Good day everyone,
This is a more generic question: I read that mirrorless cameras do not require AF microadjustment (AFMA) anymore, since AFMA is used to calibrate the focus offset between APS(-C) sensor and actual AF sensor (typically phase detection) in a DSLR body, and on mirrorless cameras focusing is done using the one and the same main sensor (using contrast detection).
Hence, most mirrorless cameras including Canon, Nikon, ... do not have an option in the menu for this, not even Canon's top of the range M5. Some mirrorless Sony Alpha series cameras appear to have an option for AFMA for A-mount (read: DSLR lenses), but not for E-mount (read: mirrorless) lenses.
So far so good - but what does that mean if I were to use a EF(-S) lens using an adapter on e.g. a M5?
I'm asking because I had major trouble with multiple iterations of adjustments for my Tamron 17-50 VC f/2.8 on my EOS 1000D some years ago - as far as I understood it back then from Tamron Support the lens had to be adjusted a *lot* to my 1000D body. It had severe front/back focusing issues. I also have e.g. a Canon 35mm f/2 (non-IS) and Canon 70-200 f/4 L - both of which I bought used and that are fine on my 1000D, but I do not know if anything was ever adjusted on those.
If I were to buy say an M5 with the EF(-S) adapter from Canon ... since e.g. the Tamron lens was adjusted, would it have a severe front/back focus again? Would it need to go back to Tamron support to zero/reset it? Or would there not be any focus problems expected?
Hope someone knows something about this.
Thanks all!
Going mirrorless means buying lenses for a mirrorless camera. Adapters are not even second best, always slower and restricted in functionality (exceptions possible). AFMa is only a thing for DSLR, means also Sony A-mount. I doubt you can make E-mount on an A-mount camera body work at all, the flange distance is shorter, so no compatibility.
Either way you get yourself a Canon DLSR with AFMA, so you're no longer depending on doubtful service which make "a lot" adjustments and can do the adjustments yourself - that means hours of wild fun  or you choose the other road and furtheron don't care about AFMA. There are other things to care with mirrorless - only the more expensive ones focus very fast and are sort of convincing with focus tracking - but each cheap mirrorless does a better, more accurate job to focus than most DSLRs in LiveView mode.
No, there is no "offset between sensor and actual AF sensor" that AF micro adjustment is supposed to address. That is a fundamental error in understanding how things work.
AF in DSLRs can work just fine with most lenses, and then you have a lens which does not give accurately focussed results, and you then blame it on an "offset between sensor and actual AF sensor"? How does that compute?
There are three things that make for occasional AF accuracy issues with PD AF in DSLRs.
- PD AF algorithms are made to avoid endless focus hunting loops. They do not check if focus is actually reached. The are geared for focus acquisition speed, basically.
- Some lenses can do inaccurate, or "crude" AF steps when the camera is going through its AF algorithms. They basically mess up what the camera accurately says about where focus should be reached. Your Tamron 17-50mm is an example of such a lens. The Tamron service center has tried to lessen the inaccuracy of the steps the lens makes, to make it step more in line with what the camera body tells it to do. The Canon 35mm f2 and 70-200mm f4 L do not make such inaccurate steps, and hence produced better focus results without the need of "calibration".
- PD AF sensors are sensitive to certain parts of visible light spectrum, and probably even to less visible IR and UV parts of the spectrum. A lens may not focus every part of the light spectrum on the same focus plane. If just a part of the light spectrum is on a different plane when focussed, and the PD AF sensor is sensitive to that part of the spectrum, the camera can see things being in focus, when you human, who sees the whole spectrum of visible light as "one" (the image) will see the image as out of focus. This is a case where AF micro adjustment can be of interest. However, it is not a perfect solution. A lens may not have the same characteristics everywhere in its focus range. And if you follow Nikon forums, you will find that some people complain that when they perform AFMA adjustments with certain lenses on certain quite recent DSLR bodies in daylight conditions, they get out of focus results with (certain) indoor light sources.
Hope that sheds some light on how PD AF works, and why you have had trouble with that Tamron.
On contrast detect AF: it works totally differently from PD AF. With PD AF you can see in which direction you have to move to get to focus. With contrast detect AF, you just check contrast from frame to frame. If contrast increases, you have not reached focus yet. If focus decreased all of a sudden, you have passed focus and need to take one step back, basically.
With cameras with live view with PD AF implemented in the imaging sensor, like an EOS M5, they combine both ways. But the PD AF done by the imaging sensor will not be influenced negatively by certain parts of the light spectrum like you see with (some? most?) DSLRs.
