Take your photography to the next level and beyond...

  • NEWS
  • REVIEWS
  • INSPIRATION
  • COMMUNITY
  • COMPETITIONS

Why not join for free today?

Join for Free

Your total photography experience starts here


AF Fine Tuning - does you camera have it?


Caisiel 4 17 Wales
13 Sep 2012 7:21PM
I have messed about with the af micro adjust on a number of my lenses only to finally revert back to
'0' af micro adjust on my Sony a850. I was worried my small lens collection was failing to accurately focus. This paranoid adjusting was applied to my zeiss 24-7" and 85mm zeiss, I soon realised that it was more inefficient focussing by me, as opposed to an issue with the optics. Hence deemed this a useless feature, until I set up and tested my older Minolta 100mm macro, and 135mm af lenses. They received adjustments of -6 and +9 respectively. Increasing my hit rate considerably. My question to those more educated ones out there, do lenses lose accuracy with AF with age?? If so this feature could prove to be essential for some of us who find it hard to leave good optics in the drawer!

Join ePHOTOzine for free and remove these adverts.

strawman 13 22.1k 16 United Kingdom
13 Sep 2012 7:57PM

Quote: unless you know of another way a lens can fool a camera into focusing incorrectly
Yes I do (well on some AF systems) as I mentioned the lens passes information on where it is in its focus range to the camera. If it measure that incorrectly then the calculation of how much to move the lens will be wrong as so a focus error will occur. This is why you have lens by lens adjustment. It is what is relied on to make classic phase AF systems.

As I mentioned earlier contrast AF works differently from phase AF so the contrast system takes a stream from the sensor and so will compensate for lens position errors while the phase system does not normally calculate the Af position more than once. It is why it is faster, but harder to get correct. It is why I mentioned earlier phase and contrast AF have different accuracies to static and moving targets and different response times. .

Depending on how the lens gets position information it may well age. For example if it uses a contact sensor technique like potentiometers. Think of it this way.

Phase AF one sample/calculation to measure focus then move the lens an amount that depends on lens info/data
Contrast AF move the lens and take a lot of sample to detect whether focus is getting closer or further away, an iterative process.
Steppenwolf 5 1.2k
13 Sep 2012 8:24PM

Quote: Yes I do (well on some AF systems) as I mentioned the lens passes information on where it is in its focus range to the camera. If it measure that incorrectly then the calculation of how much to move the lens will be wrong as so a focus error will occur. This is why you have lens by lens adjustment. It is what is relied on to make classic phase AF systems.



Sorry but this doesn't make sense to me. I see no reason why the camera should be interested in where the lens is in its focus range. The way the Minolta/Sony AF works (on DSLR/SLTs) is that the camera makes an initial calculation as to how much to move the lens focus. It then moves it and checks (when in single focus mode). If it's not in focus it makes another calculation and moves the lens and checks. It basically iterates until the picture is in focus. With a compatible lens you only hear one movement - there may be small corrections but they're too fast to detect. However, I had a Kenko TC with a duff chip once and you could see and hear the lens oscillating back and to until it finally found focus. It was plainly telling the camera the wrong information (probably on gearing, although it may have been focal length) but the camera still found focus - it just took so long it was pointless.

If a camera's in continuous focus or tracking focus mode then it seems to just make one calculation, but in single focus mode it iterates.
Sooty_1 6 1.5k 221 United Kingdom
13 Sep 2012 8:36PM

Quote:It's actually not called "manufacturing tolerance" in that manufacturers don't tolerate it


You're either being facetious or you don't know much about 'tolerances'..... I don't see any smileys..?

Everything is manufactured to a tolerance...the degree to which an item can deviate from the design norm and still be acceptable. Some items have larger tolerances than others, but even on an optical product, there is still a variation in manufacture. The fact that a 'correction factor' is added to the autofocus computation, to compensate for a measured error, does not mean the the camera/lens leaves the factory absolutely spot-on for focus. That correction factor will also have a tolerance.

