AFMA Accuracy vs. Precision

[Hydrogen]

Hello Everyone,

Just wanted to pose a question to those with sufficient experience AFMA testing and adjusting several lenses and bodies.  Specifically, I want to discuss Accuracy (how 'spot-on' focus is relative to a target) and Precision (the repeat-ability of focus).

Ref: http://www.the-digital-picture.com/Photography-Tips/AF-Microadjustment-Tips.aspx

I have found that more often than not, zero AFMA adjustment on a body for a given lens yields more precision (repeatable results) even if the depth of focus is "off" (front or back) rather than once you adjust the AFMA in either direction, even by +/- 1.  This is why I am starting to believe factory calibration of body to lens is better than AFMA tweaks.

Sometimes, but less frequently, higher precision (repeat-ability) is found on an AFMA setting other than zero, but it isn't always the "best" AFMA offset in terms of image accuracy/sharpness.

I believe this is why the FoCal AFMA testing application has a AF Consistency test, which allows you to test various AFMA values in a repeatable fashion, to seek-out the best AFMA setting with the most repeatable results.

Have any of you found the same to be true?  That sometimes you may want to trade accuracy for better precision (repeat-ability).

Thanks

[AlanF]

Hello Everyone,

Just wanted to pose a question to those with sufficient experience AFMA testing and adjusting several lenses and bodies.  Specifically, I want to discuss Accuracy (how 'spot-on' focus is relative to a target) and Precision (the repeat-ability of focus).

Ref: http://www.the-digital-picture.com/Photography-Tips/AF-Microadjustment-Tips.aspx

I have found that more often than not, zero AFMA adjustment on a body for a given lens yields more precision (repeatable results) even if the depth of focus is "off" (front or back) rather than once you adjust the AFMA in either direction, even by +/- 1.  This is why I am starting to believe factory calibration of body to lens is better than AFMA tweaks.

Sometimes, but less frequently, higher precision (repeat-ability) is found on an AFMA setting other than zero, but it isn't always the "best" AFMA offset in terms of image accuracy/sharpness.

I believe this is why the FoCal AFMA testing application has a AF Consistency test, which allows you to test various AFMA values in a repeatable fashion, to seek-out the best AFMA setting with the most repeatable results.

Have any of you found the same to be true?  That sometimes you may want to trade accuracy for better precision (repeat-ability).

Thanks

What's the point of setting your AF to be consistently wrong by trading accuracy for precision?  A highly precise focus at 0 AFMA means that you will always out of focus if your correction is, say +6, whereas a wide range around +6 means at least some of your shots are in focus.  My 5D3 is very precise at the different AFMA corrections for each of my lenses, and I don't think it is true that altering AFMA alters precision. 

[cpsico]

If the body or lense is not focusing correctly i am all for factory calibration. afma is nice for a quick fix until you can send the offending equiptment in but  I would rather have it dead on accurate, esp if you have more than one camera. I just sent my 1ds in for a shutter repair and calibration, My lense that used to use afma of -2 now is dead on at 0 and is better at all focal lengthes.

[neuroanatomist]

Just wanted to pose a question to those with sufficient experience AFMA testing and adjusting several lenses and bodies.

I've got a little bit of experience with AFMA...   

I'm with AlanF on this. I have not noticed any difference in precision with an AFMA applied (based on FoCal and on substantial empirical observation). I do notice the difference with the high-precision AF points, and also between bodies (my 1D X is more precise than my 7D, which is more precise than my 5DII).

[Skulker]

Have any of you found the same to be true?

Thanks

I've not noticed it. But I haven't looked for it either. The new lenses on the new bodies were tested by Rodger at lens rentals. I don't think he mentioned anything like this, and it would seems an unusual effect from such a system.

[East Wind Photography]

And in fact some of his tests show that this is not the case.  He routinely tests AF repeatability without AFMA adjusting cameras to lenses and although the data shows AFMA is off, repeatability is pretty much the same no matter where your AFMA is set.

One other thing he discovered is that new lenses provide AF feedback to the cameras and only the new cameras make use of it (5d3, 1DX, maybe 6D but that came out after his tests were completed).  So your best AF accuracy AND repeatability are with using a new lens with a new body.  Other combinations provide just the "standard" old-school AF.    Canon hasn't really boasted much about this but Roger was cracked open a few lenses and found the sensors.  The tests show that only the newer cameras make use of it.

There were several parts to this series but here is one that describes the AF repeatability.  I'll leave it up to the reader to find the previous and future posts on the subject:

http://www.lensrentals.com/blog/2012/08/autofocus-reality-part-3b-canon-cameras

I would also respectfully submit that in all likelihood, a camera WILL AF differently and with less consistency with varying lighting conditions and the type of light used.  I will even go as far as to submit that AF systems will vary depending on the subject target used.  However I am pretty certain that AFMA doesn't affect repeatability.

Have any of you found the same to be true?

Thanks

I've not noticed it. But I haven't looked for it either. The new lenses on the new bodies were tested by Rodger at lens rentals. I don't think he mentioned anything like this, and it would seems an unusual effect from such a system.

[Hydrogen]

Thank you everyone for your responses.  Much appreciated.

I'm aware of Roger C.'s 3-part series on AF systems and lenses.  What a wonderful set of reads.  What I found most intriguing was the variance in AF consistency between lenses tested, irrespective of newer 'closed loop' sensors.  From the set he tested, the zoom with the lowest deviation shot-to-shot was the 70-300L.

