Histograms Part 5

So now we get to it. It may seem obvious at this point what you can do with histograms (if so, great,) but just in case, I'm going to try to explain exactly how to use them.

When you look at a histogram (hopefully a three-color histogram,) you are looking for mountains of data spreading across the graph that don't bump up against the edges of the graph. If you do have a peak or a spike against an edge of the graph, you will be losing information in your file and consequently, texture and contrast in your image. Additionally, you don't want a large space between your data-graph and an edge (the right edge especially.) So here are some examples…

In one of my very first posts on exposure, I gave same ideas to help judge exposure based on what you can see on your camera's LCD. Those judgements are be based on visual cues which lead to an evaluation that is ultimately subjective. I'm going to use those same images, but this time evaluate them objectively using histograms. Here are the images with their corresponding histograms following each:

The above image is underexposed. Even though the graphs are not bumping up against the left edge very much, the fact that there is no information in the right quarter of any of the graphs tells us that this image is underexposed. Very badly underexposed images may have graphs that are more extreme than this, but those will be more obvious without a graph. This type of underexposure is where histograms are really valuable. If you see a graph like this, use exposure compensation to adjust your exposure.

This image is overexposed. You can see in the red channel, that the graph is running into the right edge - this is often the case in overexposed images of people since most skin (black, white or in between) will show up in the red channel. The overexposure of the red channel causes the neon look in the skin (notice the nose in particular.) If you see a graph like this, use exposure compensation to adjust your exposure.

This is properly exposed image. The graphs come right up to the right edge with out running into it. These are the types of graphs you should look for. Here are some other good looking histograms following their corresponding images:

This last one is a curveball - and an important lesson. The histograms are running into the right edge indicating that there is some overexposure. However, this is a properly exposed image of my daughter. What is overexposed is the snow. That is why there are patches of pure white in the snow where you can't see any texture. The only way to save the detail in the snow would be to underexpose my daughter. I didn't do that since, obviously, she is the most important part of the picture.

In many images, there will be decisions like this you will have to make. So don't just depend on histograms without thinking. If you can identify why your graphs are hitting the edges, and your ok with losing that information (like the snow,) then don't worry about it.

Read Part 6 next.