt out areas, although not all.

information your camera can capture.

Even after correction, the histogram still doesn’t look perfect, with both ends of the graph showing clipping of the blue channel. The image, however is acceptable. When I tried to completely eliminate the clipping in Photoshop®, I lost the iridescence of the blue in the right hand wing. Sometimes, you have to trust your eyes rather than the electronics.

Just an expanded view of the histogram in Photoshop®, showing the clipping on both ends of the blue channel, as well as in the red channel. When in doubt, this is one of the rare times that “shoot now, fix it later in Photoshop®” is a legitimate plan.

BEING COMFORTABLE WITH YOUR HISTOGRAM

8

original exposure,

Channel blending is not always successful, however. The luminance histogram (black and white) on an older digital camera indicated a good exposure of of this yellow warbler, with no clipping. On the monitor, however, the loss of detail in the bright yellow breast feathers was obvious. I was not able to restore this detail in Photoshop®. A color histogram would have led me to underexpose and bracket in the field, and then use either blending techniques, cloning, or layer masks to enhance and/or restore the color detail. Live and learn.

you cannot restore it in your editing software.

The histogram of this Indigo Bunting again showed a normal luminance graph, but the RGB histogram showed clipping in the blue channel. Detail was saved in the final image by underexposing slightly, being careful not to clip the shadow detail. These, of course, are the Photoshop® histograms, not the ones from the camera LCD, but the principal remains the same.