Exposure quiz poll: extreme metering

Discussion in 'Open Discussion' started by Klorenzo, Nov 7, 2015.

  1. Four different pictures.

    4 vote(s)
  2. Two groups: the white pictures are identical, and the black ones are identical.

    5 vote(s)
  3. Two groups: the "low light" pictures are identical, and the "high light" ones are identical.

    1 vote(s)
  4. Four identical pictures.

    12 vote(s)
  5. Something else completely (please post details).

    0 vote(s)
  1. Klorenzo

    Klorenzo Mu-43 All-Pro

    Mar 10, 2014
    Hi, I thought about this quiz because I think metering is an often confusing aspect of photography.

    So, here is the quiz, with this immaginary situation:
    Set the camera to Aperture priority, fixed ISO, no exposure compensation, default exposure metering mode. Take a white sheet of paper and place it under a big lamp so that it is evenly lit. Take one shot filling the frame completely with the sheet. Now repeat with a black sheet of paper. These are the "low light" shots.
    Now add more light, with a second lamp or a more powerful one, and repeat both the shots. Again is important to have a uniform light over the sheet.
    You end with four shots: white low light, black low light, white high light, black high light.

    Now answer the poll question.

    - please answer the poll before doing the test, or similar experiments, in real life.
    - please answer the poll before reading the rest of the thread
    - feel free to post your answer below but please do not add too many details about the reasoning not to spoil the game for others (not too soon at least)
    - it's just a game, have fun
    Last edited: Nov 10, 2015
    • Appreciate Appreciate x 1
  2. oldracer

    oldracer Mu-43 All-Pro

    Oct 1, 2010
    Well, if you mean doing this with automatic exposure the meter is going to try to turn all four images into something like 15% gray. Probably it will fail somewhat at this because the metering is not perfect. I'll be interested to see how people's actual experiments turn out.

    But the punch line is clearly that you can't successfully use automatic exposure if you offer the meter a target that is far, far, from "typical" reflectance. That's why we have incident light meters.
  3. esnift

    esnift Mu-43 Veteran

    Mar 17, 2013
    Boston, MA
    Real Name:
    I'm banking on the idea that it will fail at making the white and black papers an identical 18% grey in either situation, But the two differently black and white tests should be pretty close I imagine. Too the laboratory...
  4. David A

    David A Mu-43 All-Pro

    Sep 30, 2011
    Brisbane, Australia
    You might want to consider changing the instructions to include one which says to fill the camera frame with the piece of paper. The results will be different if you do that than they will be if the piece of paper only fills a small part of the camera frame because you're further away. As it stands people are likely to take the shots from varying distances and get varying results and the poll will show nothing conclusive because of that. It's only worth polling if everyone is doing exactly the same thing.
  5. Klorenzo

    Klorenzo Mu-43 All-Pro

    Mar 10, 2014
    Thanks, I forgot to specify this. I updated the text. Another thing should be to stop down the lens a little to minimize the vignetting. Probably an out of focus shot is better. To be even more precise also WB should be fixed, even better with a measured value.

    Anyway my idea for this poll is more about expectation rather then actual results. Then I think is also useful to verify first hand.

    If you try to do this do not use fabric: the threads creates bright and dark details and each fabric can be a little different. EDIT: fabric works fine too, you just get a little wider peak in the histogram.
    Last edited: Nov 8, 2015
  6. Klorenzo

    Klorenzo Mu-43 All-Pro

    Mar 10, 2014
    Mmm, not many votes here. Why not?

    Maybe I should have set a prize for the winners :)
  7. wjiang

    wjiang Mu-43 Hall of Famer

    Isn't this just another way of thinking about why dark stuff needs negative exposure comp and vice versa?
  8. Klorenzo

    Klorenzo Mu-43 All-Pro

    Mar 10, 2014
    Yes it is. But it is presented in a more basic and straight form as a thought provoking exercise.

    I have the impression that most people read this thing about dark/bright subject, gray cards, compensation, etc. in some blog/book and store it somewhere in the mind as some "fun fact" of minor importance. Like: ok, fine, next time I'll shoot on the snow as noon I'll think about it.

    I mean, you expect the camera to take the correct pictures, don't you? And if it does not you are more prone to look for complex ad hoc explanations rather then basic ones.

    This was my personal experience until I did this experiment. And I almost never bothered to use exp compensation much, I thought about it as something you use only to get silhouettes, to vainly try to save backlit portraits or to get "hipster-ish" overexposed pictures. Even shooting in M mode I trusted the exposure indicator at the bottom. It was more common for me to add compensation based on the histogram rather then on the scene, of course aiming for the perfect bell shaped curve in the middle no matter what the subject was (a thing you can find in 90% of blog articles...).

