Shimmed Mounts and Extension Tubes

Discussion in 'Digital SLR' started by Wilba, Apr 12, 2009.

  1. Wilba

    Wilba Guest

    Wilba, Apr 12, 2009
    #1
    1. Advertisements

  2. Wilba

    Me Guest

    I doubt that shimming a mount can correct front or back focus. It just
    doesn't make sense.
     
    Me, Apr 12, 2009
    #2
    1. Advertisements

  3. Wilba

    Wilba Guest

    Wilba, Apr 12, 2009
    #3
  4. According to that manual shimming or the use of undercut mounts is
    used to adjust lenses so that the infinity stop has the lens focussed
    on infinity, not to adjust front or back focus.

    The idea of correcting front or back focus by shimming doesn't make
    sense to me either. Whether the lens focusses to infinity, or past it,
    is a different question. Indeed, if shimming did affect front or back
    focus, you'd have to do something else to correct the focus error
    introduced by shimming to adjust the infinity end stop.
     
    Chris Malcolm, Apr 12, 2009
    #4
  5. Wilba

    Wilba Guest

    Reading the Canon document with that idea in mind does make sense. Thanks.
     
    Wilba, Apr 12, 2009
    #5
  6. Wilba

    Paul Furman Guest

    I have had to adjust the mount distance on an old manual super tele lens
    with a universal mount in order to reach infinity... it went past
    infinity when I was done which is not a problem really.

    It never made sense to me how autofocus could be wrong. The method of
    achieving focus is trial & error as I understand... but obviously I
    don't understand well enough because it is something people do & some
    cameras can even adjust for.

    --
    Paul Furman
    www.edgehill.net
    www.baynatives.com

    all google groups messages filtered due to spam
     
    Paul Furman, Apr 12, 2009
    #6
  7. Wilba

    Bob Guest

    -:If we accept that the phase detect autofocus performance of a lens can be
    -:fine tuned by shimming the mount (e.g.
    -:http://forums.dpreview.com/forums/read.asp?forum=1029&message=30916099),
    -:then how does adding an extension tube not throw the focus WAY out?

    it does.
    it makes the lens focus much closer,
    with greater magnification.
    used for close up macrophotography.

    -:
    -:
    -:
     
    Bob, Apr 12, 2009
    #7
  8. Wilba

    Me Guest

    There's something to do with lens aberrations - that can
    throw out phase detect AF. I have a Nikkor 80-200 AF-d F2.8 that
    focuses perfectly at f2.8 < 135mm focal length and closest focus distance.
    But at close distance between 135 and 200mm, it back-focuses as much as
    a cm or two.
    It's not correctable (using AF fine tuning) as then focus would be off
    (front focus) at other settings. That lens does have some visible
    aberrations, some coma effect at 200mm f2.8, as well as some
    longitudinal CA. The behaviour of that (and a few other other) lens is
    noted in Nikon camera manuals - but they comment on inaccuracy of the
    focus confirmation light "digital rangefinder", (in)conveniently not
    mentioning that as the same AF sensors are used, then AF is also out.
    With that lens, for me it's not a significant issue, as combination of
    200mm and f2.8 and close range allows other options - "zoom with your
    feet", or use a macro lens optimised for close focus distance. Later
    versions of this type of lens (80-200 AFS and 70-200 VR) don't have the
    focus issue.
    I don't have any other lenses where back or front focus is an issue or
    where correction is needed amongst screw driven and AFS Nikkors and
    third party lenses, but others report a need to use AF fine tune on many
    and various lenses. I suspect that some people use inappropriate AF
    targets for measuring back/front focus, and try to fix problems that
    didn't really exist until they looked for them. Perhaps a new diversion
    for compulsive fiddlers has been created.
     
    Me, Apr 12, 2009
    #8
  9. Wilba

    Wilba Guest

    LOL. No, sorry, that's not what I'm talking about. :- )

    It doesn't matter now that I understand that the shimming is about infinity
    focus, not phase detect autofocus.
     
    Wilba, Apr 13, 2009
    #9
  10. Wilba

    Wilba Guest

    There has been a long-running bitter battle on the DPReview Canon Rebel
    forum about this. On one side, experienced technically knowledgeable people
    with evidence that their 450Ds have AF errors, and on the other, Canon
    fanboys who claim there is no problem except for stupid users bashing the
    camera. What I've learnt in this thread will help us to understand what's
    going on.
     
    Wilba, Apr 13, 2009
    #10
  11. Trial and error focussing is contrast based focussing, as done
    by practically any compact camera. DSLRs (outside life view
    focussing) use a different method, phase detection. Think of
    it as a computerized split screen focussing: a linear sensor on
    either side detects a pattern, by seeing where on each side said
    pattern is, the distance the AF motor has to travel is derived[1].
    Canon in one-shot mode does supervise that the motor doesn't over-
    or undershoot (and corrects it as needed), but does not re-check
    with the AF sensors.

