Should I pay attention to the histogram

Discussion in 'Digital Darkroom' started by nikonichiban, Dec 22, 2004.

  1. I shoot RAW and edit in Elements 3.0.

    When I import a RAW file and adjust the exposure, brightness, contrast,
    shadows, to my liking, the histogram ends up representing a "poor" exposure.
    The histogram is not a bell-shaped curve and often is bunched to one side of the
    spectrum.

    Maybe I have my own eye and my own idea for what I want my pictures to look like. I
    like high contrast and dark shadows. When I adjust, I look at the picture and use my
    eye, basically ignoring the histogram.

    So if I like the way it looks, should I worry that the histogram is "wrong?" Is there a
    "right" exposure?

    Thanks
     
  2. The only "right" exposure is the one you like.
    Here's a good article about understanding histograms.
     
  3. There is no right or wrong with this stuff. Adjust the way you want. Most people want some shadow detail and some detail in the highlights. I find blown highlights to be a lot more annoying (at least with digital. Didn't seem so bad on film for some reason) than blocked up shadows.

    At the end of the day you're the artist. Create what you want.
     
  4. Yes. Pay attention to the histogram, but don't let it make the final decision on what you pictures look like. Just keep these things in mind:

    1. Your computer monitor is an active display - creating its own light. Printed photographs are just reflective. So, what you see on the screen is rarely what it looks like when it's printed. Specifically, light areas can look brighter on the monitor than they will on paper.
    2. The very left edge of the histogram represents the darkest you can possible print. The right edge is the brightest. You probably have a good idea what those shades are on printed photos. If you know that you want a photo that is skewed heavily one way or the other (perhaps even lacking either blacks or whites) that's fine. As long as you recognize what you get when you print the photo.

    My photos are also very rarely a bell curve centered around the middle of the graph.
     
  5. Thanks for the great feedback. I haven't printed anything yet, as I've been at this
    digital thing for about a week, so I hadn't considered the difference between monitor
    display and print. That's a really good point, that I hadn't considered.

    The choices I often make in framing and subject matter mean that my histograms
    may be always look lopsided.
     
  6. Wow, that link is totally messed up. Nevermind.
     
  7. David,

    Like wine, you're the final arbiter about what you like, but the histogram is important but really should be looked at at the time of exposure. When you hit the right side, you've exceded the capacity of the ccd to record an image, therefore you're excluding that part of your field of view from your picture, i.e. you've lost all data: it is simply white. Is that what you want? At the left end of the histogram, you canlose detail too but digitally it is often recoverable which is impossible at the right end. The bell curve concept is ridiculous. Between the two end of the histogram is merely determined by the color and luminence of what you're photgraphing. Uwe Steinmuller has a sucsinct article on this in the current shutterbug, but I'd recommenr you trian youirself to shoot with the histogram. If you want to blow highlughts, you can always do that later on your computer.

    Paul
     
  8. David - nice image and a good example that it is impossible to come up with any particular shape the histogram >should< have :) How much more different can a histogram be from a bell shape ^^ - a nice image just please do not try to force it into any "correct" histogram shape :)
    00AXQG-21046984.jpg
     
  9. Agree with Paul. Photography is both art *and* science, which is why comments such as "Hey just do what you like" don't cut it.

    They might be appropriate to somebody who understands and has mastered all the underlying matters of technique and technology, however.
     
  10. this website explains it the best I have ever found so far.
     
  11. The primary reason for a histogram is to enable you to place the initial exposure on a digital camera or a scan appropriately in the resolution of the sensor.

    After that, who cares if the image comes out like you want.

    On the other hand, you can get some hints at quantization effects that maybe adversely affecting the manipulated image.
     
  12. Hmmm... In the image you submitted. If you do a histogram of the entire image- well you stated the outcome already. But if you do a histogram of a portion of the image you get another distribution.

    So, I would say it is dependent on what you are intending to do in the shot. Keep in mind that your camera will likely give you a histogram of the entire sampled image.

    Have you also considered the technique known as bracketing?

    Hope I make sense.

    [If you do an experiment of this in Matlab, Octave, or your software of choice it makes more sense. But this may be beyond the scope of what you ask.]

    -Luis.
     
  13. I read "i adjust brightness, contrast" etc.. now, if you use the brightness/contrast controls of PS, no wonder you get poor histograms. These controls literally destroy your picture by clipping the information. Try it out by viewing a hist. before and after. As a rule, NEVER TOUCH brightness/contrast controls. NEVER.
     

Share This Page