21
Dec
2012

unreliable autofocus with the XF-18mm

In our “Fuji future improvements wish list poll” the option “faster autofocus” reached the second place. Although firmware updates improved the autofocus, and the 18-55mm lens with the new linear motor focuses definitely faster than previous lenses, many users are still not satisfied with the autofocus performance.

This video concerns the XF-18mm lens (click here): “The X-E1 has problems locking focus in low light. But when it does, it is simply unreliable. Look at the distance scale as the camera locks and confirms the focus to different distances while the actual distance hasn’t changed.

Our reader Sagi sent me this video. In his mail he wrote: “At first the X-E1 can’t lock the focus at all. At the same point, the Oly OMD (click here) locks times after time and fast… then I moved the focus point to a more contrasty area where the XE1 did managed to lock focus but not all the time… Notice that when it does lock, it locks each time to a different distance. This makes the auto-focus very unreliable… It doesn’t happen all the time, it seems to happen more often when using a wide open aperture and in high ISO… In most cases these little misses do not affect the outcome or at least it’s not that noticeable as the DOF covers it. But I do get inaccurate focus in real life situation.

So, problems to lock the focus, troubles in low light and backlit situations, speed… These are issues that emerge in some reviews I’ve read and are frustrating for X-photographers.

Thanks Sagi (website)… now the debate in the comments is open.

  • john burtoft

    Hi, I have an x-pro1 and the 18mm and 35mm lenses. I find the AF to be fine, never had any of the problems the video shows regards being unreliable, for instance i would have focussed on the sharks mouth which is the area of most contrast, or on the edge of the picture frame, its all about knowing how to use the equipment. In my time having the camera i have learnt to work with it, to get to know it and to develop techniques which allow me to take photos in all but the most fast action situations. Cheers Jon

    • MJr

      That’s very much true, but it does hurt knowing it *could* be better, or i guess i should say ‘foolproof’. :) But sometimes there’s just no sharp edge to focus on, it’d be nice not to need to worry about that.

      • Sagi

        It’s true, you need to learn your equipment in order to make the best of it.
        But autofocus shouldn’t be so complicated… you point, press the exposure button half way and finish to shoot. You do not need to be a rocket scientist for that :-)

    • http://www.flickr.com/nicolasraddatz Nico Foto

      What John says makes a lot of sense. If you are looking for situations to fool the camera, you will most likely find one, be it either a Canon DSLR, a FUJI mirrorless or whatever camera you choose. There is an open question with regards to the usefulness of these kind of tests in the real world.

      BUT! No matter what, AF has been from day one the Achilles heel of FUJI cameras, in fact I have an x100, and even though I learnt its quirks and find myself comfortable working with it, there’s no contest in terms of speed compared to, say an E-M5. They need to up their game with regards to that. The rest, is truly a fantastic photographic tool.

  • milan_a

    and not only try to take pictures with a flash car from the front or from behind. Apparently the reflection of light from the registration table causes the flash setting at minimum and the picture is dark

    • Sagi

      This was not a search to fool the camera…
      I am aware that the focus is a bit slow and that it sometimes fail in low light. This was reported by other before I got the camera.

      In real life situations I’ve experienced a focus misses even though the camera gave confirmation for locking the focus on the object.
      This led me to run this little experiment that may explain the focus misses I’ve encountered.

  • jr

    I agree. It’s very frustrating in low light and backlit situations! Sometimes it confirms AF (green) but it’s actually wrong lol.

    I had the EM5 for 7 months and Fuji X being CDAF is not an excuse because EM5 is also CDAF.

  • Milan Angelovic

    I have x-e1 with 18-55 lens. Distance measurement shows different values ​​for focus 18mm and 55mm. It’s annoying when I want to focus in low light, when autofocus fails.

    • Antonio G

      I only have the experience of X100, but the distance’s bar is just a guess to use for manual zone focus and at the end it doesn’t impact focusing – autofocus locks based on contrast and manual focus on visual perception on the screen or viewfinder, no matter what the bar shows.
      It is not precise? Ok, we shall ask Fuji to turn it into a reliable tool by firmware update or to improve autofocus and manual focus easy of use (focus peaking or similar) but not to make a drama out of the viewfinder’s graphic bar.

      • http://www.flickr.com/nicolasraddatz Nico Foto

        I’m not sure where you got that thing of the distance bar being a *guess*. As far as I know, the distance bar should be a reliable indicator of the camera focus plane. Can you show us some documentation or official information that says that the distance indicator is a *guess*?

