Discussion in 'Digital SLR' started by Greg G, Feb 7, 2005.

  1. Greg G

    Greg G Guest

    I'm confused.

    I understand what the ISO number means when it comes to film, but what
    about a digital sensor? Is it just a change in electronic brightness
    "gain"? If so, can I assume there is more "noise" associated with the
    higher numbers? If I choose a higher number will I have gained
    anything over processing the photo with image editing software later?

    Let me ask it another way: I assume that if you choose a higher number
    you'll get a shorter shutter speed and/or wider aperture for a given
    amount of light, but since the physical sensor is the same what has
    actually changed?

    A related question:

    In general terms how well do entry-level DSLRs do in low light
    compared with film? If I "choose" ISO 800 do I end up with roughly the
    same shutter/aperture combination that I would get with 800 speed
    film? How is the "noise"?

    What about long exposures? Reciprocity failure notwithstanding, I've
    gotten some of my nicer travel shots at night with exposures of 5
    seconds or more on film. How does a DSLR compare?

    I am of course asking for "general case" answers. I assume different
    models have different characteristics.

    Greg Guarino
    Greg G, Feb 7, 2005
    1. Advertisements

  2. Greg G

    Crownfield Guest

    your light meter reads light for either film or digital.
    iso 800 is iso 800.
    Crownfield, Feb 7, 2005
    1. Advertisements

  3. Greg G

    Skip M Guest

    ISO is ISO, the sensitivity of the sensor at any given ISO is equivalent to
    the sensitivity of film at the same ISO.
    Noise from digital tends to be less than grain from the same ISO film,
    Entry level cameras are fairly good in low light, depending on what you mean
    by "low light." And they're better now than the midrange ones were just a
    couple of years ago.
    Digital will certainly get you good shots at shutters speeds as long or
    longer than you went with film.
    Skip M, Feb 7, 2005
  4. Greg G commented courteously ...
    Yes. But its higher shutter or *smaller* aperture, or some
    combination of both, since each controls steps of 1/2 the
    light coming into the "focal plane" - i.e., the sensors.

    Short answer with my limited knowledge, it's called
    "signal amplification". The sensors don't get any more
    sensitive but you can boost the signal to do more with a
    given amount of photons or lumens (whichever you like).
    All other factors being equal, but they seldom are, more
    boost - i.e., higher digital ISO - the more noise.

    OK, everyone, fair game on Jerry for being a twit! <grin>
    All Things Mopar, Feb 7, 2005
  5. Skip M commented courteously ...
    By definition of ISO, right? Right.
    I don't have nearly enough modern-day film or digital
    experience to dispute you, and I'm not. But I'm usually
    suspicious of blanket statements that "noise fromdigital
    tends to be less than grain from the same ISO film".

    I'll make my assertion as a question: aren't there
    sometimes significant differences from one film
    manufacturer to another at any given ISO? And, aren't
    there sometimes significant differences when moving
    between ISO ranges even within the same film manufacturer?

    And, for digital, doesn't noise as well as image quality
    have quite a bit to do with the quality of the sensors (as
    measured by their cost), as well as if you always want to
    be at the bleeding mega pixel edge, you're going to suffer
    the most noise until a given manufacturer "catches up" and
    builds "quieter" sensors? I didn't expect my wife's $150
    Kodak to be as noise free, nor have as good image quality
    as my $700 (retail) Nikon 5700 (of course, the 5700 also
    has superior features).
    All Things Mopar, Feb 7, 2005
  6. Greg G

    Skip M Guest

    That's why I said "tends" not the narrower "is." Earlier digitals, like the
    D30, had very high levels of noise at high ISOs, especially 1600, much
    higher than, say, Ilford Delta 3200.
    Yes, there are, again why I tried to avoid using dogmatic phraseology.
    Yes, but. The Canon 16mp 1Ds mkII has a pretty low level of noise, probably
    one reason is costs so much. And the OP was asking about DSLRs, which
    eliminates much discussion of noise vs. sensor dimensions, since, with the
    exception of the Olympus 4/3 system, sensors range between APS and 35mm
    size, and most of the sensors are of comparable quality.
    Skip M, Feb 7, 2005
  7. There is a discussion about this very topic over on rec.photo.digital

    It seems that either the definition or the concept is not appropriate for
    digital cameras because the image quality isn't taken into account. I can
    have a point-and-shoot camera and a DLSR both work at ISO 800, but the
    image quality will likely be much better from the DLSR.

    If you take image quality into account, then you could perhaps say that a
    point-and-shoot has a usable sensitivity of ISO 50 - 200, but that the
    DSLR has a sensitivity up to ISO 1600, for example. Alternatively I guess
    you could say that DSLRs at ISO 1600 were like ISO 200 film.

