High ISO noise CCD's vs CMOS

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

  1. mswlogo

    mswlogo Guest

    Ok, I'm in the market for my first DSLR (upgrade from 35 mm SLR and
    Nikon 5700).

    I've been looking at reviews on the Sony A100, Nikon D80/D200 and Canon
    30D.

    These are both 1600 ISO (see reviews for more detailed information
    about conditions etc).

    http://www.steves-digicams.com/2006_reviews/30d/samples/IMG_8337.JPG

    http://www.steves-digicams.com/2006_reviews/d200/samples/DSC_3490.JPG

    What a HUGE difference !!!

    I see so many people say only if you shoot high ISO is it a concern.
    Unless your shooting with lights in a studio can't any one use high ISO
    at times?

    If you can get 2 (or so) stops lower on every lens and get the same
    picture, why not take it? You'll pay a small fortune for 2 stops on a
    lens.

    Are there other compromises in going higher ISO (even in the low range)
    besides noise. Do you lose color accuracy etc.

    I really favor Nikon for feel and operation. But this ISO noise has me
    leaning towards canon CMOS. I also see numerous comments that Canon IS
    is better than Nikon VR. So that to me is worth probably another stop.

    So the canon seems to have like a 3 stop advantage over Nikon.

    I have no investment in lens (my 35mm was a Pentax and I'd never use
    those boat anchor manual focus lens again).
     
    mswlogo, Aug 14, 2006
    #1
    1. Advertisements

  2. mswlogo

    Mark² Guest

    There is little dispute that Canon's CMOS produces the cleanest ISO images.
    I wouldn't dispute, however, that the D200 has a build that I wish Canon
    would emulate.
     
    Mark², Aug 15, 2006
    #2
    1. Advertisements

  3. mswlogo

    C J Southern Guest

    Problem solved - buy Canon.
     
    C J Southern, Aug 15, 2006
    #3
  4. mswlogo

    ColinD Guest

    You'll get used to the feel of the 30D; you'll never be happy with the
    crap images from the Nikon.

    No contest. Canon.

    Colin D.
     
    ColinD, Aug 15, 2006
    #4
  5. Did you look at the exif? The Canon photo is at 1/250s f/2.8 while the
    Nikon at 1/100s f/7.1. (There is no ISO in the Nikon's exif data). Also
    the sharpness in the Nikon is set to "hard", which seems a less than
    intelligent thing to do when shooting at ISO 1600.

    There is a difference between the two cameras in terms of noise (I
    spent some time trying the 20D and the D200 when I was deciding), but
    it's by no means as much as you'd think from these two samples.
     
    achilleaslazarides, Aug 15, 2006
    #5
  6. mswlogo

    mswlogo Guest

    Thanks for catching this. The summary info in the review was a little
    short on info. The photos were not a fair comparison. By the way the
    site where those photo's come from (Steves Digicam) highly favors the
    Nikon D200.

    But I think it is true that the Canon is at least 1 stop better, and I
    keep reading excuses.

    One is (only if you shoot high ISO). But if you can shoot one stop
    better which often happens for what ever reason, that seems huge. Other
    folks have said the Nikon is more conservative on in camera noise
    reduction, but I've seen reviews do everything raw and still have
    higher noise.

    Build no question, the D200 is more solid. How it operates is purely a
    matter of taste.
    But I just can't get paste this ISO thing and I have not typically shot
    high ISO because it was so bad on my 5700.

    That's why I asked is there other reasons to keep your ISO low.
     
    mswlogo, Aug 15, 2006
    #6
  7. mswlogo

    ian Guest

    As far as build quality goes the 20D and 30D are more than adequate unless
    you are demanding weather sealing. EOS 5D owners jump in here?

    As for higher ISO the alternative is spending more on lenses with a larger
    maximum aperture. That usually means brighter viewfinder and better low
    light focusing. I know the eos 20D can make use of F2.8 or wider lenses.
    An extra set of autofocus sensors come into play. 2 vertical ones. My one
    gripe with canon equipment is the tendency of flash system to underexpose.
    Nikon is said to be superior in this regard.
     
    ian, Aug 15, 2006
    #7

  8. You've answered your own question then: Get the Canon.

    Cheers.
     
    achilleaslazarides, Aug 15, 2006
    #8
  9. The point about the exposure times and f/stop being different
    has nothing to do with noise, because the light levels between
    the two different scenes are probably different. It is all
    a matter of the total amount of photons received in each
    exposure. (A laboratory with fixed lighting would make
    for better tests.)

    Noise in DSLR camera images is greatly affected by raw converter
    software, so the only true way to understand the noise is a proper
    noise analysis on raw data that has not gone through a raw
    converter.

    Examples:
    Procedures for Evaluating Digital Camera
    Sensor Noise, Dynamic Range, and Full Well Capacities;
    Canon 1D Mark II Analysis
    http://www.clarkvision.com/imagedetail/evaluation-1d2

    The Nikon D50 Digital Camera:
    Sensor Noise, Dynamic Range, and Full Well Analysis
    http://www.clarkvision.com/imagedetail/evaluation-nikon-d50

    The factors that will influence the noise between different cameras
    is directly related to quantum efficiency, fill factor,
    pixel size, and at the low intensity end, read noise
    (and for long exposures, thermal noise).

