EF 50/1.8 AF Experiment?

Discussion in 'Digital SLR' started by Wilba, Dec 20, 2009.

  1. Wilba

    Wilba Guest

    The Canon EF 50mm f/1.8 II "Nifty Fifty" has a reputation for two
    shortcomings, 1) softness at wide apertures (OK from f/2.8), and 2) erratic
    focus under difficult conditions (low light, shallow DOF).

    Many people claim that 2) is a result of the crudeness of the cheap
    focussing motor and electronics in the lens, that those components are not
    able to provide the required accuracy and control of motion of the focus
    ring.

    But I wonder if 2) is actually a result of 1) - if the AF sensors have fuzzy
    images to work with, how /could/ the system nail the focus in difficult
    conditions?

    It would be interesting to see what happens when the AF sensors have sharper
    images to work with (e.g. at f/2.8 or f/4), but my 450D refuses to AF when
    the DOF preview button is pressed, so I can't test that. External aperture
    perhaps?

    Any ideas for how these competing hypotheses could be tested? Is there a
    consequence of either hypothesis that could be disproved empirically?
     
    Wilba, Dec 20, 2009
    #1
    1. Advertisements

  2. The AF sensors pay no attention to the aperture at which you're going
    to take the picture. They do their work before the lens is stopped
    down. Their construction gives them an effective aperture of their
    own. Often this is around f6. That means that when the largest
    aperture of a lens is smaller than that they can't get enough light to
    work properly. That's why generally speaking you can't make reflex
    lenses autofocus, because for technical reasons their best compromise
    aperture is often smaller than that, e.g. 500mm f8.

    More expensive DSLRs will also have larger aperture AF sensors at the
    central position, e.g. around f3, with which they'll be able to get
    focus in lower light with lenses which with max apertures which open
    that far. It also improves the focus on very fast lenses with
    spherical aberration and corresponding aperture related focus drift,
    such as the old spherical type of 50mm f1.4 lenses.

    Since the DOF gets very thin indeed at wide apertures and close
    portrait type distances, which is often what is going on in a dimly
    lit interior, the slightest error in AF will leave the image blurred
    at the point you wished to focus on, and sharp nearby. For example in
    a portrait you might have focused on the eyes, and find that the eyes
    aren't in focus, but the tip of the nose, or the ears, are. The reason
    for that is often that when DoF gets so sharp it becomes smaller than
    the small residual error in the AF of your camera, i.e. your camera
    has a slight front or back focus in the AF sensor plane calibration
    which is larger than the DoF at these wide apertures.

    If you find a systematic error of this type in your camera than you
    either must switch to manual focus, or compensate yourself, e.g. by
    holding down focus on the eyes and then simply moving your head back
    or forwards a few cm to take up the systematic error.

    Usually the more expensive DSLRs have better AF sensors so they can
    focus better in lower light. The wider aperture AF sensors are also
    able to get a tighter focus for wide aperture low light work because
    the AF sensor itself has effectively a shallower DoF. That will also
    rein in some of the aperture related focus drift of wide aperture
    spherical lenses.

    The more expensive DSLRs are also sometimes able to read lens-specific
    focus compensation factors from the lens, and use that to trim out
    systematic errors in autofocus for that specific lens.

    The most expensive DSLRs go one better than that. They have user
    trimmable tables of focus compensation for specific lenses in order to
    get better focus with the more awkward lenses in the more awkward
    situation, in which the AF will have slight lens-specific systematic
    focus errors.
     
    Chris Malcolm, Dec 20, 2009
    #2
    1. Advertisements

  3. Wilba

    Ofnuts Guest

    AF sensors work quite well with other lenses that are not that sharp, so
    I doubt that the "softness" of the lens at f/1.8 is really a culprit.
    Read this excellent explanation of the DSLR phase-detection AF system:

    <http://doug.kerr.home.att.net/pumpkin/Split_Prism.pdf>

    To make it short, for AF sensors:
    - they are designed to work with a minimum aperture (usually f/5.6 or
    better)(and don't benefit from a bigger one)
    - the more accurate you want the sensor, the wider the design aperture
    has to be.

    IIRC in the 450D most AF sensors require f/5.6 minimum, and the central
    one is doubled with a "bigger" one that requires f/2.8 and is put in
    action when the mounted lens reports that it has a maximum aperture of
    f/2.8 or better. This allows a more accurate focus with these lenses,
    which is required since the aperture of the lens can lead to very
    shallow DoF (some entry level DSLR haven't got that second AF sensor and
    cannot be efficiently used with lenses opening at f/2.8 or better).
    The 450D is an entry-level camera, so don't expect miracles. With the
    50/1.8 the accuracy of its AF system may be a bit pushed to its limits.
    And make sure that you are using the central sensor for the the AF.
     
    Ofnuts, Dec 20, 2009
    #3
  4. Wilba

    LOL! Guest


    <Munching popcorn, watching the usenet comedy show, while I enjoy using my
    slightly slower but highly accurate contrast-detection focusing cameras.
    Just what any intelligent person wants, phase-detection focusing that
    focuses slightly faster but never accurately. All those shots, missed
    forever.>

    too fuckin' funny

    This is such good free entertainment. Beats all the comedy routines on TV.

    /me wonders if they ever realize what gigantic fools they continually make
    of themselves daily ...

    LOL!
     
    LOL!, Dec 20, 2009
    #4
  5. Wilba

    Ofnuts Guest

    It just so happens that DSLR users are a bit more concerned about focus
    than P&S users because:

    1) they get to choose which part of the frame is in focus

    2) out-of-focus parts aren't obscured by sensor noise or
    diffraction-induced blur.
     
