Backfocus with 12-40/2.8 on Olympus OM-D E-M5II ?

slncezgsi · Sep 5, 2015

Hi,

I am (a very happy) owner of Olympus E-M5II with 12-40/2.8 and 40-150/2.8 lenses since a couple of months. But while the 40-150 gives me absolutely jaw dropping detail and sharpness, I often found the 12-40 lacking a bit. I first though that I may be doing something wrong, or it is what the lens delivers as, well, it is a zoom that covers a lot of range. But later I realised (in particular in close-up photographs), that the lens tends to back-focus a little.

I will be sending the camera and lens to Olympus in October, but I just first wanted to ask whether others have experienced similar behaviour.

thanks

Digital Dick · Sep 5, 2015

I didn't think it was possible for a lens to show back focusing with a Contrast Detection Auto Focus system?

To check the focus accuracy of my lenses I photograph a ruler going away from the lens using the smallest focus square aimed over one of the ruler's numbers in S-AF mode and in Manual focus mode. I use six lenses on my E-M5 Mark II and have never seen any back nor front focusing.

****

slncezgsi · Sep 6, 2015

Yes, I read the same, but do not have enough technical knowledge on the subject to know whether it is true or not. I will do some more detailed testing once I get back.

Adamant · Sep 6, 2015

Back focus is not possible with mirrorless. You camera might be mis-focusing, but it's not back focusing.

Colin Smith1 · Sep 6, 2015

Don't really think it is back focusing but??? I have rear button focus set on my EM-1. When using the 12-24 F2.8 and focusing and releasing to re-compose, the lens will shift out of focus. Mostly happens with scenics and landscapes. I often have to shift to manual focus. Any ideas?

Egregius V · Sep 6, 2015

Can someone help me understand the claim that backfocusing is not possible in a mirrorless system? If a lens consistently misses focus, and the focal plane is consistently behind what was actually targeted, couldn't we call that back-focusing? Why does it have to have the same root cause as in DSLRs in order to be called back-focusing? Thanks much.

Question for the OP: Do you see this behavior of missing focus regardless of focus mode? I have a lens that seems to back-focus consistently, but only when using S-AF mode. In C-AF or MF, the focus tends to be spot on. (Tested on three camera bodies, so in my case it's the lens, not the body.)

Bhima78 · Sep 6, 2015

I owned three copies of the pana 15 when it first came out. All of them exhibited back focusing even in controlled tests i ran. I believe it may have something to do with a faulty contact that is telling the camera that focus has already been found when it hasnt quite found it yet.

Paul De Bra · Sep 6, 2015

We think of AF as working as follows: camera checks focus, calculates rough error, tells lens to move, checks focus again, calculates error again, tells lens to move, checks focus again and at some point the final check is focused correctly. This is what is called a "closed loop".

The reality is that in most if not all camera systems the last step is not a focus check but an instruction to the lens to move focus. As the error gets smaller the camera at some point calculates that focus will be achieved with one last small move. It just "assumes" that the move will bring the subject in focus and does not check whether that was true. This is called an "open loop". A camera with PDAF can compute the required move when the error is still fairly large. A camera with CDAF needs to first get closer and see the "error" after being still a bit too close and a bit too far. (CDAF requires the AF lens system to allow for fast reversal of focus direction.) But it can still decide that at some point the error is small enough that a small final move will be good enough. And that is where it may get it wrong...

s_grins · Sep 6, 2015

slncezgsi said:
Hi,

I am (a very happy) owner of Olympus E-M5II with 12-40/2.8 and 40-150/2.8 lenses since a couple of months. But while the 40-150 gives me absolutely jaw dropping detail and sharpness, I often found the 12-40 lacking a bit. I first though that I may be doing something wrong, or it is what the lens delivers as, well, it is a zoom that covers a lot of range. But later I realised (in particular in close-up photographs), that the lens tends to back-focus a little.

I will be sending the camera and lens to Olympus in October, but I just first wanted to ask whether others have experienced similar behaviour.

thanks

slncezgsi · Sep 6, 2015

Thanks for your points.

I am using the S-AF nearly exclusively. I have also used the MF very little so far (did not set it up properly yet), but I did not seem to have the same problems with MF.

Of course - it is possible that the problem is the user

The problems with the focus seem more pronounced in close focusing distances (like 2-3m or less). I really noticed once I was taking some semi-closeup photos at around 0.5m and wide open.

I will test more (with tripod, ruler, etc), but in general I am wondering whether it could be some calibration problem. Even if the AF itself should not be able to have this kind of problem, there is still the mechanics in between.

wolfie · Sep 6, 2015

Paul De Bra said:
We think of AF as working as follows: camera checks focus, calculates rough error, tells lens to move, checks focus again, calculates error again, tells lens to move, checks focus again and at some point the final check is focused correctly. This is what is called a "closed loop".