Your Tamron will not behave the same on the M5 as it does on your 1000D. There are some lenses which will not focus well or at all with live view, because their AF implementation is just that horrible, but probably that Tamron is not one of those. But you just have to try that out and see. The work your Tamron has seen being done to it in service centers will not have a negative bearing on how it functions on an EOS M5, or another Canon APS-C DSLR, for that matter. Your Canon lenses will focus pretty accurately on an EOS M5 (or even a EOS M (the 1st EOS M camera).
By the way, mirrorless cameras can still deliver slightly or fully out of whack out of focus results. Sometimes that is to blame on lenses (see my remark above), sometimes on the steps a lens makes (which will account for slightly OOF results), and sometimes it is a mystery on how the camera gets to a totally OOF result (some Sony A7 models/lens combinations at times give puzzling focus results once in a blue moon, for instance).
Missed focus is not an exclusive feature of mirrorless or DSLRs - all systems can fail under certain circumstances. The idea of a confused PD AF module because it is sensitive to a bigger spectrum of wavelength is something new to me, but makes sense.
"How does that compute?" good question, but in reality I see differences between all kinds of lenses on a certain DSLR, be it genuine or third party ones. I also see differences depending on distances between camera and subject. If lenses just work fine, they might be slow ones or always used at f/8 at infinity. In my experience it's a win in the lotto if one lens needs no AFMA.
Just saying - PDAF on mirrorless cameras is also accurate.
That is ... except when a lens suffers from RSAs. It depends on the camera and/or light situation whether the AF is applied to the stopped-down lens or not.
Quote:No, there is no "offset between sensor and actual AF sensor" that AF micro adjustment is supposed to address. That is a fundamental error in understanding how things work.
AF in DSLRs can work just fine with most lenses, and then you have a lens which does not give accurately focussed results, and you then blame it on an "offset between sensor and actual AF sensor"? How does that compute?
There are three things that make for occasional AF accuracy issues with PD AF in DSLRs.
...
Well, of course there is 4th reason: mechanical tolerances in case of separate AF sensor, as with DSLR. Which are quite complex due to 3 parts getting into optical path for AF aquisition: main mirror, secondary mirror and AF sensor itself.
The mechanical tolerances are there - but they don't change with the lens. However, each lens has it's own tolerance as well and after it is adjusted by the manufacturer, it depends how big the tolerance field of the camera and lens adjustment really is. And since there's more than just one reference to adjust the lens in the factory, we do face these tolerances.
Thanks for all the replies - a very interesting read into the depths of AF in general.
I dug out an email from the Tamron UK service center (Intro 2020 Ltd.) where I had my lens adjusted in 2015 (I lived in Ireland at that time, back in Deutschland now) - and this is what they had adjusted:
- The paper report stated "Adjusted AF data" - nothing more, nothing less.
- On further enquiry with their support, this is copy and paste from their email: " I have spoken with the technician who has clarified the manual adjustment you are enquiring about. The infinity point was moved along with adjustments made to the resolution power. "
So ... if there was a manual adjustment made, I am pretty sure if using the lens on a different DSLR body it may need AFMA. But what about on a mirrorless like M5? Still not quite clear to me if any manual adjustments would have any impact if using on a mirrorless body.
I'm thinking of buying either an 80D so not problem there .... or an M5 which since yesterday is really cheap on Amazon (less than 1000€ with basic kit lens, then 100€ rebate from Amazon directly + 100€ cashback from Canon) - yes I did check with Amazon support and they told me the M5 is eligible for both rebates/cashbacks 
Interesting stuff.
I am strugging with a Tamron 17-50 has well (non-VC). It has a tendency to severely front-focus at times. We're talking static shots where I point the lens at infinity, I have only the center AF point activated, yet focus is somewhere near close-focusing distance. The next shot could be spot on.
Since it's an intermittent problem (50% of shots or more) this is not something for AFMA, which fixes lack of accuracy rather than lack of precision. It irks me that in each case I get focus confirmation.
Focus in live-view works flawlessly, as does focus on my old 400D (and 300D before that, now sold).
I just had an idea. What if I started using AI servo instead of one-shot, even for static shots. It should hunt when AF goes awry. I'd lose focus-recompose unless I use back-button focus. I'll have to give this a try later today.
Sure, live view focus should work fine as long as lens is capable of performing required focus movements in real time. Being contrast based AF, it works as BC explained above.
|
