Focusing issues 50/1.4 AF G

Discussion in 'Nikon' started by albin''s images, Nov 23, 2013.

  1. Dear photo.net'ers, a quick search didn't bring up exact hits, so here is my question.

    Do any of you notice weak AF with the 50/1.4 AF G?

    I bought the lens to accompany my new D800 a year ago. As a no-compromises modern basic.
    But. . Sharpness and contrast beyond wide-open are excellent. But autofocus is not.
    It seems always unsure, searching for exact focus, and misses in low contrast situations.

    It is much worse than the 24-70, which 'only' has 2.8 but does have a shorter focus 'throw' (you have to turn less).
    Like this it is much less dependable than I hoped for.
    Any similar experiences?
     
  2. F/1.4 lenses of any focal length ain't perfect. There's always going to be some residual spherical aberration to interfere with the AF process. Spherical aberration spreads (what should be) a single plane of focus into a volume of focus, and neither the human eye nor an automated focus detection system can accurately determine exactly where the focus should be.
    If you look at the "Bokeh fringing" (sic) section of this review, you'll see how the focus shifts quite markedly on stopping down. This means that the 50mm f/1.4 AF-G Nikkor will progressively back focus as the working aperture is reduced from wide-open. In other words, it's not the camera's fault; it's the lens. And same goes for a great number of other lenses wider than f/1.8.
    Edit: BTW Sigma's f/1.4 50mm appears to have almost exactly the same focus shift characteristic according to the Photozone.de review.
     
  3. If focus shift is a known quantity, then the camera can correct for it? The camera body knows everything it needs to know to offset the focused point to where it should be when the aperture closes to the desired position and the exposure actually happens.
    When focusing in LV, the aperture isn't wide open, unlike in Phase Detect..... I'd guess this helps limit focus shift effects on stopping down?
    It's still odd that if I ask what physical aperture is set for LV focusing at a body-pre-set wish of, say, f8, no-one knows....it might be f8 or f2.8 or f22....it depends on all kinds of stuff! It might be different on the newer FX bodies, but the aperture on my D5100 is all over the place when focusing.
     
  4. My experience is that you have to gather some experience with a fast prime and a particular camera and then determine how the focus fine tune should be set for a given situation. It is not necessarily the same at all distance ranges or lighting conditions, which is why it's not something that can be set at the factory and the user basically is responsible for tuning the lens and camera to the situation. The D800 due to its high resolution brings out these issues into daylight, so to speak.
    When the focus fine tune has been set to optimal value for a situation I think 80% of shots in acceptable focus with the D800 can be obtained. Really if the aim is high resolution work, a tripod and live view focusing will yield the best quality results. It's the current state of the technology together with the ultra narrow depth of field and desire for instantaneous autofocus that limits these fast lens + high resolution camera combinations today. I would recommend persisting with it and finding what settings work best for you. I've gotten excellent results with the D800 and 50/1.4G though it may not be quite the best of the f/1.4 AF-S lenses (I think the 85mm is).
    In future generation bodies no doubt the autofocus will evolve and lenses too become better over time. However the progress is gradual and takes many years. The D800 is really a photojournalist's camera with a high resolution sensor plugged in! :) It does give phenomenal quality when things fall into place, but they do not always do that.
     
  5. If focus shift is a known quantity, then the camera can correct for it? The camera body knows everything it needs to know to offset the focused point to where it should be when the aperture closes to the desired position and the exposure actually happens.​
    The camera "knows" nothing that hasn't been programmed into it, and I seriously doubt that focus shift has been acknowledged or even considered in the programming of the camera. Whereas a lens's two dimensional aberrations and distortions seem well-documented and corrrected for, the use of distance information seems very rudimentary in Nikon cameras. In any case the issue is more one of the AF system not being able to find accurate focus at wide-open apertures in the first place.
    Look again at those smudgy f/1.4 pictures of ruler lines in the Photozone reviews. Can you tell which exact line the lenses were focused on? At f/2 I can narrow it down to 2 or 3 lines; but not so at f/1.4.
     
  6. The camera "knows" nothing that hasn't been programmed into it

    the use of distance information seems very rudimentary in Nikon cameras​
    I think it was Dieter in another post that said D information was available in EXIF if you have the right EXIF reader. I'd be interested to see how accurate it is.
    It would seem that if you showed the camera a target at known distances, calibrating D would be pretty easy for a data file for each individual lens. Maybe that's what the Sigma USB data port is for? Afterall, the cameras still have the mark that accurately (?) shows the sensor plane.
     
  7. Gentlemen,
    Thank you for your very informed posts. Much appreciated!
    Focus fine tuning.. Mmm.. I was seriously hoping to avoid that.
    On the other hand, I suppose it will not make a difference in low contrast situations. The remarks about spherical aberrations were enlightening.
     