The reality is that in most if not all camera systems the last step is not a focus check but an instruction to the lens to move focus. As the error gets smaller the camera at some point calculates that focus will be achieved with one last small move. It just "assumes" that the move will bring the subject in focus and does not check whether that was true. This is called an "open loop". A camera with PDAF can compute the required move when the error is still fairly large. A camera with CDAF needs to first get closer and see the "error" after being still a bit too close and a bit too far. (CDAF requires the AF lens system to allow for fast reversal of focus direction.) But it can still decide that at some point the error is small enough that a small final move will be good enough. And that is where it may get it wrong...

s_grins · Sep 7, 2015

Paul De Bra said:
We think of AF as working as follows: camera checks focus, calculates rough error, tells lens to move, checks focus again, calculates error again, tells lens to move, checks focus again and at some point the final check is focused correctly. This is what is called a "closed loop".

The reality is that in most if not all camera systems the last step is not a focus check but an instruction to the lens to move focus. As the error gets smaller the camera at some point calculates that focus will be achieved with one last small move. It just "assumes" that the move will bring the subject in focus and does not check whether that was true. This is called an "open loop". A camera with PDAF can compute the required move when the error is still fairly large. A camera with CDAF needs to first get closer and see the "error" after being still a bit too close and a bit too far. (CDAF requires the AF lens system to allow for fast reversal of focus direction.) But it can still decide that at some point the error is small enough that a small final move will be good enough. And that is where it may get it wrong...

--
Enjoying the Olympus OM-D E-M5.
Public pictures at http://debra.zenfolio.com/.

Your chain of events is impeccable - exponential approach that never comes to a full completion. In OP case, based on his description, everything is fine with one lens, and no so good with another. Thus, it has to be a lens, not camera, that either overdrives or underdrives the focus point. But, once OP has no photos to show, there is a real possibility that here can be operator's error, especially close-up, especially handheld.

--
Camera in bag tends to stay in bag...

Paul De Bra · Sep 7, 2015

wolfie said:
... I can understand that as an occasional event for on-sensor CDAF , but not consistently as with a out of calibration set up between AF and lens control separated systems as in a DSLR or rangefinder.

It can be a systematic problem when there is something wrong with the lens. Granted, it is less likely as with a DSLR where the positioning of the submirror that guides the light to the AF sensors is quite critical. But when the lens is off it may indeed systematically make the wrong move even in CDAF. There are more reports of systematic AF errors in one direction with m43 cameras. Recently there was a report from an E-P5 with the same problem. So something in the system makes it possible to have systematic errors in a camera-lens combination. My explanation about open-loop systems helps understand (I hope) that for CDAF to go wrong it is necessary that the very last move the lens is asked to make is not followed by the camera checking whether correct focus is achieved.

Martin.au · Sep 7, 2015

Paul De Bra said:
We think of AF as working as follows: camera checks focus, calculates rough error, tells lens to move, checks focus again, calculates error again, tells lens to move, checks focus again and at some point the final check is focused correctly. This is what is called a "closed loop".

The reality is that in most if not all camera systems the last step is not a focus check but an instruction to the lens to move focus. As the error gets smaller the camera at some point calculates that focus will be achieved with one last small move. It just "assumes" that the move will bring the subject in focus and does not check whether that was true. This is called an "open loop". A camera with PDAF can compute the required move when the error is still fairly large. A camera with CDAF needs to first get closer and see the "error" after being still a bit too close and a bit too far. (CDAF requires the AF lens system to allow for fast reversal of focus direction.) But it can still decide that at some point the error is small enough that a small final move will be good enough. And that is where it may get it wrong...

Paul De Bra · Sep 7, 2015

Martin.au said:
I looked into this a while ago. To the best of my knowledge, no camera displayed open loop tendencies.

I have experienced myself, with my previous Canon dslrs that the system is not a one-step open loop system. When focus is way off the lens will make a large step, the AF system checks focus, then performs smaller adjustment, ... and that nobody has a firm answer on is: is the very last step a step where the lens confirms it has made the final move or where the camera checks whether AF has been achieved. The fastest AF system is one where the camera does not check AF after the final move has been confirmed by the lens. That is one reason why we might assume it works this way. If a final check of correct focus is the final step then especially in a CDAF system it would be impossible for the camera to confirm focus when focus is actually not achieved.