    -Wolfgang

    [1] This includes offsets. That way, if the optical length
    isn't quite correct, for example, a corrective offset is
    applied instead of applying mechanical changes. Lenses might
    conceivably also transmit additional offsets for focus shifts
    due to aperture ...
     
    Wolfgang Weisselberg, Apr 14, 2009
    #11
  12. I've been trying to think of how to set up an experiment where in the
    brief interval between pressing the button and the lens stopping in
    focus the focussed object disappears. A squirrel on a tree branch
    kindly collaborated with me in this.

    My big heavy 500mm lens takes an appreciable fraction of second to
    whir from one end of its focus range to the other. In one-shot mode I
    got the squirrel's head smack in the centre and pressed the
    shutter. Before the lens stopped whirring along the squirrel
    disappeared, leaving blank sky in its place. The lens racked right
    out, hit the infinity end stop, turned round, racked in to the close
    focus end stop, and reported failure to focus.

    So this Sony DSLR clearly employs a closed loop AF focus.

    I know Canon in at least some lens models and bodies estimate the
    distance to focus, and if it's long they'll run the lens up to high
    focus travel speed, then slow it down for better accuracy as the
    estimated focus point approaches. I'm surprised that they don't run a
    closed loop approach which stops when good focus is detected,
    especially since some of their lenses report back to the camera what
    the bounds on good focus are, and this can be changed by altering
    switches or jumpers in the lens circuit board.

    From an engineering point of view a closed loop would seem to be
    superior and capable of better accuracy at less cost.
     
    Chris Malcolm, Apr 15, 2009
    #12
  13. Wilba

    Wilba Guest

    You're assuming that at the instant you pressed the shutter button, the AF
    sensor already had a pattern it could use to calculate the required focus?
    It's plausible to me that at the instant when the squirrel disappeared, the
    lens was racking to find a pattern that the AF sensors could use.

    I've seen something like this when using extension tubes. You can be looking
    at a well lit high-contrast image which is almost in focus, and the lens
    will go through its full range as you described, then fail to focus. I guess
    that it is not able to recognise a pattern that it can use.
    I don't see how it would cost less, but anyway, the answer is open loop is
    faster. And if the system is well calibrated (you get a good one out of the
    box or you send the system back to the manufacturer for calibration), the
    difference between a phase detect focus and a contrast detect focus can be
    similar to the difference between individual contrast detect shots. IOW,
    with a well calibrated system in good conditions, it's insignificantly less
    accurate.

    (All of the above relates to my experience with a Canon 450D.)
     
    Wilba, Apr 16, 2009
    #13
  14. No, as I mentioned above the lens was already moving when the squirrel
    disappeared. According to the description of the operation of Canon
    open loop AF given by another poster the measurement of the amount of
    focus travel required is made before the lens starts moving. In fact
    it would be very difficult to make such an accurate measurement if the
    lens was moving.
    This is the first time I've seen a suggestion of an open-loop AF
    mechanism which is based on measuring the amount to move the lens to
    focus from phase detection sensors which first requires initial
    focusing movements in order to acquire the measurement. Is there an
    on-line source which explains this?
    Either that, or what you're looking at is the operation of a closed
    loop system which was unable to find the accurate focus termination
    condition.
    The open-loop system requires very much more accurate sensing of the
    amount by which the lens is out of focus, and requires to be
    engineered in such a way that normal mechanical wear does not disturb
    the geometry on which the measurement is related to the necessary
    focus travel. Whereas the closed loop system automatically adapts to
    wear, and can use much cheaper less accurate sensors.
    True, but it's less accurate. I note you're arguing from experiences
    with a Canon. I note that Sony/Minolta is the only manufacturer to
    have an autofocus reflex lens. This is said to be because the
    Sony/Minolta AF system is the only one accurate enough to focus such a
    critical lens. Perhaps this greater accuracy comes from using a closed
    loop AF?
    You seem to be assuming that the phase detection AF of DSLRs is open
    loop, and the contrast detection of compacts closed loop. I'm arguing
    that at least Sony DSLR AF appears to be closed loop.
     
