High ISO noise CCD's vs CMOS

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

  1. I have no AF problems with the f/1.4, even in very low light.
    Good conditions != wide open?
    Maybe you are experiencing a shift in focus with closing the
    aperture (and that helps you) or a smaller aperture increases
    the depth of field, hiding inaccuracies.

    Or maybe yours is a lemom.
    Then you'd also find the problems with many of the other
    f/1.4 and f/1.2 lenses.
    Since the sensors work from the de-focussed lens image and even
    work with tele lenses (where they have much less 'sharp' images to
    work from if the focus is way off), I don't really buy that theory.

    Wolfgang Weisselberg, Aug 28, 2006
    1. Advertisements

  2. True enough, at least for moderate focal lengths - it's tough to come by
    anything below f/2.8 at 200mm. And I'd still observe that even with
    2.0, 1.7, or even 1.4, there are many shooting situations where there
    isn't enough light to get rid of subject movement, unless you start
    looking at ISO of 1600 and above.

    Marc Sabatella

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

  3. mswlogo

    AaronW Guest

    No. Sometimes the AF errors are huge that even stopped down shots are
    out of focus.

    Examples of bad condition: backlit, high flare, low contrast, etc. But
    I use the 50/1.4 and 50/1.8 under the same conditions. With 50/1.8, I
    have very few AF errors. While with 50/1.4, about half of the shots
    have AF errors. But the other half look good, so I think it is not a
    defective lens, but a design problem.
    I don't have any other f/1.4 or larger lens. I'd think that 85/1.2 has
    the same problem. I read that some other people have this problem with
    85/1.2, 50/1.4, and 50/1.0.
    Maybe the small lenses in front of the AF sensors focus it for them?
    Like the split images in manual focus are sharp themselves even though
    the lens is out of focus.

    If they really work from defocussed images, I can not believe that it
    can focus accurately from vastly out of focus images. They have to get
    more accurate reading when it gets close. So the lens sharpness when
    wide open will affect AF accuracy when it gets close.

    AaronW, Aug 29, 2006
  4. mswlogo

    Stacey Guest

    mswlogo wrote:

    Then buy the canon.

    For me things like color saturation and overall image quality is what
    matters but most gearheads seem OBCESSED with shooting at ISO 1600. Go join
    their club!
    Stacey, Aug 29, 2006
  5. You can always try for a 200mm f/1.8. :)

    Wolfgang Weisselberg, Aug 31, 2006
  6. Ok, that is hard for the AF.
    Ok, so you are saying, with the 50/1.4 you have 50% chance of AF
    error and <5% with the 1.8, right? And you say that it misfocusses
    even enough to see it well stopped down.

    Now, if that was a design problem, I'd see the very same with
    my 50/1.4, at least when shooting in the dark w/o focus assist,
    so we can probably rule out a design problem.

    It could be that on your camera the AF sensors are not exactly
    mounted where the focus points are. If that was the case, you'd
    probably focus on the background instead of the subject.
    With the minimal DOF of these lenses, it's very easy to focus on
    the center of the eye --- or an eyelash --- and have the iris be
    out of focus (at least on the computer screen). Tiny movements of
    the photographer or the object are getting very important there.
    Focus and recompose _will_ show up, as the DOF plane is slanted
    and might well move completely behind the focussed-on area.
    Nope. Unless the lens has another AF (with a lens that is focussed
    by yet another AF? And wouldn't that need yet another lens that
    focusses?), it cannot focus in the AF sense of the word.

    At best it could stop down a lot, but that would not help the AF
    (if it's sharp, it's sharp, what else can the AF say?).

    And would not ever confer better focus abilities for fast lenses,
    yet at least for it's top of the line models Canon claims
    exactly that.
    That's because your eye _can_ compensate with it's AF. The
    eye cannot compensate to a ground glass, as the glass itself
    is forming the image; what's OOF there cannot be corrected
    with any AF.
    You may observe focus hunting when you use a long fast lens
    and have the focus completely off.
    In digital P&S cameras, it's easy: move the lens and see if
    the AF areas get more or less hard borders. Continue in the
    right direction until you are better than some pre-set limit.
    The very short lenses (e.g. 6-18mm) necessitated by the small
    sensors help thanks to their deep DOF.

    In SLRs (and DSLRs), this is ... different. You cannot read
    the sensor. And you want to be _much_ faster.

    So what they use is conceptionally a split image, backed by a
    small, 1x200 or so sensor on each side. These can, even when the
    image is defocussed, usually still detect patterns (vertical or
    horizontal, or both, depending on the camera and which of it's
    sensors it is).

    Then they can see how much said pattern is shifted on the other
    sensor. That difference tells them exactly in which direction
    and how much to move the lens; the lens is ordered into that
    position and it is made sure that the position is, indeed, reached.
    However, there is no second read from the sensor. (At least that
    is how I undertand Canon works.)

    If no pattern can be detected (too dark, too low contrast, long
    tele very defocussed, ...), the lens may be moved through the whole
    focus range in the hope of picking up something in the sensors.

    Fast lenses obviously help the sensors (more light, for one).

    If the lens was 'too fast' for the sensor, the AF may be off a
    bit, but I still don't see how the lens would misfocus more than
    a slower lens (when stopped down to be identical, at least).

    A softer lens would produce more, well, softened patterns on
    the sensor, still, the sensor is _used_ to softened patterns
    (after all, we call that bukeh ... and usually train the AF on
    unfocussed parts). Worst case: it starts hunting.

    OK, the passive AF sensor is bad with repeating and weak

    Wolfgang Weisselberg, Aug 31, 2006
  7. mswlogo

    AaronW Guest

    But I get much better AF with 50/1.8 under the same situations.
    I get about 50% good AF with 50/1.4. I don't think I can get that many
    accidentally good AF with a defective lens. And I don't doubt you can
    get a lot of good AF shots with your lens. And maybe your camera is
    Many of the shots don't have anything in focus in the whole frame. It
    is possible that the AF points are not accurate, e.g., when I point AF
    to one eye it may actually measuring a spot next to the eye that has
    low contrast. But the AF confirmation light shows AF lock. And again,
    50/1.8 AF much better than 50/1.4 on the same camera, in the same
    The AF errors are much larger than DoF, even when stopped down.
    They have AF sensors spread further apart for f/2.8 lenses, in addition
    to the closer f/5.6 AF sensors. So if each side has its own lens, then
    the small AF sensor lenses won't affect the wider sensors being
    activated by f/2.8 lenses.

    AaronW, 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.