    What I'm getting at is that although from the point of view of calculating
    exposure ISO is useful, it doesn't reflect the image quality of different

    David J Taylor, Feb 7, 2005
  8. Yes, it's electronic gain of the analogue sensor signals before they are
    digitised. Yes, increasing ISO will increase noise: on small-sensored P&S
    cameras, this can be a major problem, but for dSLRs, with physically larger
    sensors, it's generally not too bad (and _generally_ better than high ISO

    The advantage of doing it in-camera with an ISO boost is that the gain
    happens before digitisation. To illustrate the benefit, suppose the
    (analogue) sensors generate from 0V (black) to 1V (light) and that this is
    digitised to 0..100 (numbers are made up to make it easy to understand).

    If you shoot a low-light shot with no ISO-boost, then the sensors might
    only produce values between 0 and 0.1V, which will be digitised into pixel
    values 0..10 (out of 100). If you post-process by scaling pixels by a
    factor of 10, then while the brightest pixel will now be 100, you will ONLY
    have pixels with values 0, 10, 20, ..., 80, 90, 100 -- there will be no
    values "in between", so the image will show "banding".

    If, however, you bump the ISO so that the analogue signal is boosted 10x,
    then the 0 to 0.1V signal becomes a full-range 0 to 1.0V signal. When this
    is digitised, you get a full-range 0..100 pixel range, with (potentially)
    all shades in-between (plus some more noise).
    Digital sensors don't suffer from the reciprocity failure that film does,
    which makes long exposure shots easier to work with (as I understand; I've
    never done this type of shot [yet]).

    As well as "amplifier noise" (above, from high ISOs), digital sensors can
    suffer from noise at long exposures. Many cameras will offer an option to
    perform "dark frame subtraction" for long exposures (an equivalent length
    "dark" shot is taken after the real shot and subtracted from the first).

    Graham Holden (g-holden AT dircon DOT co DOT uk)
    Graham Holden, Feb 7, 2005
  9. Greg G

    Alan Browne Guest

    Enjoy it for a while.
    As others have replied, ISO is ISO.

    However film and digital *are* different in many ways so a side by side will
    have some differences.
    Notably, as film ISO goes up, grain noise AND size increases.
    Digital, only the noise increases. The pixels remain the same size.
    Yep. Reciprocity rules as always. (you meant 'smaller' aperture above as IOS
    rises, all other things being equal).

    What changed? At lower ISO's (say up to 1600), the gain prior to digitization
    is increased. Above that, simply shift the bits to the left (mult by 2) to get
    3200, 6400... beginning at that point there is a drastic quantization noise
    increase, however.... (better than no photo at all).

    Alan Browne, Feb 7, 2005
  10. Greg G

    John Francis Guest

    It wasn't taken into account on film cameras, either. Nor was sensor size.
    If I put a sheet of slide film in a large-format camera, and a roll of the
    same emulsion in a half-frame 35mm camera, I'd end up with a much better
    quality image from the large-format. But, in each case, I'd need to set
    the exposure based on the sensitivity of the film. That, and only that,
    is what ISO measures.
    John Francis, Feb 7, 2005
  11. Greg G

    Greg G Guest

    Of course. That's what happens when I don't proofread.

    Greg G, Feb 7, 2005
  12. Greg G

    Bob Guest

    we all are... sometimes!
    That's it! Gain, just like any video signal in any amplifier.
    Yes you can.
    That's an interesting question! I'd say get as close as you can with the camera
    first, since you may not be able to extract data in your software. It's like a
    negative, you can go one way several stops over, but only about one stop the
    wrong way, since there would be no emulsion left. Once you start 'clipping' the
    electric image, there is no return. So stay towards the shadows where you can
    recover. As to whether or not the camera or the external software is better -
    I'd bet on the camera. But there are usually 'highlight' warnings on cameras,
    so stay below that and fix the gamma in your external software.
    The gain of the signal amplifier.
    yes, that is how they calibrated the system. there is no such thing as ISO in a
    sensor since it is amplified as a system. The 'ISO' number is just to let us
    know what the camera would do if it were film.

    Now I have a question - why no 'ISO 100" on the D70??
    Going by my memory of when I shot film (decades ago I think...) I'd say I get
    less noise with the digital. I remember shooting 400 color film and it being at
    the edge of quality, but now I shoot at 800 all the time... and 1600 to get to
    the edge.

    Or course, different film and cameras - it depends!
    It depends on the camera... I've seen long times that were great. Go to the
    review site...
    Bob, Feb 7, 2005
  13. Greg G

    Colic Guest

    Yes, long exposures are possible with all DSLRs currently on the new market.
    Long has to be defined though. In general long is (for digitals anyway) 8
    seconds to many minutes. And they can all do that to some degree.