See here: http://www.lensrentals.com/blog/2012/07/autofocus-reality-part-3a-canon-lenses

FoCal's AF Consistency test's ability to evaluate 'x' number of shots using a range of 'y to z' AFMA values though would make one presume there is a difference in the precision (repeat-ability) of shot-to-shot AFMA values for a given body/lens combination.

I have seen it first-hand using that test.  Unless FoCal is lying, there does appear to be a correlation.  I can find some of my saved PDF test reports and post them.

Not disputing others' experiences.  I am but a 9-year intermediate user of these technologies so not an expert by any means, but there are clearly so many unknown and undocumented variables that folks like Roger and others end up having to deduce/prove, that one has to wonder if AFMA behaves differently with 'older gen' lenses than the 'next/newer gen' barrels.

Since in an older generation lens there is no 'confirmation' back to the camera that the lens did, in fact step-move only 'x' points forward or backward, one has to wonder.

I know my first-hand experience with my 70-200 2.8 non-IS (which I still own before I picked up the 70-200 2.8 IS II recently) was shot-to-shot variability and after Canon calibrated it, not only was it just about perfect at AFMA zero, shot-to-shot precision (repeat-ability) significantly improved, so much so, it's as good as my IS II is.

[Rienzphotoz]

Interesting ... glad to learn something new.

[cocopop05]

I have the 5D Mark III with 24-105 f4 L.  I found photos did not seem as sharp as what reviewers on the internet were suggesting it should be. 

I tried AFMA and found my camera and lens required a +5 adjustment at all focal lengths to have sharpest possible focus when using the centre focus point.

Since making the adjustment,  my photos are now very sharp and my AF system is very accurate almost all the time.  So much so that I very rarely need to take a second shot. And I regard AFMA as an essential feature that all DSLR's (even entry level models) should have.

[barfbaggs]

I may be missing something, but I am perplexed by the suggestion of having your body calibrated to the lens. If you have several lenses which are off at varying +/- values, then should we have a dedicated body for each lens?

[ahab1372]

I may be missing something, but I am perplexed by the suggestion of having your body calibrated to the lens. If you have several lenses which are off at varying +/- values, then should we have a dedicated body for each lens?

That is taken care of - cameras store AFMA values separately for each lens (e.g. +5 for your 24-105, -2 for your 16-35 etc). With newer cameras and lenses, you can even store different AFMA values for different copies of the the same lens.

[neuroanatomist]

I may be missing something, but I am perplexed by the suggestion of having your body calibrated to the lens. If you have several lenses which are off at varying +/- values, then should we have a dedicated body for each lens?

I assume you're referring to the OP's statement about 'factory calibration of body to lens.'  That's not really what Canon Service does. Rather, if you send in a body they adjust it to match a 'reference' lens (presumably one dead center in the tolerance range), and if you send in a lens they adjust it to a 'reference' body of the type you have.  If you send body and lens, they adjust both to the reference counterpart, not to each other.

[klickflip]

My experience is sometimes focus is confirmed but in reality it is actually a fair bit out some percentage of the time.
On 5D II & III with all lenses there is a certain amount of frequent misfocus, I've AMFA's roughly most lens and I notice it a lot, enough to know that maybe 10% of my shots will be off focus, just enough to make me delete these shots.
Its really noticeable in portrait sessions often shooting at F9 on 85L, 50L and 24-105L some are out. Simple as, I just see it as a mechanical / electronic glitch you have to shot more to cover your back, even on still life static jobs I shoot 2 or 3 of each incase of the glitch.
Does anyone else see this recurring glitch ?
Oh and shooting the 50L and 85L at F1.2-1.8 is different story 1/4 are keepers focus wise (and 1 of 4 is way way off)  but often these shots are more about light and mood than focus.. but it would be nice to have a better level of consistency.

[barfbaggs]

I may be missing something, but I am perplexed by the suggestion of having your body calibrated to the lens. If you have several lenses which are off at varying +/- values, then should we have a dedicated body for each lens?

I assume you're referring to the OP's statement about 'factory calibration of body to lens.'  That's not really what Canon Service does. Rather, if you send in a body they adjust it to match a 'reference' lens (presumably one dead center in the tolerance range), and if you send in a lens they adjust it to a 'reference' body of the type you have.  If you send body and lens, they adjust both to the reference counterpart, not to each other.

Thanks for the reply. That certainly makes more sense than the way I understood it.

[AlanF]

I may be missing something, but I am perplexed by the suggestion of having your body calibrated to the lens. If you have several lenses which are off at varying +/- values, then should we have a dedicated body for each lens?

I assume you're referring to the OP's statement about 'factory calibration of body to lens.'  That's not really what Canon Service does. Rather, if you send in a body they adjust it to match a 'reference' lens (presumably one dead center in the tolerance range), and if you send in a lens they adjust it to a 'reference' body of the type you have.  If you send body and lens, they adjust both to the reference counterpart, not to each other.

That's right. If you send the pair to Canon for calibration then they calibrate for just that pair, and if you have other lenses then they will be correctly matched only if they have the same focus as the calibrated lens. But, if you calibrate the the AFMA yourself, then you can do it for all your individual lenses.