  8. Mike, Exiftool shows the focused distance data in metres, correct to 2 decimal places - sorry, make that just shown to 2 decimal places. That's the focused distance that the AF system decides is the plane of best focus. The degree of apparent accuracy won't help if the contrast and point-spread-function of the image are too diffuse to generate a positive focus lock. It's a catch 22 situation.
    In short we come back to Ilkka's observation that the only way to get accurate focus is to use Live View, which gives the actual working aperture sensor image to focus on.
    Edit: Another complication is that the focused distance is measured from subject to focal plane, and takes no account of lens internodal distance. The maths gets quite messy and requires the solution of a quadratic to sift out F, v and u from focal length and distance info.
     
  9. I used the 50 f1.4G lens for two basketball seasons and never had an issue with autofocus speed using a D700 with grip. I also used the 1.4D and after a little testing and looking at the results I liked the G lens best. Maybe you have a bad lens, my G lens is fast, maybe a tiny bit slower than the D but it's hard to tell.
     
  10. I have the lens and it's pretty slow to AF compared to a f/2.8 lens at similar focal length. Then again at f/2.8 you don't need to focus at anywhere near that kind of precision as you need at f/1.4. So from an engineering point of view, the AF have to be slower on a f/1.4 lens than it has on a f/2.8 lens - if you want more precision.
    Also AF speed depends a lot on the distance you are shooting from. If you are working close say up to 10ft it takes longer for the focus to change from near to far than if you're shooting at 10ft to infinity. DoF is also smaller at shorter distances. So what works well for some, might not work for others.
    What I've found on the D600 I have at work is that even in single focus point, the actual area used for focus is much, much bigger than what it appears to be in the viewfinder. About three times the size of the focus point square in the viewfinder, both horizontally and vertically. I've come to this conclusion having done some test in low light using a subject that the camera is unable to focus on (zero contrast), close to a subject it can focus on.
    That means that even in single focus point the AF might lock onto something close and not exactly where you want. When the DoF is razor thin it means out of focus. Using dynamic 9 or 21 or 39/51 would be an even bigger problem if we want to have a precise focus where we need it.
    So I think how you have the AF setup and how you shoot, at what distance and at what kind of subjects, makes a big difference. The way you shoot at f/2.8 might be OK but when you put on a f/1.4 lens and shoot wide open every mistake the camera (and you) makes is now visible.
     
  11. the only way to get accurate focus is to use Live View, which gives the actual working aperture sensor image to focus on.​
    RJ, without meaning to sound doubtful :)-)) but is the bit I've made bold true? Does the aperture stop down to the taking position when LV is activated?

    No-one seemed to shoot me down when I typed.....
    It's still odd that if I ask what physical aperture is set for LV focusing at a body-pre-set wish of, say, f8, no-one knows....it might be f8 or f2.8 or f22....it depends on all kinds of stuff! It might be different on the newer FX bodies, but the aperture on my D5100 is all over the place when focusing.
    ...or maybe you were all being nice?.....:)
     
  12. The IQ is not bad on that lens, but I've never used a worse lens in terms of AF performance. In bright daylight, it usually worked, but in light that was even slightly dim, it hunted and hunted. I missed so many photos that I got rid of it and went back to the AF-D, which, although it doesn't have quite the same IQ, is fairly quick and accurate to focus.
     
  13. Mike, my D800 appears to stop down consistently to the set aperture in Live View; my D700 doesn't do that if there's insufficient light. I guess there are discrepencies in the way that other Nikons work in Live View too.
    Assuming that there's ample light, I believe any modern Nikon offering Live View will attempt to stop down to the requested working aperture during viewing. I've only noticed the D700 "disobeying orders" when the light level has been quite low. In any case, once you get down to f/4 or thereabouts the image contrast and resolution should be close to optimum, and any focus shift will have levelled out. So as long as Live View stops an f/1.4 (or faster) lens down to at least f/4 when smaller stops are requested, there shouldn't be an issue with finding accurate focus. It would only be a major problem if the lens stayed wide open or was only stopped down to, say, f/2.
     
  14. RJ, yup that's what mine do in low light. I guess the video amp. can only do so much to make the screen an use at all to focus, so it opens up the aperture enough to 'see'.
    It would only be a major problem if the lens stayed wide open or was only stopped down to, say, f/2.​
    I just did an experiment with my 105mm AFS VR macro on my D5100 set to f8 in LV. The aperture changes size, ie shuts down, to f8 I guess, when i point it at my brightly light subject, but opens up to ?? when I point it into a dark corner.
    From what you say about yours, the D700 does do the above and your D800 does not?
     
  15. Jeremy John, I suppose your post directly supports my experiences.
    It doesn't exactly reassure me though..
    And I am most grateful for the others providing background information.
     

Share This Page