AF micro-adjust made easy

HoppyUK

HoppyUK

Suspended / Banned
Messages
23,200
Name
Richard
Edit My Images
No
Ingeniously simple and accurate way of setting AF micro-adjustment on this link to Fred Miranda forum http://www.fredmiranda.com/forum/topic/1187247 Some good discussion too, and also links there to same on DPReview.

It works by using the camera's own AF calibration data to pinpoint the sharpest setting, as indicated by the viewfinder AF confirmation light. As others have found, I've just tried a few lenses and it works a treat, matching my own carefully crafted settings exactly - quick, easy, accurate :thumbs:

Thanks to our Tim Dodd for spotting it :)

Edit: you don't need any fancy target, don't even need to take a picture - you just need live view.
 
Last edited:
How...obvious! One of those 'why didn't I think of that?' solutions. Thanks, Hoppy. :thumbs:
 
Interesting. Going on this test, just done now, my Sigma 35 needs +13 adjustment, where I had thought it grand at 0! Must get my eyes checked!
 
This is a very good technique, I tried it with my sigma 50 and it nicely confirmed it needs +12 on the 5Dii - which is what I had it set on to start with.
 
I've not tried it yet, but does it still require obscene distances on longer lenses?

Also can this be done indoors, ie lit by a standard bulb or should it be done outdoors in good light?
 
Always wanted to fine tune the lens to the camera.

So I take it one it is set for that lens the camera remembers it and when yuo change lens it will auto change to that lens???
 
I'm very interested in this having never MA'd before.
Also interested in the quality of light question. Do it in daylight, focus on the join in a brick wall. Or inside and a magnet on a fridge for example.
 
I would suggest that any accurate A/F adjust would need good light and contrast or it wouldn't be accurate.
 
In my opinion the best advice is to calibrate in conditions of lighting and temperature and at distances which most closely mirror your expected usage of that lens.

I don't think that calibrating every lens you have at 3m with a weedy tungsten lamp in a centrally heated home is going to necessarily yield results which will work well in the bitter winter cold in bright daylight when shooting wildlife, motorsports or aircraft.

As almost all my shooting is outdoors in fairly good daylight I calibrate my gear in the back garden, picking targets at distances which basically break down to close (~3m), middling (~6m)and distant (~18m) rather than at specific multiples of focal length. That'[s more than good enough for me. If I want even greater distances for my long lenses then I'll pick a chimney stack a few houses down or a distant electricity pylon. Anything is fair game so long as it is at a realistic distance for your needs.
 
Ok been looking through the camera settings and got my head round how to do it all now.
Just one more thing, if say on my 17-55 which focal length would be used, would I tune it at 55mm or 17mm

Thanks
 
Ideally you would try calibration at both ends of the zoom range, and perhaps even somewhere in the middle too. If you get any variation in the outcome you have to make a decision on how to compromise on the final choice you make. e.g. you might find you get +7 at 55mm and only +4 at 17mm. 55mm is going to be the most sensitive to inaccuracy due to the shallower DOF, so you'll want to favour that end of things more strongly, so you could stick with +7 or maybe reel it back a notch to +6. Either way, anything from +4 to +7 is going to be better than leaving the calibration unaltered at 0.

p.s. I guess in the case of a lens covering the 17-55mm sort of range the odds are that the short end will be used more for scenic type shots, where thin DOF is very much not the requirement and you'll be stopping down anyway, whereas the 55mm end might be used for portraiture or even a bit of action, in which case having the AF spot on when wide open will be more important still.
 
Last edited:
If it helps anyone, there is a visual representation of what's going on in the graph produced by the latest release candidate of the Focal software. Have a look here....

http://www.reikan.co.uk/focalweb/in...ease-candidate-available-for-focal-pro-users/

and scroll down the the graphic under the heading for TurboCal. The red blobs show the AFMA values tried by the software, with those at the top of the graph showing focus confirmation success and those at the bottom showing focus confirmation failures. There is one blob showing 9/10 successes for the 0 AFMA value. From the results obtained the software has plotted a graph and deduced that +6 is the sweet spot for the AFMA value. Given the 100% success rate for values of +4, +8 and +12, the 90% success rate at 0 and complete failure at -4 and +16 this seems a pretty reasonable interpretation of a midpoint figure for the successful values.

If I understand the software design correctly it is doing nothing more than the manual procedure described on the FM thread except greatly speeding the process and providing high statistical reliability through the shear number of focus confirmations attempted (10) for each AFMA value.
 
Going to give this a go on my D7000, A model which everyone seems to think has focus problems.
 
{Going to give this a go on my D7000, A model which everyone seems to think has focus problems.)


Lets know how you get on., Ian.
 
For Canon users their is also an approach to MA in the current EOS magazine that uses live view to manually focus then checking AF by looking at the focus panel to see if it moves.
It's explained better in the mag :D
 
Just tried this with my 18-55mm kit lens and 35mm 1.8 on my D7000.
The 35mm was pretty much spot on, the range was -13 to +12 which put the midpoint at basically 0 or -1.
The 18-55mm i did it at both the 18mm and 55mm ends and get very different results. At the 18mm end the range is -8 to +16 so the mid point would be +4. At 55mm the range is -12 to +11, so mid point is 0 or -1.

The D7000 does allow you to tune 12 lenses and 100 saved tune points for each lens so you can change the tune point from ones you've saved in the past. A bit of a pain on a zoom lens to constantly switch tune points though for the focal lengths but at least you can do it, and with a prime lens once tuned you don't even have to think about it.
 
Ok had a bit of spare time from work so got three of my shorter lens done.

really easy to do so happy.

