Histogram Graph

The Histogram Graph is a supreme exposure aid used by the analytical digital photographer for better exposure placement. Under some situations, the light meter is easily influenced into false exposure interpretation. Many people check their exposure accuracy by looking at their LCD screen, and this can give a false impression if the screen has not have been left at it's factory default settings. What's more, it can be hard to see under the harsh sun light.

The bars in the middle of the graph represent the middle tone values of your scene. The far right side of the graphs is the brightest highlights, and the far left is the darkest shadow details.

The height of the individual bars represent how much of your entire scene is recorded at this specific brightness value.

Do not associate the placement of the bars on the graph, with the left to right horizontal placement of light across your scene. A bright highlight on the left of your scene appears only on the right side of the graph.

What should a proper histogram graph look like?

There is no real right or wrong here, different scenes may show the distribution of light and the bar heights at different levels. There are many different lighting situations that people take pictures under, and the graph only represents what light the image senor can record.

If the bulk of your bars are on the left half of the histogram graph, then you have a low key lit scene, or it is also possible that you might be under exposing your scene. Interpretation of what you are tying to capture of the image is the only real relevance.

If you are shooting on snow in the sun you should see a shift of most of the light being more to the right half of the histogram graph if you have properly exposed for the snow. In the snow scene shown above, there a lot of trees in the sun and shadow on the snow, so the bulk of the brightnes tones of this scene are middle tones.

The histogram graph is not showing the Dynamic Range of the light that exist in the scene itself. The graph represents how your image sensor has recorded the light on the image sensor. So changing exposure is how you redistribute the placement of the light on the graph for the next shot.

At first using the histogram graph may seem quite abstract, after all it is hard to relate a region on the gram with a particular brightness value. It is true, in order to use the histogram effectively you may at least want to learn certain regions of the graph as certain bench marks that you can relate to. I will go over some of these here, but I will spend more time on this under the section of Photo Light Meter.

Histogram Graph - Proper appearance of the Histogram

Histogram - The ends of the Graph

Top Bottom

Perhaps the most important part of woking with the histogram graph is the light at the extreme ends of the graph. It is quite common that when you take a picture outdoors you run a risk of overexposing in the sky quite easily. Your Graph will show you this by the displaying a bar height that is fairly tall on the last bar to the right side of the graph.

It is okay if you get a some light in this bar, but smaller is better. This is because this last bar represents you brightest highlght, and if you want any shaping or tonality to occure in brightest highlights of your image, it must all occure below this bitghtness level, this bar is your peak brightness.

If the bar is high here, there is a strong chance that are scene highlights that are being lost do to your exposure setting. Photographers referred to this as "clipping".

Often a good exposure is one that records the full range of light of the scene without loss of details. You want all the shaping that the light can give you so that your image can reveal form. Clipped highlights can kill that, the appearance of clipping is like looking at a blank sheet of paper.

Randy Smith Photography © 2011.

This is generally to be avoided on both sides of the histogram graph. All though clipping of the shadows is considered to have smaller negative impact on the viewers than having out of control clipping of the highlights. Realistically what would be the point of lacking tonality in a photograph anyway. Your eyes can see some tonality differences in shadows and highlights even if you do not notice details in these areas, so the photogrph should also show some varyation in these areas.

It is quite natural on sunny days that the dynamic range of the outdoor scenes often exceeds the range limits of what the light sensor can record, especially if you are shooting towards the sun. It then is quite common that you might have to choose what details are more important for recording your picture. Do you favor the highlights or the shadow areas knowing you may have some clipping. Or do you change your POV (point of view) altogether by reframing to avoid those areas that are forcing you to risk some clipping of this exposure.

The answer to where to place the most exposure may depend more on what you are shooting.

Nature scenes and landscapes look more natural if we limit the amount of clipping of the highlights, so don't let the bar on the graph get vary tall.

People that photograph other people are vary concerned on the placement of flesh tones, so they ensure that these details are present somewhere near the middle of the graph ± one stop. However if the person is wearing white, like the importance of a wedding dress is to the image, the subtle highlights of that dress are quite critical, and clipping in those highlights would not be acceptable. After all, someone paid a lot of money for that dress and the bride wears it once,... hopefully.

So how can you incorporate all of the light range into an image without clipping part of these important details?

People photographers often use reflectors or electronic flash fill as light aids to boost more light into the shadows or mid range of the image, so that they can readjust exposure to a darker settings to include the important highlights while bringing up the shadows and mid tones. They in effect reduce the scene's dynamic range to fit the image sensor's dynamic range.

