Moon shots

Discussion in 'Digital SLR' started by Sheldon, Feb 26, 2005.

  1. Sheldon

    Alan Browne Guest

    You will go a good ways with the general notion that the lowest ISO for
    the conditions is the least noisy ISO, so stick to that until you want
    to squeeze out some more performance.

    Cheers,
    Alan
     
    Alan Browne, Mar 3, 2005
    #81
    1. Advertisements

  2. Sheldon

    paul Guest


    Yes, understood.


    Yes this could be done and would make sense. Again, it's not dependant
    on ISO, it could be just pushing the exposure within comfortable
    shooting speeds with the intent to darken later. In fact, it could be
    the camera's default setting to simply expose up to the max before
    blowing highlights and record that adjustment in the exif for the raw
    converter to readjust back to a normal exposure. Hmm, well, maybe there
    is another dial to add when you want. If you have a higher speed than
    you need, you would then know the "exposure cheat" factor could be
    dialed up until the speed becomes too slow. But if you are thinking with
    that in mind from the start, you would only disable "exposure cheat"
    when the speed got too slow. Then, you start adjusting ISO, then step
    off the "exposure cheat" setting.


    That's not what the test images you showed did though. I'd like to see
    this demonstrated <g>.
     
    paul, Mar 3, 2005
    #82
    1. Advertisements

  3. Sheldon

    Owamanga Guest

    The gain is increased in Photoshop with levels. (that's what I meant
    by 'push back in photoshop'. Sure, if you left it underexposed, it'd
    be a dark image with no extra noise.
    Waiting anxiously...

    People are feverishly working on it.

    It'll add some or all of these:

    ISO 100, 3200 and 6400.
    Something they call 'Composition priority' mode, where the ISO floats.
    Chanel-split color-coded histogram, so you can see each channel
    individually.
    Chanel-based color-coded highlight warning mode.
    Multiple user-curves.
    RAW+F (Raw + Fine Jpeg) mode.
    Astronomy RAW mode (eg skips the blurry hot-pixel algorithm)
    MLU & Auto-MLU
    Partial sensor (1.2Mbit) 60fps NEF movie mode.
    Flash sync to 1/8000th

    Plus you'll get to assign your own rules for each of the 'idiot' modes
    (portrait, landscape, sport etc) as to what you want it to do. This
    might be as simple as each idiot mode remembering it's ISO, aperture
    vs shutter vs composition priority, aperture and shutter speed ranges,
    assigning what the front and back command dials do, focusing mode and
    metering mode, flash sync mode etc.

    More here:
    http://groups.yahoo.com/group/nikond70hak/
     
    Owamanga, Mar 3, 2005
    #83
  4. Wow, something like this for the 20D would really be awesome.
     
    Ben Rosengart, Mar 3, 2005
    #84
  5. Sheldon

    Owamanga Guest

    Actually, Canon generally have a better history of firmware hacks than
    Nikon, in particular the EOS300D:

    http://satinfo.narod.ru/en/index.html
     
    Owamanga, Mar 3, 2005
    #85
  6. Sheldon

    Alan Browne Guest

    The point is that by doing so (as you well know), at taking time you're
    not filling the MSB, so you're not using your sensor there (and you're
    pushing the (albeit less useful) bits below capture). Loss of detail in
    the shaddows is not going to re-emerge in pushing ... the LSB's are
    filled with 0's.

    You will 'get the shot' and that's important (most important) but don't
    hope to get the full dynamic range of the sensor.
    Composition Priority[TM] is Trademarked.
    (I'm sure you threw that in there for good reason ;-) )
    As JPS suggested "CameraBASIC" would be the ticket.

    Cheers,
    Alan
     
    Alan Browne, Mar 3, 2005
    #86
  7. Sheldon

    Alan Browne Guest

    Owamanga wrote:

    What are the warranty issues? Suppose your body dies with a hacked load
    and you can't put back the original code. Gives the warranty center a
    blank check to say, "sorry, no can do". They can claim that the hack is
    the cause of the damage (and they may even be correct).
     
    Alan Browne, Mar 3, 2005
    #87
  8. Sheldon

    Owamanga Guest

    ...so we won't see noise? - just a crop of the shadows?

    Dunno if we are agreeing or disagreeing here..

    This is my opinion: A 2 stop underexposure with subsequent re-leveling
    to bring the exposure back in PS will have more noise than a two stop
    increase in ISO and correctly exposing. So you cant really emulate
    higher ISOs by underexposing.
    Indeed, we'll be wasting about 75% of the histogram.

    Correct.

