canon-nikon AF technology- the difference

Discussion in 'Nikon' started by Red_Baron, Oct 7, 2007.

  1. Red_Baron

    frederick Guest

    I've read this on forums on DPReview. Typically explained as
    inertia resulting in a lens "overshooting" the focus point.
    I tend not to believe it. There's myth and guesses about
    all sorts of things, ignoring the obvious in favour of the
    obscure can lead well up the garden path.
     
    frederick, Oct 17, 2007
    #81
    1. Advertisements

  2. Red_Baron

    acl Guest

    Well I don't know the specific case you are referring to (on dpreview;
    there's a lot of rubbish being stated there, though, and most in
    authoritative tone).

    But could you please clarify, for the present discussion, what is the
    "obvious" explanation that is being ignored, for which fact (is it an
    explanation), and in favour of which "obscure" explanation it is
    ignored? Thanks.
     
    acl, Oct 17, 2007
    #82
    1. Advertisements

  3. Red_Baron

    acl Guest


    Well I don't know the specific case you are referring to (on dpreview;
    there's a lot of rubbish being stated there, though, and most in
    authoritative tone).

    But could you please clarify, for the present discussion, what is the
    "obvious" explanation that is being ignored, for which fact (is it an
    explanation), and in favour of which "obscure" explanation it is
    ignored? Thanks.
     
    acl, Oct 17, 2007
    #83
  4. Red_Baron

    frederick Guest

    Clarify what?
    That when shooting moving subjects in AF continuous mode
    then you can't expect all photos to be perfectly focused?
    That predictive AF can not just use algorithms to extend a
    past trend to a future point, but uses a crystal ball?
     
    frederick, Oct 17, 2007
    #84
  5. Red_Baron

    acl Guest

    Let me try to express it differently, to avoid confusion: there was
    some discussion going on here. You made a remark that ignoring the
    obvious in favour of the obscure can lead well up the garden path. I
    assumed that this "obvious" explanation was one that was offered in
    this thread, as an explanation for some fact that was also mentioned
    in this thread. So I asked which explanation that was, and for which
    fact. Maybe I misunderstood?
     
    acl, Oct 17, 2007
    #85
  6. Red_Baron

    frederick Guest

    No - I was getting off topic in terms of the original posts.
    Motor in lens / motor in body IMO isn't an issue in terms of
    canon / nikon af technology difference. It might have been
    10 years ago. Af is great, but I can still unintentionally
    take out of focus photos with it. Blaming the technology,
    and getting specific about aspects of the technology that
    cause the error is stretching things a bit far.

    Some of the best examples of kidding yourself that it's all
    the fault of the technology are sample shots posted on DP
    review of "faulty" ultra wide lenses "soft in the corners".
    If I look at 90% of my real images (not brick wall photos)
    that I take with an ultra wide, the subject in each corner
    might be at say 6 inches, two feet, 10 feet, and infinity,
    the main subject somewhere in between some two of them.
    The $1800 Nikkor 14-24 is going to be just as faulty when
    "measured" that way as my cheap 10-20 Sigma (perhaps worse
    given maximum aperture and DOF with FX format at equivalant
    FOV), and given the hype, I guess that almost as many idiots
    will be buying the 14-24 as buy the 12-24. I'll wager that
    "sample variation" with the 14-24 becomes a much discussed
    issue on DPReview.

    See. I've diverted off-topic again.
     
    frederick, Oct 18, 2007
    #86
  7. Red_Baron

    acl Guest

    Ah cool, I misunderstood you, sorry.
    Sure, I agree on that.
    Yes, that's true. But dpreview isn't really a good place to judge this
    stuff; the forums are ruled by mob psychology, and there seem to be a
    small number of experts, some of whom know what they're talking about,
    while some don't. Good luck separating the good information from the
    bad.

    But faulty lenses really do happen. I just bought a Tamron 17-50mm f/
    2.8 last month (it's small and light, and one that I had borrowed
    earlier was simply excellent optically, aside from CA and vignetting).
    Well, it's blurry on the left side when focused at infinity (most
    visible at the wide end). Really blurry. Also, it doesn't autofocus
    accurately: wide open, it just doesn't focus reliably, most visibly at
    the longer lengths. If it starts focused closer than the subject, the
    focus is wrong. If it starts focused further away, it's ok. It makes
    it useless, as manual focus is impossible (it's far too highly
    geared). So it's broken. Tamron refuse to replace it, but say they'll
    fix it (I am mightily pissed: they sell me a broken lens and refuse to
    replace it?!). Let's see.

    Unfortunately for them, I'm perfectly willing to keep sending it back
    and forth until it's fixed. I'm betting it'll end up costing them more
    in labour than simply replacing it (well, unless they fix it the first
    time; I don't think they will, though).
    No problem, this is what makes this group fun!
     
    acl, Oct 18, 2007
    #87
  8. Red_Baron

    acl Guest


    Well I don't know the specific case you are referring to (on dpreview;
    there's a lot of rubbish being stated there, though, and most in
    authoritative tone).

