1. Reminder: Please use our affiliate links for holiday shopping!

Testing my m43 lenses for focus shift

Discussion in 'Native Lenses' started by johnvanatta, Aug 5, 2014.

  1. johnvanatta

    johnvanatta Mu-43 Regular

    181
    Aug 5, 2014
    Oakland, CA
    After seeing some disconcerting results from a new Panny 15/1.7 and concluding focus shift was likely an issue, I resolved to test all my m43 lenses. I'm fairly recent to m43, and was hoping this particular harpy had been banished from the realm, but my results indicate that it isn't, at least not entirely. I think these tests are probably useful to others, so I'm sharing here.

    Here's a quick introduction on focus shift for those not familiar: fast lenses with uncorrected spherical aberration can focus at a slightly different points wide open and stopped down. The EM-5 (and I believe, other m43 cameras) preferentially focuses wide open, just like DSLRs. So what appears to be a properly focused image in the EVF may end up backfocused. The fact that the EVF uses contrast-detect doesn't solve the problem unless you use the DOF preview button while manually focusing. For more information about focus shift in general, see here: http://photographylife.com/what-is-focus-shift

    The testing setup: EM-5 on a tripod, in purely manual focus mode. I manually focused wide open using the LCD with magnification, then took a picture both wide open and stopped down. The target was slanted at about 40° so that the depth of field is apparent. The vertical lines and Hs and &s was printed at a slanted angle, I was hoping it'd help but it didn't. I tried to keep everything level with the camera directly facing the paper target but there is some slop, I wasn't using rails. It shouldn't be enough to affect the test substantially.

    Focus was on the center of the X and the line directly through it. Let's start with the Olympus 60/2.8 macro in close focus (about 1:1.3 magnification) since it's easiest to see the shift here.

    Wide open, f2.8, the focus is nearly on the center line through the X. It looks slightly behind it, so I missed by a touch.
    60open.

    Now, stopped down to f5.6, the focus plane has shifted backwards even more, at least halfway between the center line and the one above it. This corresponds to a shift of about .5mm, which is probably not very noticeable, even in macro. But the magnification makes it easy to see what's happening overall, it's harder at wider angles.
    60stopped.

    Now for the Panny 15/1.7, which kicked off this entire testing suite.
    Wide open. Note that the colors change very close to the point of focus.
    15open.
    Stopped down to f4. It's a little hard to tell where the exact focus point lies, but using symmetry and color shifts, I'd estimate it somewhere between the & and the H behind the X. That's about 9mm. This is enough to severely alter an image's impact. If you're having trouble seeing the shift here, try downloading the images and viewing them in separate photoshop layers (or preview on a mac), which let you toggle between them. You should be able to see the zone of focus "sliding" backwards.
    15stopped.

    Here's the Olympus 12/2.0 wide open. Rather hazy but still possible to focus accurately with care.
    12open.

    And now, stopped down. I can't detect any focus shift at all. :smile:
    12stopped.

    And here's everyone's favorite lens, the Panny 25/1.4 wide open. Again I may be slightly backfocused here, it's hard to say with the green and violet fringing.
    25open.

    And stopped down. I'd say that peak focus is around the bottom of the & behind the X. That's a shift of about 4mm. Testing of assorted objects around my room (not test charts) indicates that it's just enough to be noticeable if precise focus is required.
    25stopped.

    I also tested my Oly 12-40/2.8 zoom at 12mm, 25mm, and 40mm; the Oly 45/1.8, and the Oly 60/2.8 further away. For the sake of brevity I'll omit the pictures and just say that no focus shift was detectable in any of these cases. It seems Olympus' designs prioritize spherical aberration, a laudable approach.

    Conclusions:
    If you get the Panny 15/1.7, be aware that if you aren't shooting wide open, you'll get backfocus. These tests were necessarily done at close focus (sensor was about 30cm from X for the 15mm), but I did some landscape testing near infinity, and saw substantial shift there as well. One workaround is to manually focus while holding the DOF preview button down, which works within its many limitations. Personally I'm very disappointed in this lens design and returned mine. I'd give up some overall sharpness for the ability to accurately focus easily. For the Panny 25/1.4, consider doing the same for some subjects, though it's far less critical. For the Olympus lenses, it looks like we're all clear, let's hope they keep up the good work.
     
  2. Paul80

    Paul80 Mu-43 Veteran

    254
    Jul 6, 2014
    Just wondering, are you seeing focus shift or just the optical effect of depth of field, which extends farther the the rear than in front of the focus point.

    Paul
     
  3. johnvanatta

    johnvanatta Mu-43 Regular

    181
    Aug 5, 2014
    Oakland, CA
    Good point. I assumed the DoF would be symmetric at this range (and it certainly appears to be in the 12mm shot) but it's worth checking.