In short, allowing the light to slope down to the vary far right of the histogram graph, or to the vary far left, is a good placement of exposure. It is when we are letting the light in the end bars climb up that the image is being hurt by loosing important highlights or shadow details.

There are a few acceptable highlights that it is okay to let completely blow out. A bright light source like the sun, or a specular reflection off of a shiny object or the water, that is mirroring the light of the sun. Most bright street lights and or car lights are lights that should carry some color and tonality to them. A vary small measure of haze or clouds immediately next to the bright sun may also be beyond the range of your exposure. What determines how much clipping is ultimately is acceptable is based on how does the final image looks to you.

If I am shooting a snow scene, I take extra caution to not let the last white of the image touch anything but the smallest portion of the last bar on the histogram graph, and I may not even let snow reflectance reach this bar at all. Some snow conditions can present mirror like reflections and sparkles, but usually snow is not thought of as a light source it self, and should contain tonality throughout.

I will also mention here that if you save your images in the camera as the 8-bit jpeg format, that your image only has these 255 level of brightness for each color channel to work with.

If you save your images in the camera as the 14-bit RAW file, each bar of the histogram represents about 63 levels of brightness. This type of image file gives you much more room for controlling the highlights later when you are editing. This means that even the brightness values of the last bar can be pulled down into lower brightness values that can then add to shape and definitions of details.

Slide film has about the same dynamic range as the digital image sensor, but when shooting slide film photographers never had the luxury of looking at a histogram graph to aid in exposure placement. There use to be a saying when shooting with slide film, "Expose for the highlights and let the shadows fall where they may".

Limiting clipping of shadows

If you're shooting a low key image, good lighting and exposure control will mean you don't have clipping in the shadows, even if the details that are present in the shadows are barely recognizable as usable details. What good is jet black space over large areas of the image to a photograph.

It generally is also a good idea to have at least some bright highlight somewhere in the image that will help serve as a reference point for our eyes, such as a bright reflection coming off a shinny object, like catch lights in the eyes, or a specular light source like a light bulb. This is also making good use of the dynamic range of the scene.

If your image has just medium gray values and dark shadows, it is not going to be a vary appealing look. Think back to the B&W movies of the spy in the dark shadows of the street scene, the street light at the end of the block is not a low brightness value.

Histogram Graph - Ends of the Histogram

Histogram Graph - Exposure Brightness Placement

Top Bottom

Making good use of exposure placement usually means you have more control of shaping of highlights or shadows when you are editing for the final image, which can help create a more appealing image. So getting close to your desired exposure placement will help you in the editing of the image.

If you shoot an exposure of snow and you accept that the exposure meter is going to under exposure here, then you can use Exposure Compensation to adjust for more exposure to get those highlights back up to where they should be.

Where you place the light level of the scene for any particular subject is a matter of personal preference. Your just trying to use exposure placement to expresses your subject favorably within the framed scene. Ansel Adams was great at taking middle toned scenes on B&W and expanding the highlights and shadow values of those mostly middle tone subjects, out towards the ends of the dynamic range of the final printed image.

When shooting scenes intended to have dark shadows, it does not hurt to shoot the scene with a bit more light into those dark shadows then you are intending to present in your final image. During editing you can easily adjust the brightness cure and pull those details down into darker regions of your image to your desired effect. It is much more difficult to get good image quality trying to pull shadow details up that were shot too dark, into a bright tonality so that you can see those details. This can result in bringing up stronger the pixelated signal to noise ration problems an image can have.

Regional brightness reference

The Histogram graph does not come with a stepped gray scale across the bottom of the graph, so I set up few examples that might help with understanding regional brightness in steps of ± one stop increments.

To begin with, your light meter may use complicated algorithms and regional sensitivity differences within the viewfinder to read the light of the framed shot.

Histogram Graph - Middle Tone
ER = Exposure Reading in this example, it is the exposure meters interpretation of light, it allways strives to return a reading for a middle toned value between Black and White.

In this histograpm example, I photographed a plain White towel and let the light meter determine the exposure. I dropped out the normal background of the histogram graph and I am showing here the white towel.

The camera light meter has placed the resulting exposure of this towel as a mid tone image, not as a white. This is the basic job of the light meter, to average the light of the scene and set the exposure for the middle tone, so the white towel is placed between the brightest white and the blackest black.