        • Antonio G

          When I said “a guess” it was for the user not for the camera, obviously.
          I got it just by using X100 manual focus fine adjustments and realizing that the “answer” is not linear, no matter how slowly you rotate the lens ring the indication on the distances bar doesn’t show a continuous and stable movement.
          As all focusing by the wire systems, when you move the ring it sends a signal to the camera to process and to compute the graphical indication in the viewfinder information (distances bar, focus point and DoF range indications).
          At the end the camera can focus precisely besides the viewfinder indication is not as exact as it should and I would like it to be.

    • Sagi

      EXACTLY!!!!

      I’ve noticed those misses as well.

  • Antonio G

    Strange “headline” as it says that “autofocus locks at different distances”. Is this opinion based on a graphic in the viewfinder or on the results?
    On thing is the graphic indication and the other is the real distance and it seems that the user is focusing the same point several times and the autofocus locks on it, how can he talk about “different distances”?
    Is this a real problem or just “internet issue’ finding” willingness?

    • Milan Angelovic

      I’m in the dark, I’m going to take pictures with flash object for example distant 3m, I do not see anything in the viewfinder, so I want to set focus distance in meters, but when indication in the viewfinder is not accurate, so the focus is inaccurate

      • Sagi

        See my other comments to answer your question:
        Is this a real problem or just “internet issue’ finding” willingness?

  • Oded S.

    While this video is indeed some cause for concern, note the lighting conditions it was taken under: 1/15 sec. at f/2 on ISO 800. This is about three stops dimmer than average indoor artificial lighting (without daylight streaming in through the windows). How likely are you to take photos under such dim conditions, and what would be realistic expectations of the autofocus system? If these occassions are relatively rare, maybe you could just focus manually.

    Note also that the entire range of AF detected distances in this video is about 80 to 100 cm. Not a huge difference, although at f/2 you’ll probably see it.

    What’s more interesting to me is the low-light AF performance with the kit zoom. I’m thinking about getting an X-E1 or an X-Pro1 with this lens. Does anybody have some comparable experience with it?

    Oded

    • http://www.flickr.com/nicolasraddatz Nico Foto

      On the contrary, ISO 800, 1/15, f/2 doesn’t seem that bad lighning imho. I would say it is a rather average low light situation – certainly not a *worst case* scenario.

      • Sagi

        Obviously, you don’t really know what you are talking about when it comes to low light…

        The room was lit with a spiral bulb with the equivalent power of 100W with daylight white balance. this is hardly considered low light…. You should check the conditions in a club while trying to shoot a live band performance… (this is something I will be testing as well)

        For the purpose of the test, the shutter speed is irrelevant.

  • GreyOwl

    I’ve just tested my X-E1 with 18 mm lens at ISO 1600. The focus locks every time, but the indicated distance varies. I repeated the test with my 35 mm lens (same ISO 1600); focus locks every time, but with less variation on the distance scale. The ultimate proof of a problem, real or imagined, lays of course, in the final photograph.

    • Antonio G

      You’re in the dark, so let’s see:
      (1) you use manual focus, adjust your focus in advance using something at 3 m and shoot your subject
      (2) even if the bar is not as precise as it should be, as your subject is at 3 m you can use F:8 and adjust the focus indication at the bar to 3 m.
      As you’re using a 18mm lens, your hyperfocal distance is 2,04 m and you’ll have a DOF from 1,21 m to ∞.
      If the indication error is 1m it will mean that effective focusing is being locked at 2m and your DOF is from 1,01 m to 94,2 m or, from 1,35 m to ∞ if focus is locked at 4 m.
      As flash exposure calculated on the sensor I guess the camera will adjust it with no big trouble.
      This implies you have neither light for manual focus using the viewfinder and subject magnification for precise focus adjustment nor conditions for the camera to lock autofocus using the assist lamp, so you’ll be in the same situation as someone using a rangefinder and forced to use zone focus just looking at the marks on the lens barrel.

      • Antonio G

        Sorry, my last post is a reply to Milan Angelovic and not to GreyOwl (as it shows under his post).

        • Milan Angelovic

          Thanks Antonio, ok I understand, but it is not entirely clear solution

  • Gary

    john burtoft writes:
    ” its all about knowing how to use the equipment.”

    Yes. It appears, after my short experience with the X-E1, that low light focusing is difficult when the focusing frame is too small.
    It works as well as I could hope with really extremely low light (1/2 s f/2.8 ISO6400) and the largest size of the focusing frame.