    Alvin Chia-Hua Shih used the term in the realm of Photography back in
    1994. BTW, to hold a trademark for any period of time, you have to
    demonstrate that you are using it (countries differ of course, can't
    speak for Canada). We get letters from the US Trademark office every
    few years chasing up our ones.

    Your use is a better fit for the phrase than Alvin's.
    Indeed.
     
    Owamanga, Mar 3, 2005
    #88
  9. Sheldon

    Alan Browne Guest

    You'll see the noise as it occupies higher bits, and is pulled higher in
    the pixel color channel, below it, the data will fill with 0's (for a
    doubling in the intensity of that pixel, in PS, of course, when you
    simply 'haul' up the levels it is not neccesarilly a doubling or 4x..)

    (eg: when you amplify noise you get *amplified* noise.)
    We're agreeing. (I think).
    Well, when you do the re-leveling on an underexposed image, you are
    amplifying noise as well as signal, but you're also really working with
    less information. The information that pads underneath is not noisy per
    se, it is just "non-information".

    3 stop under example (no R,G,B pixel in the entire image has an on bit
    in bit positions 11, 10 or 9. So let's grab red (or green or blue)
    pixel somewhere and play with it:

    Here's an under exposed 12 bit red pixel: (bits 11 down to 0).
    1 9 0
    000101011101

    Increase the exposure by 1 stop (0 pad the LSB)

    1 9 0
    001010111010

    Another stop

    1 9 0
    010101110100

    Another
    1 9 0
    101011101000

    So, (and assuming we haven't clipped blue or green), we've brought the
    signal up 3 stops, but the lowest order bits have been 0 paded. (You
    could put in random 1's or 0's if you like).

    Leveling of course doesn't go neccesarilly in doublings, so the padding
    will be whatever products result... but they are non-information from
    the original image ... they are just numerical artifacts. No
    information has been lost, but certainly NONE has been gained either.

    The point? Is that that lowest order bit [0] in the first instance
    (which can be assumed to be noise) has now moved up to bit 3, or 8 times
    the 'signal' it had before. Quantization noise/shaddow detail blocking
    occurs.
    As the case may be.
    I am humbled by your recognition and it is appropriate to state that it
    was one of your posts that triggered the phrase (the thought's been on
    my mind for a long time, there are just situations, such as studio
    shooting where it would be of negative value).

    I did a google and the term does not come up in a photographic sense.

    Cheers,
    Alan
     
    Alan Browne, Mar 3, 2005
    #89
  10. Sheldon

    JPS Guest

    In message <>,
    Oh yeah, right; I suppose that just parrotting "use the lowest ISO that
    you can get away with" like most everyone is sounder scientific method?
    Maybe part of the problem is with the reader. Many people are incapable
    of thinking outside the box they're in, and interpret things in a
    distorted way to bring them back inside the box, no matter how clearly
    stated. I can't control what assumptions people come to the table with.
    My posts are often like the little pink or blue boxes in a text book
    with the more detailed truth for those who care to know. Until you can
    understand the ramifications of these truths, fell free to "use the
    lowest ISO that you can get away with".
    --
     
    JPS, Mar 4, 2005
    #90
  11. Sheldon

    JPS Guest

    In message <>,
    In theory, yes, but the noise standpoint is not the only one to hold.

    The images may be blurred. Exposing at ISO 25 in dense fog for
    hand-held pictures is very compromising. You either have to shoot wide
    open, compromising the optics and DOF, or use long exposures, which may
    cause motion-blur.

    If you had the camera set to ISO 100, and were in dense fog, with the EC
    set to 0, you might suddenly think "expose to the right", after seeing
    the narrow, mid-screen histogram hump. What I'm saying is that you can
    leave the aperture and shutter speed where they are, and just move the
    RAW data to the right, by using a higher ISO combined with positive
    exposure compensation.
    I don't think so. I think that the confusion comes from the inefficient
    paradigms already in place.

    Exposure on the sensor comes from subject lighting, aperture, and
    shutter speed, alone. (Gain-based) ISO is just a way to map that
    exposure to numbers, by scaling with amplification.
    --
     
    JPS, Mar 4, 2005
    #91
  12. Sheldon

    Matt Ion Guest

    There is no dark side of the moon, really. Matter of fact, it's all dark.
     
    Matt Ion, Mar 5, 2005
    #92
  13. Sheldon

    JPS Guest

    In message <[email protected]>,
    Explain to me why you're mad.
    --
     
    JPS, Mar 6, 2005
    #93
  14. Sheldon

    paul Guest


    Even if you are not mad.
     
    paul, Mar 6, 2005
    #94
    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.