    Using the formulas for D_N and D_F given on wikipedia (halfway down the page, right above the Hyperfocal distance section)
    https://en.wikipedia.org/wiki/Depth_of_field#Derivation_of_the_DOF_formulas

    Values: .01mm for c, the DoF circle of confusion (roughly twice the pixel pitch), and 240mm for s, subject distance (an approximation, since the lens is not a single element)

    Using these equations, I calculate that for the 15mm at f4, similar depths of field are 9.2mm forward and 10mm behind the point of focus. A different of only .8mm. The blur should be essentially symmetric. Even the 12mm has only about 2mm of difference using the values given, and I can't say I see any DoF asymmetry in that image. For the longer lenses the asymmetry is negligible, since they are even closer focused relative to their focal length.

    So, I'm confident that it is indeed reasonable to assume DoF is symmetric here, and focus shift is the culprit.
     
  4. DennisC

    DennisC Mu-43 Regular

    76
    Jan 24, 2010
    Cheshire UK
    I saw similar back focus effects with my Olympus 14-42.EZ lens.

    I would suggest that increased pixel count has meant that our attention to resolution is greater than ever and issues like shutter shock and focus shift are easier to detect.
    In my case I concluded that just because my E-M10 shows a green focus square doesn't always mean that it is focusing exactly on that spot.

    Anyway after a camera firmware update I gave up worrying as things looked better
     
  5. johnvanatta

    johnvanatta Mu-43 Regular

    181
    Aug 5, 2014
    Oakland, CA
    Dennis, I agree that higher resolution sensors are exposing the limitations of the lenses and focus systems that were once masked. But what you were seeing sounds more like autofocus accuracy than focus shift. Focus shift generally only happens on fast lenses, f/1.8 or maybe f/2.8. And I've noticed that the contrast-detect likes to land backwards of where I want it. But focus shift is insidious because it still happens even with precise manual focus, as my testing hopefully shows.
     
  6. MayaTlab

    MayaTlab New to Mu-43

    7
    May 20, 2014
    At least that's one thing that is common between Panasonic Leica-branded lenses and real Leica lenses : focus shift ! It's been quite a major issue for Leica in the digital age - to the point that they redesigned several lenses to take care of that issue such as the 35 lux.
     
  7. Replytoken

    Replytoken Mu-43 Hall of Famer

    May 7, 2012
    Puget Sound
    Ken
    I have a copy of the 12-40 that has been giving me on and off fits at close focusing distances, and I am still trying to determine how much is lens, how much could be user error, and how much could be attributed to other factors. It is still an amazing lens at less than close distances, and I am not sure that I am experiencing a focus shift issue, but I would not say that Olympus if immune from any possible issues.

    I am still need to read Nasim's article when I have the time to digest the entire piece, and I am curious as to what results might be obtained if I manually focused with the DOF preview engaged.

    Thanks,

    --Ken
     
  8. fortwodriver

    fortwodriver Mu-43 Top Veteran

    959
    Nov 15, 2013
    Toronto, Ontario, Canada
    Frank
    Um... even 30,000$ Cine lenses focus differently depending on their apertures.
    Focus Shift seems to be related to the optical formula used to create the lens.
    There's no real magic there.

    On the other hand, if a lens is having trouble reliably focusing at close distances, that may be firmware or the granularity of the encoders used to report back how far the lens has moved through its focus range. Some old AF Nikkors were so coarse, that you could kind-of predict when they would fail to focus - but that was also with earlier PDAF open-loop focusing.
     
  9. Replytoken

    Replytoken Mu-43 Hall of Famer

    May 7, 2012
    Puget Sound
    Ken
    I had a chance to read Nasim's article, and while he talks about LIve View in DSLR's he did not touch upon mirrorless cameras. I realize that Live View is a contrast-based focusing method, but is it identical to the contrast-based systems in mirrorless cameras? Not that that may make a difference WRT this issue, but it would still be nice to know if there are any differences.

    --Ken
     
  10. BigOwl

    BigOwl Mu-43 Regular

    61
    May 3, 2013
    San Antonio, Texas USA
    Al
    Help me out here. I have tried to follow this thread, but am a little confused. Let's say I have a 15mm 1.7 and set the aperture ring to 1.7 then manually focus on a convenient target. Then without touching the focus ring, I move the aperture ring to 4.0. Are you saying I will no longer be focused on the target? If that is the case, was there movement of the lens elements? In my mind, the aperture ring should only move the diaphragm blades. Thanks for some clarification.
     
  11. Replytoken

    Replytoken Mu-43 Hall of Famer

    May 7, 2012
    Puget Sound
    Ken
    I will not speak for the OP, but if Nasim's article is correct, what you are seeing is the effect of spherical aberration. Closing down the diaphragm reduces the effect of the speherical aberration, and as the focus point "concentrates", it seems to do so at a different plane of focus than if the lens were wide open.

