Histogram

Now that you know how to manipulate your camera into getting a proper exposure for a given image, or specific mood you want to convey in your image, lets have a look at the feedback your camera gives you.  This feedback comes in the form of a histogram that displays on the LCD on the back of your camera.  It is also available in photoshop  and in many different imaging software programs you can use to  do the final processing in your images. The histogram is basically a graph layout of your image colors.  The horizontal or (x) axis displays the different shades of color in the image and the vertical or (y) axis displays the amount of pixels having that particular color. (and you thought you were all done with that math stuff!)

Bald-Eaglesunder1

 

Here is and example of a eagle sitting on a branch on a snowy blustery day.  The normal exposure the camera decided on made the image with rather dark looking snow.  Do you remember the black dog in the snowy field example?  The same principle is at work here.

 

The  image to the right is the histogram from photoshop to show how an underexposed image might Bald-EaglesHistogramunder1look on the LCD of your cameras histogram display.  You can see that the large peak representing the white (snow) portion of the image is in the middle of the histogram and therefore in that mid-tone grey region that your camera would like to turn everything into.  The color channels are separated for illustration purposes.

Bald-Eaglesover1

 

Now if you were to add one stop of light to this scene the dark dreary looking grey snow is once again rendered white as it should be.

 

 

 

Bald-EaglesHistogramN

 

You can also see that the peak representing the snow has moved towards the right hand side of the histogram and spread out across the (x) axis in the following image.

 

 

SMHistogramBW

 

To simplify this it may help to think in black and white again.  The left hand side of the graph represents complete black.  As you move toward the right hand side of the graph you go from black to increasingly lighter shades of gray and eventually to complete white on the right.  The following luminosity histogram shows an average scene with a complete range of shades from complete black (0) to complete white (255).

 

Think again of the black dog sitting in the snow covered field.   A histogram of this image would have a sharp upward spike from the black dog on the left side of the graph. Then it would quickly drop back down to the SMbaseline because there are few gray shades in the scene.  It would have a taller upward spike on the right side of the histogram to represent the snow in the image.  On the other hand if you imagine the white cat against the black wall, you would get a tall spike on the left hand side of the screen for the black wall, drop back down to baseline in the middle, and  spike back up again to a lesser degree on the right side to show the white pixels of the cat.  But because the world is not totally black and white, the average histogram displays a bunch of peaks and valleys all the way across the horizontal axis of the histogram.   This depicts every different shade that is present in a given scene and the amount of pixels of any particular shade in the vertical axis.  This is the picture that was represented by the previous histogram.

Now we can add color to the mix.  The sensor in your camera is set up to capture the light in three color channels;  red green and blue.  Besides just showing the brightness of each shade of white, gray, and black the camera can also display a graph of SMHistogramCeach color in the image in the form a RGB graph.  On the left hand side of the scale is pure black which is given a value of 0 on a RGB graph.  On the other end is pure white which is given a value of 255.  Going back to the black and white histogram, you are representing the different shades from black to white from left to right.   If you now overlay the red channel on this histogram you can see the number of red pixels from a dark shade of red on the left or “0” end of the scale to a light shade of red on the right or “255” side of the scale.  You can do the same for the green channel and also for the blue channel.  Because our eye perceives some light as brighter than others you will notice the histogram of each color is in a slightly different pattern and place on the graph. This image is a histogram displaying separate color channels of the previous picture.  The top histogram shows all 3 colors overlapped.

Now lets mix things up a bit.  Somewhere in your past you probably remember running into the color wheel and the 3 primary colors of red, green, and blue.  You also probably remember sitting there in your kindergarten class sweating bricks because your thesis on why green painted dogs are cuter than red dogs is due in 30 minutes, and you have broken all your crayons rendering them unusable except for the blue and yellow.  You realize that your future hangs in the balance.  You either come up with a solution quickly so you can graduate and move on to higher education with your classmates or get left behind to spend another year in kindergarten with the little kids.  Suddenly you get a brain wave.  If you put down a base of yellow and put some blue over top you get a sort of green color which will exonerate your thesis.  And what’s more, if you press harder you can change just how green it is and make a multi green colored dog.  Ok! so I got a bit carried away….back to the histogram.