    And I lived happily with my "middle gray"-ish pictures because we are so used to this that you simple do not notice it. Our brain "fixes" the exposure but the impact is lost.
  9. mattia

    mattia Mu-43 Hall of Famer

    May 3, 2012
    The Netherlands
    This is exactly where the EVF wins - real-time feedback on what the shot will look like. What I really wish cameras would do is a live histogram based on RAW so I can expose the highlights better, although I'm far less worried about exposing to the right than I used to be back when ISO 800 was noisy and ISO 1600 was just barely acceptable for any use whatsoever.
    • Agree Agree x 2
  10. Klorenzo

    Klorenzo Mu-43 All-Pro

    Mar 10, 2014
    Yes, EVF wins as soon as one pays enough attention and get used to see the actual tones on the tiny screen. Like with composition, level horizons, dead cows in the background...is all there when you take the shot but often one does not notice, especially if you do not expect to have to worry about it.

    See current results, votes are quite spread.
  11. Klorenzo

    Klorenzo Mu-43 All-Pro

    Mar 10, 2014
    Ok, here is what I found out doing this experiment a few times.

    You get, as many of you has guessed or knew, four identical pictures. With identical I mean that you have to look at the EXIF shutter speed to tell them apart.
    These are mixed black, white, gray shots:


    To say it in a different way you can take any uniform colour thing, place it anywhere and you'll get the same picture/histogram (after desaturation). Red, green, gray18, at night, outdoor, etc. The camera has no problem at all doing this by varying the exposure time or other parameters. And it really wants to do it.

    I did this with the E-M10, Pana LX-5 and Canon 5D. Only with the Canon there was a slight variation in the black vs white pictures, I suspect due to the strong vignetting.
    The main difference between these cameras was the position of the peak in the histogram: with the Oly and Canon it was a little on the dark side, while the Pana was exactly in the middle. This on OOC jpegs. I do not know where the peak is supposed to be, if right in the middle or not.

    Another similar experiment is this: find a uniformly lit scene, point it somewhere randomly but do not include the sky or strong light sources in the frame. Set the camera to Manual mode, ESP/Matrix/Multi-zone metering mode and tweak it until the exposure indicator at the bottom is right in the middle. Now slowly move the camera around and notice how much the exposure indicator moves, last time I did it went from +0.7 to -2.3, about 3 stops. Given the uniform light this variation is only due to the different reflectance of the subjects. Ideally there should be no variations, this means that only one out of those ten values was "technically exact".

    In a real scene with a lot of of many small various elements the chance to have a "lucky average" is high, but when you have big patches of things you can clearly see how much the exposure gets shifted depending on the framing. Often the light is also not uniform. And light sources, like the sky, can complicate things.

    Then, of course, what you choose to expose for is personal, is a creative choice, etc. but is nonetheless influenced by the basics of the camera metering.
    Last edited: Nov 18, 2015
    • Informative Informative x 2
  12. PakkyT

    PakkyT Mu-43 Top Veteran

    Jun 20, 2015
    New England
    Dang, I failed the test. Back to school.
  13. oldracer

    oldracer Mu-43 All-Pro

    Oct 1, 2010
    Thanks. I would have expected them to be a little less identical just due to metering nonlinearities (as I said in post #2). I guess I was too pessimistic. Nice to know.
  14. Klorenzo

    Klorenzo Mu-43 All-Pro

    Mar 10, 2014
    First time I did this I expected the same, I had to repeat this one more time to be sure. It was like: no way a black and a white thing can look the same.
    And the interesting thing is that this, of course, happens with the mighty Manual mode too if you follow its indications. You really have to say: no, wait, let me do it, trust me.
    The AEL also made more sense: it is quite frustrating to use compensation when the exposure changes every time you move or slightly change the framing.
  15. oldracer

    oldracer Mu-43 All-Pro

    Oct 1, 2010
    Here's a totally useless thought: The experiment wouldn't work if the "black" target was truly black -- zero incident light being reflected. The meter would call for an infinitely long shutter speed at any aperture, but with no photons coming off the target, no exposure would be sufficient to get a gray result. The experiment only works because the black target is actually reflecting a lot of light.

    I learned photography with Nikon F meter prisms, which center weight the reading and that's still the only metering I use in digital cameras. I avoid spot metering in the camera because it's only useful in very special cases and will screw me up big time if I forget to turn it off. I avoid the fancy multi-metering modes because I have no idea what they are picking out to meter. Occasionally I will meter an area in the center weight setting, then lock the exposure and reframe. But that and occasional bracketing is about as fancy as I have found to be necessary. If I shot a lot in a studio setting with continuous light sources, I would probably buy a hand-held spot meter. I would still not use the in-camera spot.
    • Like Like x 1