    CCD quantum efficiencies tend to be slightly higher than
    CMOS sensors, so the advantage there is the Nikon (by perhaps
    10%).

    Fill factors are essentially 100% by the use of micro-lenses
    over the detectors (CCD or CMOS), so no advantage to either.

    Read noise: Canon's CMOS has <4 electrons on good cameras like the
    20D (and by extension 30D; same sensor). CCDs are typically
    7 to 15 electrons (the D50 above is ~ 7.5 electrons.
    So the CMOS has an advantage of ~2x at the very lowest signals,
    not the main things you see in the steves-digicams.com images
    which are much brighter.

    That leaves the major factor in noise: the pixel size.
    The important factor is delivering photons, and to do that,
    you need aperture. See:

    The f/ratio Myth and Digital Cameras
    http://www.clarkvision.com/photoinfo/f-ratio_myth

    The D200 has 6.1 micron pixels versus the 30D at 6.4 micron
    pixels, so a small difference (actually area is the important
    factor: 37.2 versus 40.1 square microns, again not much
    difference).

    There should be a slight advantage to the 30D but it should be
    small. (I would choose the camera based on other factors.)

    I will be evaluating a 200 in the next couple of months.

    Other sensor data are located in Tables 1-3 at:
    The Signal-to-Noise of Digital Camera images
    and Comparison to Film
    http://www.clarkvision.com/imagedetail/digital.signal.to.noise

    Roger

    Roger
     
    Roger N. Clark (change username to rnclark), Aug 15, 2006
    #9
  10. mswlogo

    C J Southern Guest

    Gosh - you sounded just like me when you said that.
     
    C J Southern, Aug 15, 2006
    #10
  11. Hello,
    I don't disagree with your points below (and indeed the photos were not
    even taken on the same day, so the light levels must have been
    different), however, the noise in the D200 image is much worse than I
    see in my photos, therefore I suspect it was underexposed. Hence my
    remark about the different exposures. The point here being that these
    samples aren't representative of the difference between the two
    cameras, at least in my experience with both of them (and various raw
    converters which I used to test them).

    My conclusion in the end was that the difference is in high ISOs and in
    shadows. Probably, from a brief look at your analysis of the D50 (lack
    of time, drowning in work), due to higher read noise. By the way, do
    you have any idea which factors affect the read noise (ie why would the
    D50 have higher read noise than the 1D)? The amplifiers (seems unlikely
    to me)? Presumably, this is random (eg thermal), otherwise they'd just
    model it and remove it.

    I've though of taking photographs of printed targets with various kinds
    of noise on them just to see what kinds of noise reduction are done on
    the raw data. [The Nikons certainly do this, see
    http://astrosurf.com/buil/d70v10d/eval.htm
    and I can't see how the Canons get their noise so low, but haven't
    confirmed anything.]

    Do you have any idea if anybody has done this? I could not find
    anything on the web.

    Cheers.
     
    achilleaslazarides, Aug 15, 2006
    #11
  12. mswlogo

    Mark² Guest

    True. Although in spite of my whining, I have yet to even experience my
    first sensor-dust speck on my 5D, much less any trouble with other seals.
    Would I prefer 1D type sealing? Absolutely. Is it life or death? No.
    Should Canon provide it anyway...on a $3K DSLR? No question. -Gaskets are
    super-cheap. Nikon proves that.
    I've had excellent flash performance with my 5D and the 580EX. Improvement
    over my 10D w/ 550EX is significant. Still not perfect, but it's an
    improvement. With full frame, I've had to get back in the habit of flipping
    down the built-in wide screen on the flash...but it works well.
     
    Mark², Aug 15, 2006
    #12
  13. Be aware that differences in in-camera sharpening, and in noise reduction
    algorithms may account for a large part of the difference you are seeing.

    By the way: I recently had the chance to compare my Nikon 8400 with a
    top-of-the-range Canon 5D. The viewfinder on the Canon was awfully dark,
    even though the image quality was (obviously) better. People tell me that
    was one of the better cameras, as well! It would seem to me that buying
    these cheap lenses with a maximum aperture of f/5.6 may be a mistake, and
    you should factor in the price of better (brighter) lenses,

    David
     
    David J Taylor, Aug 15, 2006
    #13
  14. mswlogo

    Mark² Guest

    Ah...but that is entirely dependant upon what lens you had attached to the
    5D. Your words below seem to indicate some awareness of this, but I wonder
    if you realize how dramatic a difference it is...

