canon-nikon AF technology- the difference

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

  1.  
    Wolfgang Weisselberg, Oct 15, 2007
    #61
    1. Advertisements

  2. Your expectation is wrong then. What the heck do you think happens when
    a user sends a camera with several lenses into the service centre for
    calibration against EACH lens, at different focal lengths (in the case
    of a zoom lens) and at different apertures? The service centre can
    reprogram the LUT for each specific lens, taking account of the SA in
    that particular lens - this is not (currently) available in user
    accessible micro-adjustment. Getting specific lenses (or all lenses)
    calibrated on a particular camera happens all the time and the reports
    from those users who have gone through it is that it is very successful.
    I think you will find that most kit lenses aren't wide enough to suffer
    enough spherical aberration to make significant AF errors. The only
    people I have seen reporting problems needing calibration have been
    using much faster glass than a typical kit zoom. So I expect that the
    complete opposite of your suggestion is closer to reality.
    You are confused enough - don't add confusion of cynicism with
    stupidity!
     
    Kennedy McEwen, Oct 15, 2007
    #62
    1. Advertisements

  3. Red_Baron

    frederick Guest

    So with a LUT for each lens as described the for sure, I
    agree It can be done!
    Does every dslr come with such a feature? Canon or others?
    Sure, I was thinking more of users with entry dslrs who buy
    aftermarket fast lenses in particular. Say typically 16/18
    - 50/55 f 2.8 zooms, and primes such as the Sigma 30mm f1.4.
    This could explain (at least in part) the large number of
    complaints about user variability - so-called "sample
    variation".
    I don't think I am. If a LUT storing calibration data for
    lens (es) at different fl/focus distance in particular, or
    sometime even aperture isn't used, then I stand by my comments.
     
    frederick, Oct 15, 2007
    #63
  4. Red_Baron

    acl Guest

    But what does this LUT contain? Information on eg the amount of SA for
    each lens, allowing the AF system to focus and then compensate for the
    f/stop just before the shot is taken and so on? If that is the case,
    it should be possible to demonstrate simply by using manual focus with
    a lens with lots of CA and seeing if the AF indicator is less reliable
    than it is if using AF (by shooting at different f/stops etc).
    Unfortunately the only lens I have with any SA to speak of (judging by
    the shape of OOF specular highlights) is the Tamron 90mm, but it may
    not be enough; I'll try to make some time to check this in the near
    future. Meanwhile, has anybody tried/observed this?

    Although it seems to be the only way I can think of explaining the
    need for calibration on cameras with non-mechanical AF linkage (for
    screw-driven AF systems I can easily imagine that some information
    must be given to the camera about the lens' mechanics).
     
    acl, Oct 15, 2007
    #64
  5. Certainly Canon dSLRs have that facility, but only at service access
    level - and only for Canon lenses. The user micro-adjustment on the
    latest camera does not give the user access to that. I have no idea
    what Nikon (or other manufacturers) do, just as I don't know what
    Nikon's AF specification is, since I don't know anyone at a Nikon
    service centre, but I expect it is much the same.
    As I have said throughout this thread, your comments are wrong -
    certainly as far as Canon AF is concerned, which is the only
    specification that has been mentioned or disputed.
     
    Kennedy McEwen, Oct 15, 2007
    #65
  6. Except for one problem in your assumption - just like the AF, the
    viewfinder (especially bright viewfinders on AF cameras) aren't very
    good at manual focus and certainly don't show much light from the outer
    regions of the lens which is where the SA dominates.

    The reason for this is pretty obvious and explained early in Doug Kerr's
    document I cited at the start of the thread. The ideal focus screen
    would scatter light evenly in all directions so that an image could be
    viewed as it appears on the sensor or film. This is easy to focus on
    the screen and accurately judge the depth of field changes as the lens
    stops down. The downside of that is that the viewfinder image would be
    dark, because the f/# of the eyepiece in a typical slr is f/8 - f/11,
    so to get brighter images the focus screens doesn't scatter as much as
    that. This means that some of the light reaching the eyepiece is a
    direct aerial image, which can't be used to judge focus very well at
    all. More importantly, the light from the lens periphery to create that
    aerial image doesn't even get to the eyepiece.

    That is why, when you stick a fast lens on your modern dSLR and compare
    the viewfinder image in closing it down 1 or 2 stops from wide open the
    brightness change is much less than is visible going from, say f/8 to
    f/11 or f/16. Its the same change in light coming through the lens, but
    you don't see the same change in the viewfinder image - because the only
    light from the periphery of the lens that gets to the eyepiece is from
    the scattered component, not the aerial component.