My 17-55 lens was a bit out at 17mm it needed to be +1 and at 55mm it needed to be -8, so I just knocked it back a couple of increments and that will be better than it was.

My 100mm macro was out a bit it needed -4 and my tamron 10-24 was nearly the same at both end bar 1 notch so all is good.

Just need some nice weather to do the 100-400 and 70-200 now

spike
 
I'm not sure, that version requires judgement and whenever I've nudged back on forth with those arrows it takes a while to decide which is sharper, so maybe for some its quicker but I doubt it will be for everyone.
 
JohnN said:
I'm not sure, that version requires judgement and whenever I've nudged back on forth with those arrows it takes a while to decide which is sharper
Click to expand...
If the difference between two MFA values isn't obvious then it makes no difference which you end up using.
 
I'm not sure either, but whichever method you use, the target, and distance, is critical. The target must be flat, and square to the camera. There is no guide about distance that applies universally, but it's got to be realistic and not artifically close. Use your judgement.

This is a target I use. It could hardly be easier and has the major benefit that you can see on the rule where the focus actually is, how far it's out, in which direction, and when it's right.

Edit: if the camera has a decent LCD, that's fine for checking. Just zoom in on that - no need for computer.

IMG_3381-1.jpg
 
Last edited:
Excellent Hoppy, I will be using that set up when I get round to it. Thanks for sharing.
 
I got a 50mm f1.8 mk1 today and went to do that one too.

Now on my 50D the range goes from -20 to +20 and as by the way to do it above it went from +3 to -20 so set it to -8.

Now what happens if you get to -20 or +20 an that is not enough????
 
SpikeK6 said:
I got a 50mm f1.8 mk1 today and went to do that one too.

Now on my 50D the range goes from -20 to +20 and as by the way to do it above it went from +3 to -20 so set it to -8.

Now what happens if you get to -20 or +20 an that is not enough????
Click to expand...

If you read all the links, and the links on the links, there's a suggested way around that, based on the assumption that the final adjustment setting will be within the +20/-20 range, even the total range covered by the AF confirmation light extends beyond that in one direction. Basically you have to nudge the focus slighly out as a starting point to bring the range within +20/-20, and then do the tests.

Not tested a MkI 50 1.8, but I surely hope it's got better AF than the notorious MkII. I don't bother with AF micro-adjust on my MkII - it's so inconsistent shot to shot it's not worth bothering :eek:
 
Thanks for that Richard. To be totally honest I did not read all the links or even the links on the links:D as all my lens where well within the range.
 
Am I missing something here?
Isn't this method only testing and showing the adjustment difference between contrast and phase detection focusing?
Sorry if I sound a bit thick, I watched the video and can't get my head round how this is fine-tuning a particular lens.
JohnyT
 
Johnytuono said:
Am I missing something here? Isn't this method only testing and showing the adjustment difference between contrast and phase detection focusing?
Sorry if I sound a bit thick, I watched the video and can't get my head round how this is fine-tuning a particular lens.
JohnyT
Click to expand...

Yes, it sounds like you are ;)

Contrast-detect AF has got nothing to do with it, though it can be a useful reference to check against because it reads off the actual focal plane (sensor).

Phase-detect AF works to a simulated focal plane distance. AF microadjustment ensures the simulated plane is in fact at the same distance as the actual focal plane.
 
Okay, how's this for easy. I've just seen dot tune will be included in magic lantern, so attach lens, set up and it does four full sweeps of the -20 to 20 range, gets the value and its done. All fully auto.
 
Video of ML implementation here....

[YOUTUBE]58enf18Q0l4[/YOUTUBE]

Just think what this will do for "in the field" calibration. :D
 
Last edited:
damn, beat me to it - I was just coming to link that :)

looks good doesn't it!

Also it looks like with ML they can take the AF value to range to -100 to 100, although thats very much in testing.
 
Last edited:
tdodd said:
Video of ML implementation here....

<snip>

Just think what this will do for "in the field" calibration. :D
Click to expand...

Wow, super-cool :thumbs:

Edit: Demo shows a prime lens on Canon 5D3, so I'm guessing they're still working on a version that will check and install at max and min ends of a zoom, as the 5D3 allows. Which begs the question, could they introduce this max/min zoom facility to other models that don't normally have it? Or max/min and middle? Or do something similar for near and middle distance focusing? That would be pretty special :thumbs:
 
Last edited:
Ok, i've just performed a dot tune on my 70-200 f2.8 (mk1) I have performed the test at 70mm and again at 200mm, and, just out of curiosity i did it again at 135mm. Anyway, from this i've got 3 different values:

@70mm = -1
@135mm= +8.5
@200mm= +2.5

So which one should i go for? or should i pick somewhere between the 135mm & 200mm value? so +5? or maybe somewhere between the 70mm & 200mm result? +3 maybe?

Hmmmm....
 
Last edited:
hehe, the curse of the extra test!

First thing I would do is retest at 135 and confirm the results as this seems a little odd, although I'm sure I have read this before somewhere.

Also keep an eye on Magic Lantern as their including dot tune and it might make things a bit easier.
 
haha... yes i know, even before i did the extra test i just knew it would throw up a weird result just to confuse the issue! it was bound to wasn't it. I will re-do the test at some point at all 3 lengths, as i did it a bit rough and ready anyway as i was in a hurry, i didn't have the tripod collar (so not quite as stable as it might of been) on the lens and i used a brick wall rather than a chart. Other than that, the test was very controlled! lol
 
Not to go off topic, but have they sorted Magic lantern yet? on the 7d you had to reinstall the ml firmware each time you turn the camera on. Have they gotten round that issue yet? it was putting me off installing it....
 
You must log in or register to reply here.
Back
Top