    If you are looking at doing up to about 30 seconds or so don't worry about
    it, pick the one you like for other reasons, because they are all pretty
    much OK to this point, some slightly better than others. If you want times
    of over 5 minutes shutter then you need to look at each one on a case by
    case basis. I think the current long exposure champ is probably the Canon

    Here is a 10 minute Canon 20D exposure, at 400 ISO. The originator of the
    this shot claims it is right out of the camera, and was originally shot as a
    small fine jpg.

    Colic, Feb 8, 2005
  14. Greg G

    DoN. Nichols Guest

    [ ... ]
    At a guess, it is because the CCD saturates and starts
    distorting or clipping the exposure curve with that much light.

    And which came first -- ISO or ASA? I knew the ASA long before
    I heard of ISO in film speeds. The only European film speed which I
    knew back then was the DIN rating, which certainly did *not* match the
    ASA of the time.

    I would *really* like to have ASA/ISO 100 (or even down to ISO
    50) for my old "Medical Nikkor" at the really close focal distances.
    (Yes, it does work on the D70 with the flash sync adaptor module in the

    Or -- I would like the necessary information to make a
    "resistor" cord from the flash power supply to the head in the lens.

    I'm rather reluctant to add a neutral density filter to the
    (potential) stack of close-up lenses which are part of that system.
    When using only a single close-up lens, I guess that I would be all
    right, but some of the particularly close ones are a stack of two, and
    the closest one has no secondary filter threads, so the filter would
    have to be closer to the prime lens than at least one of the close-up

    As it is, I have gotten some interesting close-ups to a funnel
    weaver spider (lion spider) who took up residence between the storm
    window and the main window of our bathroom. s/he got enough to eat
    because of the bugs which were attracted to the lights and got through
    the crack left at the top of the storm window.

    The spider has now passed on, and the webs are cleaned out. And,
    the storm window is fully closed this time. :)

    DoN. Nichols, Feb 8, 2005
  15. Greg G

    paul Guest

    Thanks for this excellent explanation. If I follow correctly, the way to
    check this is to get a 'full' looking histogram in the original capture
    rather than brighten it up later.

    On a somewhat related note; does anyone know if the exposure adjustment
    in PS CS RAW plugin amounts to a real change in exposure or more like
    any other software brightness control?
    paul, Feb 8, 2005
  16. It's even worse. Eyes are logarithmic (and so is jpeg), but
    the sensor are linear. Assuming you can record 5 steps of
    brightness (zone 3 to 7 inclusive), zone 7 would be 0.5-1V,
    zone 6 0.25-0.5V, zone 5 0.125-0.25V, zone 4 0.0625-0.125V and
    zone 3 0.03125-0.0625V. (and not much discernible below)

    Which would not be a problem, but for the fact that the sensor
    will record 0.03125V as 3, not as 3.125 ... There currently is
    no commercial viable way to make the sensor record 1..10000 for
    making up for the low voltage differences.
    And that mostly in the darker part, not evenly distributed,
    i.e. even more noticable.
    .... thus bringing the sensor into a favourable working point,
    where it can use it's whole resolution on the problem.

    Wolfgang Weisselberg, Feb 8, 2005
  17. DIN (Deutsche (German) Industrie Norm) is something you'll find
    in many places, e.g. DIN A4 (paper size).
    I assume the American Standards Association is similar in intent
    as an organisation.

    Pertaining to film:

    15° 25 25/15°
    18° 50 50/18°
    21° 100 100/21°
    24° 200 200/24°
    27° 400 400/27°
    30° 800 800/30°
    33° 1600 1600/33°

    So in DIN +1 is 1/3rd of a step.

    Wolfgang Weisselberg, Feb 8, 2005
  18. Greg G

    DoN. Nichols Guest

    I understand this, and am familiar with many DIN standards
    (mostly from the fields of electronics and metalworking). And I realize
    that many other exist of which I have no knowledge.

    But I was in particular speaking of the standards for film
    sensitivity. I grew up with ASA, and encountered DIN with certain
    imported films, such as Isopan Record.

    And I was then away from photography for many years, because I
    worked for a US Government organization in which a security clearance
    was necessary, and cameras were totally forbidden (even one just stored
    in my car), so my standard practice of carrying at least one camera and
    two lenses (50mm f1.4 and 135mm f3.5) was not possible, and photography
    fell into disuse.

    Now, I am retired. I still have too much to do to spend the
    hours in the darkroom which I used to do, but the Digital cameras deal
    with that quite nicely.

    But -- when I came back to photography, I discovered that ASA
    film speeds had been replaced with ISO film speeds, with apparently
    identical values (for familiar films such as Tri-X and Ektrachrome, the
    two which I used most often.)
    Certainly. As is the ISO, which I know from computer language
    standards and the such.
    That table would save me a bit of looking up, if I still used
    film instead of digital focal planes.

    DoN. Nichols, Feb 8, 2005
  19. Greg G

    m II Guest

    oh, the irony...<g>

    m II, Feb 11, 2005
    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.