Help me choose my first dslr

Discussion in 'Digital SLR' started by Padu, Aug 14, 2006.

  1. Wolfgang Weisselberg, Aug 28, 2006
    1. Advertisements

  2. In theory, but one can simply make sure you design the thing with enough
    bits to handle the range of inputs it will actually be given.
    Presumably the output range of the A/D converter on the sensor is a
    known quantity, as is the range of ISO settings handled.
    OK, I guess that does technically count as "predictable" when looking at
    the big pictures. I was talking about individual pixels on a single
    shot, though
    It can influence one's *perception* of noise. That is, you might not
    notice the noise in a severely underexposed picture, but push it to the
    point where the values are closer to, say, 18% gray, and you're more
    likely to notice the noise. At least, that's been *my* experience.
    Absolutely; I agree with this. But I think the *reasons* for this are
    slightly different than presented. I'd make the case by appealing to an
    extreme example. Say you've underexposed so severely that you never
    have a chance to collect more than 1 photon per site. Your sensor could
    do this perfectly, and have a perfect A/D conversion to yield "0" or "1"
    as the value of each pixel prior to de-mosaicing. You could apply
    absolutely perfect digital multiplication to push this up into the
    maximum dynamic range supported by the RAW file. The resulting image
    would *still* look terrible. The issue isn't that the digital
    multiplication added noise, the issue is that it was given
    information-starved data to work with.

    Marc Sabatella

    Music, art, & educational materials
    Featuring "A Jazz Improvisation Primer"
    Marc Sabatella, Aug 28, 2006
    1. Advertisements

  3. Yes, that's what I said in the first place.
    Not sure what you're asking here. In painting, you simply don't paint
    all the detail you see. It takes a considerable amount of training and
    discipline to decide where to paint detail and where not to - and how
    much dynamic range to use in painting the detail that does get painted -
    in order to create a painting that is perceived in the art world as
    "convincing". In photography, if you simply shot the scene as is and
    did no processing, you'd get results that many artists would consider
    excellent. But it seems many photographers are into this "high dynamic
    range" thing where they deliberately go to great lengths - either
    through the use of shining fill lights into the shadow areas when
    shooting, or through post-processing - to create images that a painter
    would find much less convincing and aesthetically pleasing than the
    original version.
    Artists don't usually think in terms of "stops" - instead, we use a 10
    or 11 step "value scale", where 0 or 1 represents black, 10 white (some
    reverse the sense of this). A typical painting would use only values
    1-9 at most - pure white and pure black are both pretty rare - but most
    paintings would indeed use much if not all of that 1-9 range. The trick
    would be how you map those values to elements in the scene. The typical
    mistake beginners would make is to have the darkest thing in the scene
    be 1, the lightest thing 9, and scale everything else evenly throughout
    that range, so detail is spread evenly throughout the painting at the
    expense of the illusion of light. Much more effective is to decide to
    concentrate your detail in either the lights *or* the shadows. If you
    elect to have the detail in the lights, you might decide that values 3-9
    will be for the light areas of the painting, and values 1-2 for the
    shadow. You paint more detail, and have a larger dynamic range
    available, in the lights, and deliberately compress the dynamic range
    and simply detail in the shadows. Or the other way around.
    I'm not familiar with zone system terminology, but I'll assume that is
    roughly equivalent to the idea of having an object represented with
    little detail and little dynamic range at the extremes of the value
    range. I'll simply observe that while photographers may value that
    choice, in my experience, most of the time, they will choose *not* to do
    so, and indeed, will go to some lengths - including the use of fill
    lights and post-processing for HDR - in order to *counter* an effect a
    painter would have been likely to choose to *exaggerate*.

    Marc Sabatella

    Music, art, & educational materials
    Featuring "A Jazz Improvisation Primer"
    Marc Sabatella, Aug 28, 2006
  4. Since the sensor is "unpredictable", looking for 100%
    predictability is locking the barn after the horses got out.
    "Well, there may be no noise visible, but, aesthetically,
    black on black images have been ... outré for years."
    You're also likely to actually *see* what's in the picture.
    OK, make the counter-experiment:
    Your sensor and A/D-converter are again perfect, but the
    booster introduces 1% noise. Since you get 0000 0000 0000
    +- noise and 1111 1111 1111 +- noise, it`s trivial to remove the
    noise. And it would be possible even up to noise up to half
    the converter range.
    In that special case, the problem isn't noise. We have to
    look at the real world, with non-perfect sensors and
    non-perfect A/D-converters.

    But for that, look at
    for really low-light conditions.

    Wolfgang Weisselberg, Aug 31, 2006
  5. Please respect the context. You were talking about
    photograpy and painting, *I* was talking _only_ about
    Assume a scene. 3 Rocks. A dark one, a snowcapped white
    one and a middle-gray one.

    I photograph them. I can choose:
    - Paint the dark rock completely black (because otherwise there'd
    be shadow detail in it's texture) --- and maybe treat the darker
    parts of the middle-gray rock the same.
    (no, I need the white to be white and then the dark parts
    will still show detail)
    - blacken them in post-processing (photoshop etc)
    - accept that there _is_ shadow detail in the scene
    That's a painting. Not a photography.
    .... or through multiple, bracketted exposures and merging ...
    Photography is a different medium from painting.

    A painter painting a night scene in the city could:
    - paint only the street lights in detail and leave the rest
    - paint the street lights in pure white, obliterating detail
    and getting some detail outside the pure shadow
    - flatten the contrast (like a photographer might)
    - _not_ paint this.
    flattened contrast/blue hour
    little but discernible shadow detail
    clear shadow detail
    flattened contrast or interior lighting, for else you'd loose
    the interior details (wall would be black, easel, backlit
    grille, ... ). Basically, someone used a flash here :)
    clear shadow detail
    lots and clear shadow detail
    look at the tree/bushes ...
    quite like a photograpy, don't you agree?
    look at the dark graduation at the right side, for example
    Even the trees have shadow detail! (use curves on it to see
    it clearly)
    While the houses are black (except for the windows), but look
    at the sky.

    .... and many more.
    Look at the zone system. 0: blackest black, X (i.e. 10): paper white.
    sounds like most photographs ... look at "normalizing".
    i.e. playing with curves.
    Aeh, no.
    It's a method of pre-visualizing density, graduation and texture
    visible in the final print.

    It uses a 11 step scale (using roman numerals), where V is
    middle gray, and every step up or down is a _stop_ up or down
    (though that will be stretched or compressed in development (or
    in printing with the choice of the grade of the paper (or using
    multi-graded papers)) to get the contrast you need.

    It is based on black-and-white printing (but that doesn't
    mean it cannot be used for digital color photography with
    some adjustements for the technology.)

    The extreme ranges (0 and X) are pure black/white.
    The extreme but one ranges (I and IX) are showing light tonality,
    but no texture. (which would be somewhat hard for paintings:
    the brush stroke gives texture, for example)
    The extreme but two ranges (II and VIII) get tonality and first
    The extreme but three ranges (III and VII) show average dark/white
    subjects with full texture
    The middle ranges (IV, V VI) are, well, the middle, with V being
    18% gray,

    Between each range there's one fstop.

    Pages like
    seem to explain it well.
    So? You are accusing photographers to not be painters and to
    use a different language? To try to use new technique? I guess
    expressionism/cubism/abstract paintings were not hailed as the
    greatest thing since sliced bread at their beginning, either.

    And because *you* are so unhappy with that, it's not a problem
    if digital cameras have bad shadow detail?

    Wolfgang Weisselberg, Sep 1, 2006
    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.