    A 2.8 lens lets in FOUR TIMES as much light at a 5.6 lens, so if you had an
    f5.6 lens attached (for example), it would show you a viewfinder image that
    exhibits only 1/4th the light of another lens. Heck...if you had the 50mm
    1.4 lens atached, that would be a whopping 16x brighter than a 5.6!
    Every stop doubles or halves the difference in light collected by lenses one
    stop different. A 2.8 is 2 stops faster than a 5.6, meaning it doubles the
    light (f4) and then doubles THAT at 2.8, for four times the light. A 1.4
    lens doubles at f4 (2x the light), doubles that at 2.8 (now 4x), doubles
    again at f2 (now 8x) and again at f1.4 (16x!!).

    All that to say... You can't judge any viewfinder on a DSLR until you know
    what lens is on it.

    -Mark

    Yes... So...which lens was on the 5D?
    :)
     
    Mark², Aug 15, 2006
    #14
  15. Mark² wrote:
    []
    Yes, I used to own a Nikon F3 with some f/1,4 and f/2.8 lenses.
    One of the f/5.6-ish zooms. The owner had both wide-angle and telephoto
    zooms. I only handled the camera for a new moments, but as I have been
    using my Nikon 8400 just before, the brightness difference was shattering.

    If the 5D is supposed to be a "bright" camera, I hate to think what the
    others are like. Certainly nothing as pleasant as my SLR experiences of
    10-15 years ago. And, of course, no split-image focussing, micro-prism
    etc.!

    Just goes to affirm what I always suggest - try the equipment for yourself
    before purchase.

    David
     
    David J Taylor, Aug 15, 2006
    #15
  16. mswlogo

    tomm42 Guest

    The Nikon 8400 has a video based view finder, so the brightness of the
    viewfinder is based on the video gain, not the lens. So it is apples
    and oranges. One of the problems with having auto focus and variable
    ISO in digital cameras, and that the higher ISO results are so good, is
    the slow lens (f3.5 - f5.6) is making its come back. With film SLRs we
    were sold 50 f2 as the entry level lens, now we have a f3.5- f5.6 zoom.
    With this we get a dim viewfinder. The less expensive Nikon and Canon
    cameras also have mirror based viefinders dimming the view even more.
    It is actually a good thing that the D70 and D50 Nikon can't take
    manual focus lenses as it would be difficult to focus with these lenses
    with the dim view finders.
    When I bought my D200, the first DSLR I owned, I bought a 24 f2 mf
    Nikon lens for it, mainly because I was used to 35 f2s on my film SLRs.
    Not a big brightness difference between my D200 and my Canon F1n with a
    35 f2 lens on it. There is a visible frame size difference, but the
    brightness is about equal. The D200 screen though makes it tough to
    accurately focus a wide angle, may get a Katz screen for it.
    Back to high ISO, the D200 is only about a stop different from a Canon
    D30 IF you nail the exposure, under exposure is a killer with that
    camera in high ISO. Unfortunatly working in high ISO necessary environs
    often is where you are shooting quickly rather than figuring out each
    shot, if you are shooting in that evironment alot get a Canon. Other
    wise which ever body is ergonomicly good for you.

    Tom
     
    tomm42, Aug 15, 2006
    #16
  17. mswlogo

    Jeremy Nixon Guest

    There isn't enough noise with any of these cameras that it should even
    be a factor in your decision. I could make a picture with the Canon that
    has far more noise than that Nikon "example"; does that mean it's a noisy
    camera? No, of course not.

    If you don't expose properly you get noise. If you do expose properly,
    it's just not a factor. I shoot with low light and high ISO with a Nikon
    D2x quite a lot, and noise just doesn't enter into it. I haven't seen
    enough noise to even tempt me into getting any noise reduction software.
    But, naturally, if you underexpose, then there's noise. So don't do that.

    Most "noise comparison" shots you see are total nonsense. Comparing noise
    from one camera to another is very nearly impossible to do in a way that
    means anything at all.
    Get past it. It's not worth worrying about.
     
    Jeremy Nixon, Aug 15, 2006
    #17
  18. The point, however, is that the D200 at high ISO reacts badly to
    underexposure (ie worse than the 20D). If it gets enough light, there
    is no visible noise. But of course a photograph with a wide dynamic
    range will include shadows. Still, with postprocessing, I found that
    the difference can be minimised, so it's a non-issue in reality (for
    me).

    Of course, someone else's definition of reality might be different from
    mine.
     
    achilleaslazarides, Aug 15, 2006
    #18
  19. mswlogo

    Robert Brace Guest

    Of course it's not worth worrying about, and any thinking photographer
    understands that. That is anyone who actually USES THE EQUIPMENT IN
    QUESTION!!!
    Think about it though -- if we didn't have these esoteric "noise" threads
    periodically, it would get dull around here and all the "pixel peepers"
    wouldn't have anything to discuss, let alone getting out to actually
    photograph anything.
    Bob
     
    Robert Brace, Aug 15, 2006
    #19
  20. mswlogo

    Ole Larsen Guest

    mswlogo skrev:
    This comparison is ridiculous - the second is underexposed at leat one
    step - more likely two steps. In Denmark we could call such a comparison
    fraud without any riscs.
     
    Ole Larsen, Aug 16, 2006
    #20
    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.