Tag Archives: histogram

What is a good histogram?

In my last entry, I introduced the histogram and described its basic function. I even showed a couple of examples. If you read the entry you should have a good understanding of what a histogram is, but there’s probably a great big question in your mind right now: “What is a good histogram?”

The short answer is that there’s no such thing.

You don’t like that answer much, do you? Nope, me either, so even though I could save myself a bunch of typing I won’t leave it at that. I don’t want people running around saying that I’m a meanie.

GoodFlowerHistogramThe longer answer is that what a good histogram is depends on the subject of your photo. For “typical” images, those with an average range of light and dark areas, a good histogram will basically look like a bell curve that’s centered around the center of the graph. And look! Here’s one now. Isn’t that convenient?

Looking at that histogram, you might conclude that there’s nothing in the image that’s 100% bright white, and a very tiny amount that’s pure black. The photo has a good range of tones from light to dark, and there are more moderately-light pixels than moderately dark ones. Let’s look at the image and see if we’re right.

GoodFlowerHey, we guessed pretty well. The flower has a range of tones from black to nearly-white, and there are pretty big light-colored areas.

Note that the exposure looks pretty good to our eye and on the histogram. There are no areas where it looks like we’ve cut off dark colors and made them black, or lost information because the pixels have become pure white. If we had, you’d see spikes at the left or right side of the histogram.

In general, that’s the mark of a good histogram for a typical, average photo. You generally want the majority of the pixels to be somewhere in the center of the graph, without a lot of stuff piled up on one side or the other. If you’re shooting something that has a significant amount of pure black or pure white in it, you’ll sometimes get a spike on the left or right side. That’s OK, but you’ll probably be OK if most of the pixels are somewhere near the center.

When I shot this flower, I took a bunch of pictures with the lighting changed slightly, and most of them didn’t turn out as good as this one. Let’s look at a couple of bad ones now.

DarkFlower LightFlower

If you’re like me, you’re probably thinking about the three bears right now– “This one is too dark. This one is too light. And this one is juuuussstttt right.” If you look at the bottom right corner of the images, you can see the histograms for them. (Or click for larger versions of the dark and light ones respectively.) The histogram for the dark flower has the whole graph left of center, and there’s a big pile of pixels right up against the left edge.  That means that all of the colors in the image are dark, and the part against the edge shows that there’s a bunch of solid-black pixels.  That’s not good– it means we cut off a bunch of information.

The too-light flower has almost all of its pixels on the right side of the graph, which usually indicates that the image has been overexposed. However, if you look closely you can see that there’s nothing against the right edge of the graph. That’s a good thing– it means that you can probably use an image editing program like Photoshop to fix the image and get something pretty good. The dark image might be recoverable as well, but it will never be as good because some of the information got lost when the picture was taken.

If you’re taking a picture of something that’s mostly black or very bright, or something with a narrow range of brightnesses, a correct histogram may look strange. In my next lesson I’ll show you some examples of those, and answer the question about the all-on-the-left histogram at the bottom of my last entry.

And here’s something for you to practice: get into the habit of looking at your histograms. Take a few pictures, and review them on the camera. Check your manual to see how to view the histogram, but on my Canons I get it by pressing the “Info” button while viewing an image. The best way to get familiar with it– pretty soon you’ll have a good eye for what’s right and wrong.

Leave a comment

Filed under Lesson

Histogram: What is it?

I confess… I’ve been neglecting Stop Shooting Auto! for a while.  I’ve been traveling constantly for the last couple of months, and that plus a full-time job has kept me quite busy.  In fact, I’m writing this while sitting at SFO while I wait for my delayed flight to take off.

AverageImageHistogram I promised I’d address the histogram next, so let’s do it.

Maybe you’ve seen something that looks like this, either in Photoshop or other image editing software, or on your camera when you reviewed an image.  Maybe you even took a guess at what it is and how it works, and if you did you probably guessed right.  That’s the histogram.

Histogram is a scary, technical-sounding word, but it’s actually a really simple concept.  The histogram is just a graph of how many dark and light pixels there are in your image.  At the far left you’ll see how many dark pixels there are, then moving to the right you’d see count of lighter and lighter pixels until the far right was pure white pixels.

AverageImageIn the histogram above, you can see that there are a very few solid black pixels, lots that are a little bit darker than average, a bunch that are a little lighter than average, and a chunk that are pure white. Let’s look at the image and see what it looks like.

There are a few pure black or very close to black pixels, mostly around the trash can and along the left wall– that’s the tiny hump on the far left.

The pure white pixels are mostly on the sign, in particular the words “San Francisco Marketplace” are pure white, and a tiny bit overexposed.  That’s the vertical bar right along the far right edge.

The dark of the sign, the dark carpet, and the dark part of the wall are the peak on the left. The peak on the right is the lighter part of the walls.

Most histograms that you will see don’t take color into account, only brightness.  That means that a very bright blue will look just like a very bright white.  Some cameras have the ability to do histograms with the colors separate, and of course most postprocessing software can do so as well.

DarkImageHistogramThis one is kind of fun. Without seeing the image, what can you tell me about it?  Obviously you can’t tell me what it’s a picture of, but can you make any guesses about what the image looks like?  Is it light or dark?  Does it have lots of different shades or only a few?  Are there any pure black or pure white areas in the image?

Do you think this image is well-exposed, underexposed, or overexposed?

OK, they’re calling my flight now.  Ciao!


Filed under Exposure, Lesson