    Ofnuts, Dec 20, 2009
    #5
  6. Wilba

    Ray Fischer Guest

    I'd ask you to list all of those cameras that do "contrast-detection
    focusing" that can shoot 8 frames per second, but you're just a lying
    chickenshit little troll.
     
    Ray Fischer, Dec 20, 2009
    #6
  7. The simple answer, as the sentence above which you quote indicates, is
    no. But the question was raised in the context of AF which becomes
    unreliable in dim lighting at high apertures, and there are a number
    of technical problems and issues here, some due to the properties of
    spherical lenses (or incompletely aspherical ones :), some due to the
    way AF sensors work in different cameras, and some due to not uncommon
    small AF calibration errors only apparent with very shallow DoF.
     
    Chris Malcolm, Dec 21, 2009
    #7
  8. Wilba

    Wilba Guest

    Thanks for your effort. Unfortunately, there is no answer to my question
    within it. :- )
     
    Wilba, Dec 21, 2009
    #8
  9. Wilba

    Wilba Guest

    It's a good document which has influenced my thinking.
    Thanks for sharing your thoughts. Unfortunately, they don't answer my
    question.
     
    Wilba, Dec 21, 2009
    #9
  10. Wilba

    Wilba Guest

    Sure. These were shot a while back for comparison with PD AF shots to show
    the calibration error I had at the time -
    http://www.users.on.net/~alanw/Usenet/f1.8LiveViewFullsize.jpg (3.6MB)
    http://www.users.on.net/~alanw/Usenet/f2.8LiveViewFullsize.jpg (3.4MB).
     
    Wilba, Dec 21, 2009
    #10
  11. Wilba

    Wilba Guest

    Do I sound like a woman?! I'll try to butch it up a bit. <hwock ptooey>
    :- )
     
    Wilba, Dec 21, 2009
    #11
  12. Wilba

    Wilba Guest

    It's an alternative (Australian-sounding) spelling of Wilbur.
    Eh? What?
     
    Wilba, Dec 21, 2009
    #12
  13. Huh? I thought all Strines were called Curly (bald) or Blue (redhead).

    WTF???

    Good on ya, Wilba.. I think stateside we might spell it with an h at the
    end, assuming your spelling is more how it's pronounced- or does that
    spelling append a different meaning? Is your given name Wilbur?
     
    John McWilliams, Dec 21, 2009
    #13
  14. Wilba

    Wilba Guest

    Those shots were focussed using contrast detection. Any error you might
    perceive is of the order of the typical shot-to-shot variation.
    More information please.
    Yeah, hard to tell much with that much DOF.
     
    Wilba, Dec 21, 2009
    #14
  15. Wilba

    Wilba Guest

    Yeah, we're not all called Bruce.
    Right, just like it looks - alba, amoeba, caramba, samba, tuba, etc.
    It's a nickname from 1969, after the cartoon character Wilbur the Worm. It
    stuck so I made it my own by changing the spelling. It matches my character
    better than my given name. :- )
     
    Wilba, Dec 21, 2009
    #15
  16. Your qestion doesn't make sense because you don't understand enough
    about what might be going on with your specific camera and your
    specific focus difficulties. Your questions need to be revised in the
    light of an improved understanding, part of which must come from doing
    some experiments of your own to discover exactly what with your
    specific camera and focus problems the most important problems
    are. For example, your problems could be simply not enough light for
    your camera's AF sensors to work properly. If so there's nothing you
    can do with the way you use your lens to improve that. But there are
    several ways of helping your AF to work better in poor light. On the
    other hand your difficulties could be due to a small AF calibration
    error which starts to matter under those circumstances with your
    lens. If so there are a number of things you can do about that.

    But it's not worth going into all the details of all the possibilities
    until you have found out more about which particular limitation of the
    several possibilities your camera and lens are coming up against.
     
    Chris Malcolm, Dec 21, 2009
    #16
  17. Wilba

    Wilba Guest

    I'm not having focus difficulties. The question is stated in the 5th
    paragraph. All the rest is preamble.
     
    Wilba, Dec 21, 2009
    #17
  18. Wilba

    Wilba Guest

    Thanks, no worries with all that, I just hadn't heard the term before.

    What you say fits with my thinking, the problem is finding a way to test the
    theories.
    It sure feels and sounds like a crude sloppy mechanism with the 50/1.8, and
    I'm sure that has a lot to do with people inferring that that causes the
    focus performance they observe.
     
    Wilba, Dec 21, 2009
    #18
  19. Wilba

    Paul Furman Guest

    If you are using the center AF point, sharpness shouldn't be a problem.
    In fact the shallower DOF should improve accuracy.

    One factor could be the phenomenon where focus shifts when stopping
    down, but I suspect they've included corrections for that in the
    firmware. There could also be issues with curvature of field, if you are
    concerned about something off to the side, etc.

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

    all google groups messages filtered due to spam
     
    Paul Furman, Dec 25, 2009
    #19
  20. Wilba

    Wilba Guest

    I can interpret that to mean several different things. What do you mean
    exactly?
    I think that's part of why the "crude mechanism" theory is so popular, but I
    can't think of any way to test it. I'm not interested in theory or
    speculation except that which leads to an experiment which proves something.
    If that were a factor I would expect to see a consistent mis-focus, but
    that's not what I get.
    I'm not. :- )
     
    Wilba, Dec 25, 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.