Which is better: Higher ASA or Longer Exposure?

Discussion in 'Digital Photography' started by Gary Edstrom, Dec 20, 2004.

  1. Gary Edstrom

    Gary Edstrom Guest

    Given that object and/or camera movement is not an issue, which gives
    better results in a low light condition: Higher ASA setting, or a longer
    exposure time?

    Thanks, Gary
    Gary Edstrom, Dec 20, 2004
    1. Advertisements

  2. What are you trying to achieve? A higher ISO setting will give more
    noise, a longer exposure may require dark frame subtraction.

    David J Taylor, Dec 20, 2004
    1. Advertisements

  3. I would guess longer exposure is better.

    The ISO span is not large. If you use a consumer camera
    then 100 is good, 200 is acceptable, 400 is disturbing
    and 800 is unacceptable. The difference between 1/8 second
    and 1 second is not that large. A DSLR has a larger span
    (200-3200 or so), but I assume that it still holds.

    Roland Karlsson, Dec 20, 2004
  4. Gary Edstrom

    Jon Pike Guest

    Look on google for "reciprocity failure"
    and compare it to "iso noise"
    Jon Pike, Dec 20, 2004
  5. Because this is a digital group, I'll assume you are
    referring to digital cameras (if so ignore the other poster
    citing reciprocity failure, as it is not relevant).

    The trade with high ISO is higher signal with increased
    noise maybe requiring a shorter exposure versus low
    ISO with lower signal requiring longer exposure but
    with the possibility of increased dark current noise,
    and with some sensors, higher "popcorn" noise (CMOS
    sensors often have this problem).

    To get the ideal solution, you must think
    outside traditional exposure versus ISO trade space
    and consider another possibility with digital:
    averaging multiple frames to increase signal-to-noise.
    Astronomers are doing this quite effectively.
    The trade space then includes how many exposures
    for a total exposure time. For example, for 60 minutes
    total exposure, is ten 6-minute exposures better
    than 20 3-minute exposures, and at what ISO?

    Finally, temperature plays an important role
    because electronic sensor noise is very temperature
    dependent. So we have ISO, exposure time, number of
    frames, and temperature as the variables.

    I've explored this problem for the Canon 10D camera,
    and the results are at:

    At 70 F, I find you get the best results (highest signal
    to noise images) by taking 1-minute exposures at ISO 1600.
    At ~50 F, the exposure increases to 2 to 4 minutes, with
    ISO 400 doing about as good as ISO 1600. At 33 F, the
    time increases to 4 to 6 minutes per exposure.

    Thus at 70 F, if you need 60 minutes of exposure at ISO
    1600, it is best to do 60 1-minute exposures and add them
    (this all assumes 12-bit raw file recording).
    If you changed to ISO 400 and exposed for 60 minutes
    with 30 2-minute exposures, your signal-to-noise would only
    drop by about 10%. But if you did 20 3-minute exposures
    at ISO 100, your signal-to-noise would be only about 60%
    that you could get with the ISO 1600 case.

    So, to answer your original question, higher iso and shorter
    exposure time, but you must do lots of them and add (or average)
    them together to get the best results unless it is very cold,
    in which case, longer exposure, but still at the higher
    ISOs. The trends may be very different for other cameras,
    especially higher noise P&S cameras. Other DSLRs are probably

    Photos at: http://clarkvision.com
    Roger N. Clark (change username to rnclark), Dec 21, 2004
  6. Gary Edstrom

    Alan Meyer Guest


    Your study is wonderfully thorough. Even apart from its
    conclusions, and as a non-astronomer, I found it interesting to
    look at just to see how you've analyzed a difficult problem,
    reduced some pretty abstract data into meaningful terms,
    thought about what the variables are, and normalized the
    measurements for easier comparison.

    But I'm wondering if its application to ordinary, non-astronomical,
    situations is limited. By shooting with the lens cap on, I presume
    you were trying to find out when photographs of the empty parts
    of the night sky might appear to be non-empty due to camera
    artifacts - and how to minize that one problem.

    In non-astronomical situations, the amount of light will normally
    be greater, the "blacks" will not be as black since most surfaces
    reflect something, and the typical shutter speed ranges might be
    from fractions of a second to a few seconds. Also, the main
    issues in the final results may not be related to dark noise. They may
    have more to do with faithful reproduction of colors, or getting
    a useful and representative range of luminance and color.

    In the non-astronomical situations, do you think your analysis of
    the ISO / exposure time tradeoff still holds?

    Alan Meyer, Dec 22, 2004
  7. Alan,
    Yes, I believe they apply equally well to non-astronomical
    situations. The reason is that the electronic sensors are
    linear and do not suffer from reciprocity failure like film.
    My tests quantify noise at its lowest level. If you take an
    image and analyze brighter portions (e.g. anywhere it is not
    totally black), the signal-to-noise will simply be higher.
    In these higher signal areas, the trade space parameters
    would then be less sensitive to the ISO, exposure time,
    and temperature effects, but the darkest portions of the
    image will follow what is shown in my tests.