You may also notice that this curve is not dead center in the graph, and this is rather common for most all digital camera manufactures. Camera light meters are not calibrated for a purfect 18% middle gray value, they are generally a bit less. We go over that in the section Photo Light Meter.

The narrow inverted bell shaped cure shown on this histogram graph is showing that the dynamic range from the brightest values in this image to the darkest values is rather narrow, and perhaps only a stop and a half wide. This image is rather flat looking as far as light range is concerned, and this is a histogram graph example of a low contrast image.

You can see the curve is made up of little vertical bars, and just to rehash what we talked about, there are 255 brightness values represented across the graph.

So if you take a picture of a plain white card you get a gray picture of a white card. If you take a picture of a plain black card, the camera will attempt to give you a gray picture of a black card. Luckily, if you take a picture of a gray card, you get a picture of a gray card.

To make this white towel look like a white towel, an exposure compensation correction is needed, or you can be satisfied with the gray towel and try to brighten it up later during editing the image. For best image quality sake, it is best to shoot the white towel as a white towel.

In this next graphic I superimpose a number of graphs on top of each other, to show you where the different graphs would be if I only adjust the light meter in ±one stop increments as I take an additional pictures of the same towel. The lighting was not change.

Histogram Graph - ±1 Stop Increments

You can see where the peaks of the bell curve of these one stop changes in exposure are in relation to placement on the graph.

It would be nice if there were some small marks outside of the histogram graph that indicate these steps in exposure, but since we don't have that you can kind of get an idea of how much your exposure adjustment will shift across the graph based on the amount of exposure change you add for the next shot.

The histogram on your camera most likely has these line dividers that break up the graph into five parts like the ones that I show here. The only meaning of those is to divide the histogram up into equal sections. So 255 brightness levels are divided into five groups of 51 steps of light.

If you can remember the distances of least where the peaks of the ± 1 stop of light changes are then you can intuitively see how much correction may need to be added to, or subtracted from any additional exposures in order to place exposure values where you want them.

So back to my white towel example

If I wanted my towel to appear white to begin with in the first shot, I would take an exposure reading and then make an exposure compensation adjustment of about +1.5 to +2 stops of light, this would place the towels exposure more in the visual range of were we might see the towel with our interpretation of white.

I took this example to +2.5 stops increas over the normal exposure reading to show you that this towel is now white, and the graph shows that there is no clipping in the whites on the histogram.

Histogram Graph - Placing white where white should be

Just how bright this white should be is somewhat subjective, but your eyes and mind work together to make you feel that a white towel is still white when you are in any variety of ambient light brightness levels. I don't want to push the towel exposure to far to the right because I want to preserve the texture of the towel which comes from the shading around the Terry Cloth strands. This would mean watching that I don't let the last bar on the right of the graph get taller than maybe one small step upwards.

In this towel example, I have filled the frame with just one subject, and it completely covers the entire frame, this is not a vary common example for a real world scenario of most photographs you will shoot.

If you are shooting a picture in winter on the snow, you might have a wide range of features at different reflectance values. You need to be aware of which type of exposure metering style method you are using, so that you can interpret what the light meter is likely to see and be judging in that framed shot. Is the area of the shot composed of more trees than snow? Is the snow in the sun or is it in the shadow? All of these brightness values influence the light meter's exposure placement.

If you feel there are too many variables, it's okay, let the meter set the exposure, take a picture and examine the LCD screen, and the histogram. Watch for the ends of the histogram for potential clipping of details, and then adjust the exposure according to the way you want to see the image.

Rest assured though, you can eventual get vary familiar with how a light metering style is reading a scene, and you can become a good enough judge of what amount of area a brightness value dominates the scene, and then make an exposure compensation adjustment on the first shot. The section Photo Light Meter will help you develop that skill.

You may have noticed in the above diagram of the histogram graph that there is about 6 stops of dynamic range total from end to end for the image sensor if you are shooting in the jpeg format.

If you are saving your in camera images in the RAW file format, the appearance of the histogram graph will not change, the image appearance on the LCD screen will not change from that of the jpeg file, it is only during the editing phase that you can take advantage of the benefits of the RAW file format to gain an additional 2.5 stops of recoverable details shared between the shadows and the highlights.

You may have also noticed that you have about 2.5 stops of adjustment towards the highlights and 3 stops of light adjustment towards the shadows. So you need to watch your highlight just a bit more carefully than the shadows or they can get lost off the bright end of the dynamic range.