Way back I had a Canon 300D with the Sigma 18-125 lens. Most often the lens would front-focus but sometimes also back-focus. Overall it was erratic. If focus confirmation by the camera's AF system would be the last step then any error in focusing would be systematic. This was obviously not the case. Now, Sigma does not fully understand the Canon AF protocol, so when the lens was told to made a certain move it would move but that could well be the wrong move. But the lens would confirm it had made the move and then: in a final step open loop system the camera would believe the lens' confirmation that it made the right move and would take the out of focus shot, whereas in a final step closed loop system the camera would check focus again, see that it is off and continue to give commands and check focus until focus would become confirmed. To me it is clear: the inconsistent misfocus of the 300D with Sigma 18-125 proves that at least in this case the camera did not check focus after the lens confirmed its last move.

The dpreview article shows that with many cameras they could only verify that the AF system was not completely open loop, but could not verify whether the *final step* was open loop or closed loop.

--
Enjoying the Olympus OM-D E-M5.
Public pictures at http://debra.zenfolio.com/.

wolfie · Sep 7, 2015

slncezgsi said:
Thanks for your points.

I am using the S-AF nearly exclusively. I have also used the MF very little so far (did not set it up properly yet), but I did not seem to have the same problems with MF.

Of course - it is possible that the problem is the user

The problems with the focus seem more pronounced in close focusing distances (like 2-3m or less). I really noticed once I was taking some semi-closeup photos at around 0.5m and wide open.

I will test more (with tripod, ruler, etc), but in general I am wondering whether it could be some calibration problem. Even if the AF itself should not be able to have this kind of problem, there is still the mechanics in between.

slncezgsi · Sep 7, 2015

I will try ti shoot some samples to get better idea about the issue. But that will have to wait until we come back from the upcoming vacation. I just wanted to ask whether someone has similar experience.

Concerning the move between focusing and re-composing: that is always possible, but I have been trying to be careful and I have also experience with different systems and formats, so a user error of this type should not be happening too often. But again - proper test should allow to rule on that.

BigB · Oct 14, 2015

I've been having the same issue with the 12-40/2.8 on my E-M5 (Mk1). Olympus wants me to send them some sample photos where the issue is prevalent so we'll see what's happening (camera, lens, user). Did you ever get a resolution to your issue? On mine it's a very distinct back-focus issue. I understand the dynamics on DOF, but it happens even when the DOF is wide enough that the focus plane is distinctly off. I do use the small center AF spot as well, so it's hard to miss it when the focus spot is an eye or a nose or an entire head and the result is focused flowers behind the head or a field of trees behind the subject. It's haphazard and I cannot always recreate it when I try, but then I go out and take some normal photos and while they look in focus via the viewfinder AF, consequent review shows some photos with the off plane focus.

It's pretty disheartening when you get better, more consistent results from a plastic 40-150mm lens or the incredibly slow 12-50mm than what is considered a Pro lens. In fact, I had much better results with my old E-300 and 14-54mm lens than with this current combination.

TomFid · Oct 14, 2015

Paul De Bra said:
We think of AF as working as follows: camera checks focus, calculates rough error, tells lens to move, checks focus again, calculates error again, tells lens to move, checks focus again and at some point the final check is focused correctly. This is what is called a "closed loop".

The reality is that in most if not all camera systems the last step is not a focus check but an instruction to the lens to move focus. As the error gets smaller the camera at some point calculates that focus will be achieved with one last small move. It just "assumes" that the move will bring the subject in focus and does not check whether that was true. This is called an "open loop". A camera with PDAF can compute the required move when the error is still fairly large. A camera with CDAF needs to first get closer and see the "error" after being still a bit too close and a bit too far. (CDAF requires the AF lens system to allow for fast reversal of focus direction.) But it can still decide that at some point the error is small enough that a small final move will be good enough. And that is where it may get it wrong...

TomFid · Oct 14, 2015

Egregius V said:
Can someone help me understand the claim that backfocusing is not possible in a mirrorless system? If a lens consistently misses focus, and the focal plane is consistently behind what was actually targeted, couldn't we call that back-focusing? Why does it have to have the same root cause as in DSLRs in order to be called back-focusing? Thanks much.

Historically, the primary cause of front/back focusing in a DSLR was misalignment between the plane of the image sensor and the plane of the AF sensor due to their different optical paths. In a mirrorless camera, the sensor is the AF sensor, so that can't happen. But other things can presumably still go wrong.

Backfocus with 12-40/2.8 on Olympus OM-D E-M5II ?

Well-known member

Senior Member

Well-known member

Leading Member

Senior Member

Leading Member

Senior Member

Forum Pro

Forum Pro

Well-known member

Senior Member

Forum Pro

Forum Pro

Forum Pro

Forum Pro

Senior Member

Well-known member

Leading Member

Senior Member

Senior Member

Similar threads

Keyboard shortcuts