    Chris Malcolm, Apr 16, 2009
    #14
  15. Wilba

    Wilba Guest

    "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."
    http://www.zen20934.zen.co.uk/photography/Canon AF System.htm

    I can't vouch for the accuracy of Lester Wareham's statement, but it fits
    with my experience.
    I doubt it, since everyone who seems to know what they're on about seems to
    agree that Canon phase detect AF is open loop.
    Correct. I'm assuming that since lots of people say Canon's phase detect AF
    is open loop, and I don't recall anyone with any serious credibility saying
    it's closed loop, that that's probably true.
    I understand. I also see an alternative explanation (based on my Canon
    experience), for what you observed. I don't have any experience with Sony
    gear, so I'm just speculating from the possibility that it is also open
    loop. The explaination I get from that is as plausible as yours. :- )

    Can you try this? Put a pile of extension tubes on that lens, and see what
    it does when you try to focus on a well-lit high-contrast target. Does it
    rack in and out like it did in the absent squirrel shot?
     
    Wilba, Apr 16, 2009
    #15
  16. Wilba

    Bob Larter Guest

    That is, by definition, a closed-loop process.
    A fully closed-loop system wouldn't require much in the way of
    individual lens calibration, which is an expensive process. Canon lenses
    have a lookup table in ROM that's created when the lens is calibrated in
    the factory.
    That's very dependant on the electronics for each. You can't say, as an
    absolute, that either is faster than the other, as there are so many
    variables.
     
    Bob Larter, Apr 16, 2009
    #16
  17. Wilba

    Bob Larter Guest

    Again, this behaviour is closed-loop.

    Cf: <http://en.wikipedia.org/wiki/Control_theory>
    Also: <http://en.wikipedia.org/wiki/Servomechanism>
     
    Bob Larter, Apr 16, 2009
    #17
  18. I'd use a computer screen and a computer controlled remote
    switch ...
    Especially long tele lenses when defocussed a lot will prevent the
    linear sensors detecting a usable pattern ... and thus cause the
    fall back method of hunting once through the entire focus range:
    Not proven.
    Because of "focus and recompose", maybe?
    If you *have* to wait until focus has been archived, you spend a
    lot of time waiting while the subject is getting away ...
    even if you have decoupled AF from the half-pressed shutter.
    How so?
    Some lenses report a roughly stepped distance, which is
    mainly used for ETTL-2 "on camera hotshoe" flash photography.
    Is that what you mean?
    If you want a closed loop, don't use one shot! Try AI servo
    instead and live with it's limitations. You may want to decouple
    the AF activation from the shutter, since you cannot do any focus
    and recompose otherwise.

    -Wolfgang
     
    Wolfgang Weisselberg, Apr 16, 2009
    #18
  19. That could explain your experience. And if it was what happened in my
    squirrel experiment it would re-interpret that in a way which meant it
    couldn't be taken as evidence for closed loop autofocus. The reason I
    think that unlikely is that the squirrel's dark head against the light
    sky provided an excellent contrast edge for autofocus. I've just
    recalibrated the autofocus mechanism of my camera because it was
    backfocussing with that very critical 500mm f8 lens. I've been
    assessing the accuracy and speed of the adjusted autofocus with that
    lens by photographing the local squirrels a lot. When the squirrel has
    held its pose against the sky in dozens and possibly hundreds of shots
    I've never seen it fail to lock focus on a squirrel against the sky
    immediately.
    My problem with that idea is that the details of autofocus are so
    intricately technical that almost everyone I've read expounding on it
    has got at least bits of it wrong, and many otherwise very well
    technically educated photographers sometimes have quite serious
    misconceptions about it. In other words it's one of those rather
    confusing counter-intuitive topics that you really have to work to
    understand yourself rather than rely on the words of an authority.
    I don't have any extension tubes, and for autofocus to continue to
    work on my camera they'd have to carry the electrical contacts through
    and so wouldn't be cheap. What I can say is that with this camera I've
    never noticed that racking in and out on a high contrast subject with
    any lens. I'd have noticed that, because it does often happen if a
    very low contrast area happens to be placed on the AF focus area, in
    which case I make a point of shifting the camera to get something high
    contrast at a suitable distance to focus on, then locking focus and
    returning to the original composition.

    When doing close macro photography the only times I can recall the
    behaviour you describe is when I'm working right on the edge of
    closest focus distance, and happen inadvertently to have moved the
    camera too close.
     
    Chris Malcolm, Apr 18, 2009
    #19
  20. And at least some of their cameras also have a ROM table of lenses
    which they match with the information from the lens to determine the
    specific autofocus behaviour for that lens. As do at least some Sony
    cameras. Which doesn't argue against AF being closed loop, because the
    closed loop which runs (if it does) is based on taking readings from
    an AF sensor of a specific aperture. There are ways in which for some
    lenses the best critical focus of the image is sometimes slightly
    offset from the best position as measured by the AF sensor. For
    example some lenses have aperture related focus drift, and so need a
    little aperture-related focus adjustment for best focus when the
    aperture you're using for the photograph differs from the AF sensor's
    effective aperture.
     
    Chris Malcolm, Apr 18, 2009
    #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.