    So, the flaw in your proposed "test" is that you are unlikely to see
    most of the SA in the viewfinder image that you are using to manually
    focus with.
    That would be irrelevant anyway, since 3rd party lenses can't be
    compensated for.
     
    Kennedy McEwen, Oct 15, 2007
    #66
  7. Red_Baron

    acl Guest

    Well, I did not suggest to see the change in the viewfinder. What I
    suggest is this: Assume there is a LUT with data for the focus shift
    as the lens is stopped down. Take a lens which autofocuses accurately
    and has significant focus shift (although you presumably wouldn't know
    if you used AF, since it is automatically corrected by assumption).
    Then focus using the focus indicator in the viewfinder and take a
    shot; since the LUT can't possibly be used, as the indication happens
    before the camera knows which aperture you plan to use (*), we should
    be able to observe less accuracy than if AF had been used.

    One problem is that it may be difficult to find a lens with enough
    focus shift to cause greater error than the AF tolerance. I don't
    know.


    (*) Alternatively, if the focus shift is taken into account, set the
    aperture to something, focus using the indication, change the
    aperture.
    I understand how focusing screens work reasonably well (I recently
    thought a bit about it). I meant to take a shot after focusing,
    although maybe I didn't mention it. Apologies for the confusion.
    OK, another question: how do you know there's such an LUT?
     
    acl, Oct 15, 2007
    #67
  8. Red_Baron

    Paul Furman Guest

    How does the viewfinder get a different aperture? Because it's enlarged
    somehow? I figure that's due to not being a 100% view but that's just a
    wild guess.
    What do you mean by 'aerial component'?
     
    Paul Furman, Oct 15, 2007
    #68
  9. I suspect that your alternative may be necessary to show the effect you
    are looking for.
    Most very fast lenses should show it since SA usually dominates the on
    axis performance when wide open. A fast lens without SA would be
    incredibly sharp.
    Indirectly by reading a service manual and trying to follow what was
    being programmed.
     
    Kennedy McEwen, Oct 16, 2007
    #69
  10. The viewfinder has a different f/# from the image lens because the
    viewfinder is a different optical system - a combination of your eye and
    the eyepiece. If that has a higher f/# than the image lens then only
    the light coming from the central part of the image lens can reach your
    retina - the amount of light is always limited by the highest f/#
    optical component in the system. Look at the size of your eye pupil and
    the focal length of the viewfinder eyepiece and you get a higher f/#
    than most lenses at full aperture.
    I mean the component of the image reaching the viewfinder eyepiece that
    comes from an image formed in free space "floating in air" at the focus
    screen, which is what you would see through the eyepiece if the focus
    screen wasn't there scattering the rays in all directions. A proportion
    of the image you see in the viewfinder is formed this way, whilst
    another proportion is formed by rays which are scattered into the
    eyepiece by the focus screen.
     
    Kennedy McEwen, Oct 16, 2007
    #70
  11. Red_Baron

    frederick Guest

    That's not very convincing.
     
    frederick, Oct 16, 2007
    #71
  12. Red_Baron

    Paul Furman Guest

    Interesting theory and it makes sense except that where I notice this
    effect most obviously is the size of large OOF circles is smaller on the
    focus screen than the captured image and what goes on the focus screen
    is separate from the optical system which refocuses that for your eyes
    through the viewfinder. The focus screen recieves a flat impression, not
    a clear optical path... the viewfinder works on that as if the ground
    glass were a print (I think).
    I don't follow this. Scattered light, like flare, reducing contrast?
    Maybe you are talking about what I tried describing above, where a
    portion of what you see is the flat 'print-like' image on the ground
    glass and a portion is direct light from the scene as if the glass was
    clear... that makes sense, but then it should be possible to see a faint
    shadow of the larger sized OOF circle.
     
    Paul Furman, Oct 16, 2007
    #72
  13. Red_Baron

    acl Guest

     
    acl, Oct 17, 2007
    #73
  14. Not necessary, at least for Canon, as the setting of the lens is
    closed loop and thus corrected against under- and overshooting.
    The lenses report the amount of angle change of the internal
    "focus ring" (note that the real focus ring on the outside of
    the lens does not necessarily turn)[1].

    -Wolfgang

    [1] while a great many lenses do report a (probably rough)
    distance, Canon chooses not record that distance in the
    meta data.
     
    Wolfgang Weisselberg, Oct 17, 2007
    #74
  15. Red_Baron

    frederick Guest

    That would seem the same for Nikon.
    People comment about supposed inertia effect with screw
    driven lenses, but usually refer to continuous mode AF,
    where they're probably comparing to faster USM/AFs lenses.