If you assign a value from its location on the (x) axis for each color (red, green, and blue) you get a value that represents the amount of mixing of each of the colors giving you a completely new color. Remember creating that green color in kindergarten class?  For example light red would have a value of 255 red  0 green and 0 blue.  If you change any one of the values you slightly change the combination of mixing and wind up with a different color.  In the previous example of red if you change the value to 128/0/0 (not adding any green or blue to the image) you would have a darker shade of red because the value has move from the right light side of the graph to middle.  255/255/0 would give you yellow by combining light red and light green values.

turqoise

 

To the left is an example of a turquoise color. The mixture of colors comes from 0 on the red channel and 153  on both the green and blue channels.

purple

 

 

To the right is another example for a light purple color to show where on the RGB scale each color mixed from.

 

 

You can see how different colors can be represented by the mixing of the 3 channels.  255 different shades of Red X 255 different shades of Blue X 255 different shades of Green can represent over 16 million different colors!  Your camera and computer screen cannot display that many colors, but enough to give you the impression of smooth transitions different shades of the same color. The following right histogram shows how this overlap of colors is displayed compared to the luminosity histogram on the left.

colorscale2

Now to put all this to use.  A histogram of a well exposed average scene will have a histogram with peaks and valleys across the complete (x) graph from 0 to 255 RGB values.  If you overexpose an image the resulting image will become lighter.  If you go back to the   camera simulator   you can prove this to yourself.  This would push the resulting histogram values towards the right side of the graph. overexhistogram If you overexpose enough the graph will eventually be pushing off the right side completely.  This is known and having blown highlights.  Once you approach the right side of the scale the resulting whites become increasingly washed out and will eventually have no detail in them.  This means that you can no longer see those nice folds in that white wedding dress you overexposed.  This is what an overexposed image would look like on the histogram.

Once you have blown highlights they are next to impossible to recover, so keeping an eye on your histogram is critical for proper exposure.  There is an adage in digital photography to “shoot right”.  This means that you want to shoot so your histogram is as close to the right side of the graph as possible without going off the end.   This will give you the maximum brightness to work with when processing the image and still retain the detail in the highlights.  It is best to view your histogram in the color overlay mode as the red channel often tends to get blown before the others.  Most cameras have a setting to show you a blinking overlay of the blown highlights somewhere in the menu.  This can be handy as a quick reference point to adjust your exposure.

underexhistogramAt the other end of the scale are the dark colors.  If you underexpose an image the histogram will start bunching up towards the left hand side of the screen.  If you are trying to do a silhouette shot this may be the look you are aiming for with no detail in the blacks at all.  But on the average image where you still want so see the nice creases in the grooms pants   (I’m not sure how we got into a wedding mode here but….)   you will have to watch that the pixels that represent the black pants are not off the left end of the graph.  This would give you what is termed “blocked up shadows” or areas of dark with no detail in them.

noiseRemember in the first lesson we mentioned that noise tends to give a “mottled” look on the image.  This is one of the  unwanted side effects of underexposing an image.  The dark areas can have a tendency to display color noise when they are underexposed.  If you try to recover the image to the proper exposure after the fact by making it lighter (in photoshop for example) the noise often tends to be accentuated. Here is an example of slight noise from the previous scenic image.

There are some really good noise reduction programs out there that can do wonders with reducing noise and saving an image, but there is always a trade-off.  If not properly used, or if noise is really objectionable and a large amount of “cleaning” needs to be done, noise reduction software can smear fine details in the image.  NIK Define is my current noise reduction software of choice for its ease of use and ability to target specific areas of the image, but there are many other good programs as well.

So in conclusion….

When shooting the average scene remember the adage “shoot right” and keep an eye on the histogram (preferably with the color overlay) to keep an eye on the right hand side of the graph.  Blown highlights are much harder to deal with than blocked up shadows.  If you want to do creative photography like silhouettes or high key (images tending towards really light colors) the histogram can be invaluable in showing how you need to adjust your exposure.

In the next section we will have a look at  depth of field