    Thanks for the compliments.
    Roger N. Clark (change username to rnclark), Dec 22, 2004
  8. Eh? Solid state sensors have no reciprocity failure.

    Roland Karlsson, Dec 22, 2004
  9. Gary Edstrom

    Jon Pike Guest

    really? got some evidence of that?
    Jon Pike, Dec 22, 2004
  10. But how do ISO 200 and 1/8 second compare to ISO 1600 and 1 second?
    And even more interresting how do ISO 200 and 1/800 compare to ISO 1600
    and 1/100 second?

    Roland Karlsson, Dec 22, 2004
  11. Have some evidence of the contrary?

    Roland Karlsson, Dec 22, 2004
  12. Oops - invert those numbers :)

    ISO 200 1 s vs iso 1600 1/8 s
    ISO 200 1/100 s vs iso 1600 1/800 s

    Roland Karlsson, Dec 22, 2004
  13. Gary Edstrom


    Pike has a learned a few technical terms and has put them together in a
    sentence without any idea whatsoever as to their meaning. This is
    classic negative intelligence at work:


    "A crackpot, however, seems to grasp a subject less and less the more
    , Dec 22, 2004
  14. There will be more noise in the higher ISO shots because of the smaller
    number of photons required for full exposure (and perhaps more noise
    because of the consequential higher gain between the sensor and the A-D).
    Few photons means more variance in the number of photons as a fraction of
    the total (photon-limited noise). This noise can be reduced by averaging
    multiple frames.

    However, there may be more noise in the longer exposure shots because of
    higher dark current noise in the sensor. This will depend on the
    temperature of the sensor, amongst other things. Some of this noise can
    be removed by dark-frame subtraction.

    David J Taylor, Dec 22, 2004
  15. Roland,
    Assuming dark current is small at these exposure times, then
    the lower ISO would might a better image. But if the signal
    is really low, so that at 1 second the image is still dark,
    then increasing the ISO would help. An important
    parameter to know for the camera you are using is:
    Gain in electrons/DN (DN = output data number),
    Read Noise (in either DN or electrons),
    dark current noise (e.g. in electrons /second),
    full well capacity (in electrons)
    (too bad manufacturers do not give this information).

    This page shows some gains, full capacities, and read noise
    values for several cameras:

    David's response is correct if you are photon noise limited,
    and in the <~ 1 second range, I would expect most DSLRs would be
    photon noise limited. Say your camera gave a full-well
    image in one second at ISO 1600 with a Canon 10D camera.
    The gain is 0.7 electrons/DN, so you are getting all
    the information the camera will offer and noise is
    photon noise limited. But at ISO 200,
    your signal in one second will be 1/8 the intensity it was
    at ISO 1600, and because the camera gain is 5.5 electrons/DN,
    the camera does not record all the intensity details--it loses
    some in the DN quantization. So the image would not be
    a good. Note however, that due to posterization effects,
    in some areas the image might look smoother, but not really
    due to less noise.

    If, however, you can expose the same above scene at
    ISO 200 for 8 seconds, then the image should be much better
    than the 1-second ISO 1600 image (again assuming dark current
    noise does not begin to dominate, which it won't on Canon
    10D, 20D, rebel, 1DII, and probably not on the Nikon D70).

    Roger N. Clark (change username to rnclark), Dec 22, 2004
  16. There is excellent evidence from the astronomy community
    that modern digital cameras do not suffer from reciprocity
    failure. Exposure times, including how many photons are
    collected can be calculated with good accuracy for very
    long exposure times. Photometry and spectroscopy are being
    done with digital cameras, and no reciprocity issues have
    been reported using short to very long exposure times
    (<1/1000 second to tens of minutes).

    Roger N. Clark (change username to rnclark), Dec 22, 2004
  17. Gary Edstrom

    Jon Pike Guest

    still touting your own 'research' as though it's done in a competant
    that's amusing :)
    Jon Pike, Dec 22, 2004
  18. Gary Edstrom

    Dr Stone Guest

    Still link-pimping for your little imaginary animal companion?
    That's pathetic. By the way, did you take down your Web
    archive? It was blank when I tried to read about your mail-
    order bride. Oh, and you misspelled "competent" in your
    sneer, which is kind of an instant Karma hoot!
    Dr Stone, Dec 23, 2004
  19. Yepp - it is all very good to be sceptic. But to just
    say "really? got some evidence for that?" when it is a well
    known fact I found somewhat annoying.

    Roland Karlsson, Dec 23, 2004
  20. Roger is competent. I don't always agree - but thats life.
    I consider myself rather competent also - but I might be wrong
    sometimes nevertheless.
    The smiley there is totally misplaced. You don't use smileys
    when you are dripping of sarcasm - only when you are being

    Roland Karlsson, Dec 23, 2004
    1. Advertisements

Ask a Question

Want to reply to this thread or ask your own question?

You'll need to choose a username for the site, which only take a couple of moments (here). After that, you can post your question and our members will help you out.