    The distance information in "D" lenses isn't accurate for
    for use in focusing. It can only report focus ring position
    in a number of zones.
     
    frederick, Oct 17, 2007
    #75
  16. That is exactly what you expect when the image you view is a combination
    of the image ON the focus screen and the aerial image passing through
    the focus screen and reaching your eye through a higher f/#. So your
    experience isn't at odds with the "theory" at all, it is precisely what
    it suggests. Now, if the focus screen was ground coarser and scattered
    more then it would pass less direct light, reducing the proportion of
    the aerial image that you see and hence your viewfinder image, although
    darker, would be more like the captured image in terms of DoF.
    Once again, can I suggest you read the first section of the document I
    referenced at the start of the thread.
    http://doug.kerr.home.att.net/pumpkin/Split_Prism.pdf

    Read up to page 4, ensuring that you understand the principle of the
    ground glass focus screen, which scatters the light so that some of it
    reaches the eyepiece - unfortunately drawn in Figures 3 & 4, although
    Figure 5 shows it on page 5.

    When you have understood that the screen scatters light in all
    directions and only some of that reaches the eyepiece, read on to the
    second paragraph on page 6 and notice what happens to light that only
    comes from the edge of the lens when no screen is present - as shown in
    Figure 5. This light fails to enter the eyepiece. Only light from the
    centre of the lens (blocked out on Doug's Figure 5 because he is
    starting to explain how a split image prism viewfinder works) would
    actually reach the eyepiece - and even then, it would be from a much
    more central area that actually reached your eye pupil. So the
    viewfinder is using a much higher f/# segment of the lens when a clear
    focus screen is used.

    The fine ground surface of an AF screen permits both types of image to
    be formed by the eyepiece, the dark scattered image from the ground
    glass and the clear aerial image. Because your eye is focussed on the
    ground glass surface you don't get a sharp view of the aerial, hence it
    isn't clear that there are two separate views, but that is why you see
    much more DOF in the viewfinder than appears in the captured image. It
    is also why a coarser glass screen is used for manual focus on those
    cameras which have interchangeable screens.
     
    Kennedy McEwen, Oct 17, 2007
    #76
  17. Red_Baron

    acl Guest


    If you think about it a bit, you'll probably conclude that it's very
    unlikely that it's a closed loop (at least, some things are very
    difficult to explain if it's a closed loop but completely natural if
    it's open).
     
    acl, Oct 17, 2007
    #77
  18. Red_Baron

    acl Guest

    Really? Where? I don't remember seeing any such complaints, and I am
    curious.
     
    acl, Oct 17, 2007
    #78
  19. Red_Baron

    Guest Guest

    canon's autofocus is summed up in the following article, with several
    references:

    <http://www.fredmiranda.com/forum/topic2/241524>

    In one "look," it calculates the distance and direction the lens must
    be moved to cancel the phase differences. It then commands the lens
    to move the appropriate distance and direction and stops. It does not
    "hunt" for a best focus, nor does it take a second look after the
    lens has moved (it is an "open loop" system).

    If the starting point is so far out of focus that the sensor can't
    identify a phase difference, the camera racks the lens once forward
    and once backward to find a detectable difference. If it can't find a
    detectable difference during that motion, it stops.

    Although the camera does not take a "second look" to see if the
    intended focus has been achieved, the lens does take a "second look"
    to ensure it has moved the direction and distance commanded by the
    camera (it is a "closed loop" system). This second look corrects for
    any slippage or backlash in the lens mechanism, and can often be
    detected as a small "correction" movement at the end of the longer
    initial movements.
     
    Guest, Oct 17, 2007
    #79
  20. Red_Baron

    Paul Furman Guest

    OK thanks, I didn't know those diagrams were relevant to this issue
    also. What finally made it click for me is the part about the size of
    your eye pupil: obviously much smaller than the film/sensor plane. The
    double circles I described (pasted again below) should theoretically be
    visible but in practice it all just looks pretty fuzzy.

    I'd say the majority of what I see is the aerial view, not the
    projection on the ground glass but maybe it's 50/50 so the aerial OOF
    circles would be much smaller & the circles I see are a hybrid.

    Now I understand why a more diffuse ground glass screen is better for
    focusing with fast lenses but too dark for slow lenses. It makes me
    wonder if there is another technology which could be applied to this
    issue though.
     
    Paul Furman, Oct 17, 2007
    #80
    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.