    --Ken
     
    • Like Like x 1
  12. BigOwl

    BigOwl Mu-43 Regular

    61
    May 3, 2013
    San Antonio, Texas USA
    Al
    Thanks, Ken. Then perhaps I could tweak the focus slightly after stopping down. If the image is bright enough to do so, that is. And if the increased depth of field does not make that too difficult.
     
  13. johnvanatta

    johnvanatta Mu-43 Regular

    181
    Aug 5, 2014
    Oakland, CA
    Replytoken, it's definitely worth checking for focus shift, though I'd expect it not to be the case if the lens is otherwise good. But sample variation is always a pain, and always unpredictable. Also, my testing was done at fairly close focus (28-47cm from the sensor), but that's nowhere near minimum focus for the pseudo-macro capable 12-40. It's possible there's a small shift that I deemed negligible that becomes more problematic at minimum focus/macro subjects.

    BigOwl, yes, the key is that stopping down removes spherical aberrations, changing the exact point of focus without moving anything. If you have a 15/1.7, try it out, making sure to keep the framing exactly the same (tripod very helpful). There should be enough shift on that lens to be readily apparent on close examination. Pick a target with some clear detail just behind the chosen focus point, since the shift is backwards.

    Compensating by dead reckoning is pretty tough to do. Focusing stopped down in the EVF is possible, though it requires some finger dancing, and it isn't always easy to get peak focus beyond f/4.0 since the DoF is large.
     
  14. Replytoken

    Replytoken Mu-43 Hall of Famer

    May 7, 2012
    Puget Sound
    Ken
    I will check a bit more, as when I bought my refurbished 12-40, I thought it seemed to be focusing behind the plane of focus that I selected with the single point area in the EVF when I was focusing on close objects at 40mm. In many images it seemed like the area just behind the subject (usually the center of a flower) seemed to be in sharper focus than the subject itself. At normal distances, this was not evident. Some close focus images did appear to be fully in focus, so I want to sort out potential camera AF issues, from lens issues, from user issues. I am normally not one to pull out the tripod to test, so I usually just shoot a lot and see if any patterns start to emerge. Not scientific, but if an issue is there and is impacting my images, I can usually identify it. Solving it, however, can be a whole different can of worms.

    --Ken
     
  15. johnvanatta

    johnvanatta Mu-43 Regular

    181
    Aug 5, 2014
    Oakland, CA
    I noticed that the contrast detect likes to go to the back of the flower too, so now I'm trying to get in the habit of manually focusing for the last mile when I'm able to. With the MF-assist mode on the EM-5 enabled, its really easy, autofocus to get close then just tweak the focus a bit at the automatic 5x magnification. I love that feature. Best advantage of the EVF IMO.
     
  16. Replytoken

    Replytoken Mu-43 Hall of Famer

    May 7, 2012
    Puget Sound
    Ken
    That is the same technique that I primarily use, but it also means that I have have to be even more still after the focus, as I do not want to lose my framing, and that means waiting for the magnified frame to close.

    --Ken
     
  17. Clint

    Clint Mu-43 All-Pro

    Apr 22, 2013
    San Diego area, CA
    Clint
    The big question is - how much does it affect your every day photography? And if it does, do you know your equipment well enough to make necessary adjustments to compensate?
     
  18. T N Args

    T N Args Agent Photocateur

    Dec 3, 2013
    Adelaide, Australia
    call me Arg
    Not the macro, that looks like a bigger problem than the others. Macro focus is so critical.
     
  19. T N Args

    T N Args Agent Photocateur

    Dec 3, 2013
    Adelaide, Australia
    call me Arg
    John, I'm probably not the first to suggest this (seeing as how you tend to take your questions to multiple forums simultaneously in a 'shotgun' approach), but IMO you probably got a suspect example of the 15mm lens.

    I don't think your conclusion "If you get the Panny 15/1.7, be aware that if you aren't shooting wide open, you'll get backfocus" should have been in the form of advice to others. You have no clue that your lens is a good sample. You *do* have a clue that it is a bad sample.
     
    • Like Like x 1
  20. johnvanatta

    johnvanatta Mu-43 Regular

    181
    Aug 5, 2014
    Oakland, CA
    Macro focus is critical indeed, but a half millimeter is still a pretty small change. Take an insect eye, for example. The eye itself is going to be at least a millimeter or two deep. Even the area covered by specular highlights is probably about half a millimeter. I don't know if it matters much where in the specular region the focus hits at f/5.6, and so far I'm not seeing any real-world problems with it. Though my macro tends to be handheld spray and pray :)

    As for the 15mm, two things make me believe that it isn't just a bad sample. I think focus shift is generally not something that sees much sample variation. The lens checked out otherwise, so it'd have to *only* a spherical aberration problem. Second, I had a brief discussion with Lloyd Chambers, whose testing is even more rigorous than mine; he also saw substantial focus shift in his sample. Unless I see tests specifically for focus shift yielding negative results (link me if you have some), I'm standing by my conclusion that there is a problematic amount of focus shift in the lens design.