    To adjust the focusing frame size, refer to page 44 of the manual (same page number in English and French versions).

    For me, the focusing is correct and reliable, whatever the indication at the bottom of the viewfinder. It is no real problem for people who want to take photos, just another “internet issue”.

  • http://Flickr GLYA

    I just get my XE1 and its true that thre is AF problems in low light condition. I’m using it with 18-55, and also with legacy lens (canon FD, Nikon old ones, and LTM 39, Minolta MD, Contax G…) manual focus of course and it works perfectly. There only a problem with the FOTODIOX adapter LTM 39 to Fuji X: the adapter is mouted upside down, so the aperture ring is facing the ground….I’ve ask FOTODIOX to change this and I wait for answers…..

  • http://www.lets-ride.de Rico Pfirstinger

    Not the AF but only the AF distance scale is unreliable with the 18mm. It has always been this way and I have been puzzled since February 2012 why nobody seemed to complain about it.

    I documented this issue as fas back as in early summer and sent it to Fujifilm. Then again, it only affects those who do rely on the red marker and the white DOF bars. We all know that the DOF bars are based on a CoC of 0.005mm, which is fine with me but too conservative for many in the “zone focussing” crowd. The latter seems to prefer something like 0.02mm.

    In any case, the AF is very accurate no matter what the distance scale is showing. I guess the problem may have its roots in the high DOF of wide angle lenses. Basically, everything between 4 and 6 meters will be in perfect focus, anyway, so the camera simply can’t measure any contrast difference between those distances. Of course, that’s just an educated guess.

  • JC

    I only have experience with the X100. I bought it when it first came out and to focus in low light was a nightmare, so I got rid of it. However, it was still the camera I wanted (small, optical VF, 35mm equivalent, great IQ…) so I decided to try the improvements of the new firmwares and finally decided to buy it again a couple of months ago. I couldn’t be happier, the focus is very snappy and locks focus even at low light, especially in AF-C, 99% of the times. I guess with a firmware update Fuji will be able to improve the focus of the XE-1 as they did with the X100.

  • Sagi

    OK, some updates.
    Let’s start by explaining why I did this test.
    I was shooting in real life situations with f/2 and high ISO (around 1600 to 3200), it was in a covered flea market so those settings were needed for good exposure at decent shutter speeds to freeze movements as I was photographing people.
    When I checked my results I saw that in some cases the focus was inaccurate, there were misses even though the camera gave focus confirmation. (As mentioned by jr in his comment, this is not just me…).

    So the test that led to this video was to try and understand if this is a camera/lens issue or is it a user fault.
    It is only then when I started to some tests that I’ve noticed that distance scale shows different focus distances for the same focus distance.

    What I did next is I switched to manual focus and placed an object in different distances from me which I’ve measured and marked with a tape measure. I then manual focused on the object from the known distances using the enlarged viewing capability of the camera to make sure the focus is accurate. the camera distance scale showed the correct distances at all time.
    This proved to me that it is not a graphic issue but rather real.
    Although it could be working differently in AF and in MF.
    I’ve seen those distance scale variations on the X10 as well but never gave it much thought as it was always spot on!

    It is true that in some conditions and distances, the variations will not affect the outcome as it might be still within the DOF of the given focal length and the aperture. But I did have miss-focusing issues.

    I’ve tested this again in real life situation and this time I actually did took pictures and not just did focus. The issue of the variance distances on the scale accrued again but when looking at the pictures, focus was spot on.

    SO, the intermediate conclusions are:
    1. The distance scale might work differently for AF or MF. While it is accurate for MF, it should be ignored for AF. It is still an issue because I find it useful sometimes. I think that Fuji should check it out and fix it.

    2. While in manual focus mode, it pressing the exposure button half way, it’ll sometimes move the distance scale a bit. This should not happen!
    It seems that the movement of the aperture moves some elements in the lens… the lens element mechanism for focus should not be so sensitive. This might explain the focus misses some have experienced (including me).

    3. And last… accidentally, I dropped the lens…. it was a soft drop from about 40cm on a padded carpet. Since then NO focus misses…. maybe it should be part of the routine of un-boxing the lens… hehehehe (DO NOT TRY THIS AT HOME!!!).

    I will still continue to explore these missed focus issues and try to see if I can find a common denominator to those misses (in real life, I’m done with such experiments) and update you all if I find something.

    Thank you all for your comments.
    The XE1 is a wonderful camera and fun to use with great image quality.
    I hope that all bugs will be fixed soon and that we will get a better RAW support.

    Cheers,
    Sagi