    But could you please clarify, for the present discussion, what is the
    "obvious" explanation that is being ignored, for which fact (is it an
    explanation), and in favour of which "obscure" explanation it is
    ignored? Thanks.
     
    acl, Oct 18, 2007
    #88
  9. So why can I observe lenses overshooting during focussing and
    then correcting it immediately? Because it's open loop?
    Hello?

    -Wolfgang
     
    Wolfgang Weisselberg, Oct 18, 2007
    #89
  10. Red_Baron

    acl Guest

    Well I wasn't going to reply (as, from experience, thisis surely going
    to become long, tedious and stupid), but, conveniently enough,
    http://www.fredmiranda.com/forum/topic2/241524
    Unless you meant that the physical motion of the lens (as opposed to
    the whole AF process) is controlled by a closed loop , in which case I
    misunderstood.
     
    acl, Oct 18, 2007
    #90
  11. Wolfie is partially correct. The system stays in "open loop" for
    approximately 90-seconds or until it is up to normal operating temperature.
    If sensors don't sense a proper F/A ratio the system will stay in "open
    loop" and not run at peak efficiency.







    Rita
     
    Rita Ä Berkowitz, Oct 18, 2007
    #91
  12. That's exactly what I meant. One-Shot AF is open loop, but
    physically setting the lens to the calculated offset is
    closed loop.

    (Non-One-Shot AF settings can be closed loop.)

    -Wolfgang
     
    Wolfgang Weisselberg, Oct 19, 2007
    #92
  13. Wolfgang Weisselberg wrote:
    []
    Wolfgang,

    Let me see if I understand you correctly. In one-shot AF, the system is
    closed loop until the required accuracy is achieved. At that time, the
    focus is fixed, and as accurate as the designers set and the system
    allows. This is not "setting the lens to a calculated position", but
    measuring the focus error and correcting the lens position until the error
    is nominally "zero" (within tolerance).

    After the end of the focussing, the system is open loop in the sense that
    no further adjustment to the focus is made.

    Is that what you mean?

    Cheers,
    David
     
    David J Taylor, Oct 19, 2007
    #93
  14. Red_Baron

    acl Guest

    Aha, I misunderstood then.
     
    acl, Oct 19, 2007
    #94
  15. Nope.
    There are 2 subsystems to consider:
    a) phase detection
    b) setting the lens' focus

    a) is open loop. Once the phase has been detected (that may
    include forcing the lnes to traverse it's focus range),
    that's that. There is no check in one-shot if, after b)
    is done, a) is actually in focus.

    b) is closed loop. Over- or undershooting lenses are
    corrected, till the calculated offset from a) is reached.

    Since a) is open loop, the whole process of a) + b) is open loop,
    but setting the lens' focus (b)) is still closed loop.

    -Wolfgang
     
    Wolfgang Weisselberg, Oct 20, 2007
    #95
  16. Wolfgang,

    I'm not sure I either understand you or agree with you!

    Isn't the very act of checking "actually in focus" part of a closed loop
    system? If the lenas has to traverse its focus range, the phase detection
    must be repeated after the traverse has been made, mustn't it?

    Cheers,
    David
     
    David J Taylor, Oct 20, 2007
    #96
  17. That's OK!
    Phase detection is active WHILE traversing the focus range,
    but traversing the focus range is a fallback option and,
    basically, a red herring.

    What happens (in one-shot mode):
    phase detection successful?
    YES NO -> traverse focus range while detecting phase
    | Phase detection NOW successful?
    | YES NO
    | | |
    | ,----------------' V
    | / GIVE UP
    |/
    |
    V
    caculate lens steps to focus
    |
    V
    order lens: set focus so-and-so many steps closer/farther
    Did the lens over- or undershot?
    NO YES => order lens so-and-so many
    | ^ steps closer/farther to
    | | compensate for it
    | | Did the lens over- or undershot?
    | \ YES NO
    | \ | |
    | '-----' /
    | /
    | ,-----------------------------'
    | /
    |/
    |
    V
    Say "Focus successful"
    Exit focussing

    You see no arrow loop back to phase detection (open loop).
    You see an arrow loop back from lens over/undershoot (closed loop).


    In modes _other_ than one shot, there is no "Exit
    focussing", but control flow returns to the top. That's
    closed loop.

    -Wolfgang
     
    Wolfgang Weisselberg, Oct 20, 2007
    #97
  18. Red_Baron

    acl Guest

    But if we're within some distance from correct focus, the system can
    detect how far and in which direction it is out of focus, and simply
    send a signal to the lens "move to this distance" and stop there;
    according to WW (and the website cited earlier), this lens motion is
    closed loop, while the AF system doesn't check if it is in focus after
    it is done. This is supported by my experience with SLR AF systems
    (Nikon and Minolta, at least).

    The lens traverses the focus range only if the AF system can't detect
    in which direction it is out of focus. This happens if you're a long
    way from correct focus (or not so long if you're doing macro).
     
    acl, Oct 20, 2007
    #98
  19. Red_Baron

    me Guest

    Nope, just doesn't work that way. Even with a single sensor selected with a
    D200 on a stationary high contrast target, as I did this am.
     
    me, Oct 20, 2007
    #99
  20. Red_Baron

    acl Guest

    Well I'm not psychic, so you'll have to tell me what way it doesn't
    work...
     
    acl, Oct 20, 2007
    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.