Histogram Graph - Exposure Placement

Histogram Graph -Examples

Top Bottom

Tuffted Puffin

Randy Smith Photography © 2011.

In this image of a marine bird, the histogram shows two different areas of the graph with their own peaks, instead of one single bell shaped curve. That's okay, the graph is just showing what brightness values exist in this image as well as the quantity of the pixels at these brightness levels. Notice at both ends of the graph that the peaks are vary low, so it is likely that there is vary little or no pixel brightest values in this scene that exist beyond the dynamic range of this image capture.

Notice also the delicate white shading and shaping of the white feathers on the Puffin's face, none of that area in the picture appears to be blocked up as a solid block of single bright tone, this is what you want to see in your highlight, shaping and smooth textures. The same is true for the low light areas of this image. The only place where we might loose some shadow details is in the pupil of the eye, and that should be black, and likely the darkest black of the image. If it were a light black tone, then the bird would appear to have Cataract problems in the eye.

Marsh Scene

Randy Smith Photography © 2011.

This fall scene is actually more than one picture, it is an HDR (High Dynamic Range). It was made from three images taken at different exposures, and is later blended together during editing. The reason for three different exposures is because the natural dynamic range of the scene exceeded that of what the image sensor could capture alone in one single shot.

Notice in this HDR image, that the ends of the bell shaped curve do slope down to the out side edges of the graph. However, the last bars on both ends of the graph do show that a lot of pixels are represented here as the height of the bars are climbing up taller than surrounding pixels. This suggest that there are pixels in the natural scene that are still outside the range of what the image sensor captured and that those brightness areas are not represented in this image. So this means we have clipping. These are areas of the image that exhibit no details, we have just pure black or pure white.

Clipping of brightnes and details are not normal to our eyes in nature, and generally you don't want to see it in our images either.

The tall spike on the right of the graph is the sun, and flare of light from from the sun. It is not too surprising that we have a spike in the histogram graph when you include the sun in your image, and you should expect that this light will be represented like this, it is far too bright to include into the scene as anything but the brightest part of the image. To tone the sun down by exposure on a nice clear bright day like this would mean making the rest of the natural scene would go much darker, and I wanted to see the grasses and other details throughout this scene easily.

If you do not have the sun, or a bright specular light source, or reflection in your scene, and you are getting this kind of spike on the right, you are way over exposed. Even if you are shooting a high key scene, you should see shaping details in your brightest whites, and the last bar on the right should not be climbing up the sides of the graph.

RGB Histogram Graph Display

You can also display on your camera the histogram graph showing a different graph for each of the independent Primary RGB color channels. There is no difference here with how you read this graph than the other histogram graph display.

RGB Histogram Graph for the Crocuses image above.

If any of the RGB color channels has a tall bar on the far right or left of the graph, then that color has reach is maximum saturation for that color and brightness, and it means the color is clipping. If you want details in those colors then you need to make an exposure change so that you can be sure to capture those details.

Just like with any exposure that has some clipping, you would reduce exposure to ensure capture of details in those brightness ranges.

The left side of the graph means that that color is less represented in the image, and so not vary prominent.

So ideally you don't want any of the color clipping, and each of these RGB graphs often times look quite different than the other.

A red ball would will show less presence of bars in the graph in the blue and green frequencies.

If you looked at the link Color Images under the Photography Basics section, then you are familiar with the three Primary colors of light being Red, Green and Blue, and that these three colors are all we need to blend together to make any color of light that we see with our eyes. The image sensor only records light brightness in these three wavelength or RGB. Where you see an over lap of the Positive Primary colors of RGB you get the other Subtractive Primary Colors of Yellow, Cyan and Magenta. This explains the yellow and cyan and magenta areas shown in the RGB Histogram as overlapping regions of RBG values.

If you took a picture of a truly gray or white ball, the three color histogram graphs of RGB should appear to have almost equal bell shaped curves. If the graphs are not similar then this could also be an indication that your color balance from your light source is not balanced for your picture color balance or picture style setting.

If I am looking at any other color in the scene other than a true gray or white, I certainly could not tell much about color balance, I would not know how much was suppose to be real color of the object as shown in these three color channels, and how much of that color was a light balance issue. However, if we are looking at the image on the LCD scree, even the less discriminating observer can usually tell if our color balance is way out of whack.

Return to Digital Photography Basics from Histogram Graph