The question is really "how much is too much" to fail quality control. The reasons that old Soviet cameras had such varied quality was the agricultural production methods used gave rise to large tolerances, as opposed to the West Germans with their highly skilled and accurate processes. Thus with the Eastern European stuff, if you got two items at the same end of tolerance you got a good product, etc etc.

As Mike alluded to previously, if you have items at the opposite ends of the tolerance scale, there could well be noticeable errors.

Nick
Sooty_1 6 1.5k 221 United Kingdom
13 Sep 2012 8:38PM
The other thing in lenses is backlash in the focus drive it might depend which direction the lens is focussing how much error you get.
strawman 13 22.1k 16 United Kingdom
13 Sep 2012 10:25PM

Quote: The way the Minolta/Sony AF works (on DSLR/SLTs) is that the camera makes an initial calculation as to how much to move the lens focus. It then moves it and checks (when in single focus mode). If it's not in focus it makes another calculation and moves the lens and checks.
If that is true then that may a big difference from some of the Nikon and Canon AF strategies and perhaps may be why there are differences in sport/wildlife performance. Are you certain the modern Sony cameras do this for fast photography like for example sports shooting and for motor drive operation?.

I know on my camera (and its 5 years old) I have the choice of letting it do a double check type focus or go for speed. The Nikon and Canon systems have the capability of one lens movement. If it were a constantly iterating system then yes it would work as you say but it would be slower and to be honest you would have to ask on an SLT system why not just use the contrast video AF like the m4/3 and NEX cameras do. Does the new sony use the SLT based AF sensor to do the gross focus calculation and use pixels on the main sensor to do local contrast AF to trim??

The advantage I can see of the Phase type AF is it can in one calculation and one movement get to the correct focus. But to do that movement in one it has to know information from the lens. And that will be very sensitive to getting good data from the lens and AF sensor and more prone to calibration issues. If it iterates then yes you can take out the need for the lens data but you also end up in the slower iterative focusing system. It is possible that with old screwdriver type lenses they do not have the capability to support one shot AF and so are driven in an iterative manner. It could also be that your converter forced the camera into a backup focusing mode, or it is possible your camera system at the time could not perform one shot AF. It is possible that this was technology that Minolta/Sony did not have. Also you have to check that it has not entered sweep mode. If the focus is a long way off they can have a mode where they sweep the lens through its focal range till it start to see enough detail to go for a focus calculation.

As ever the camera companies like to keep secrets but have a look into some of the articles where people reverse engineer the systems. I am certain you will agree that if you can know where in the lens range the focusing motor is and can tell how out of focus it is then you have a better chance of moving the lens to the correct point quickly. If you look at the latest Canon lens patents they have items for precisely knowing the rotations of the motor.
Steppenwolf 5 1.2k
14 Sep 2012 9:50AM

Quote:It's actually not called "manufacturing tolerance" in that manufacturers don't tolerate it

You're either being facetious or you don't know much about 'tolerances'..... I don't see any smileys..?

Everything is manufactured to a tolerance...the degree to which an item can deviate from the design norm and still be acceptable. Some items have larger tolerances than others, but even on an optical product, there is still a variation in manufacture. The fact that a 'correction factor' is added to the autofocus computation, to compensate for a measured error, does not mean the the camera/lens leaves the factory absolutely spot-on for focus. That correction factor will also have a tolerance.

The question is really "how much is too much" to fail quality control. The reasons that old Soviet cameras had such varied quality was the agricultural production methods used gave rise to large tolerances, as opposed to the West Germans with their highly skilled and accurate processes. Thus with the Eastern European stuff, if you got two items at the same end of tolerance you got a good product, etc etc.

As Mike alluded to previously, if you have items at the opposite ends of the tolerance scale, there could well be noticeable errors.

Nick



I questioned "Mike's" allusion to "the tolerance scale" in the case of lenses because I wanted him to explain (which he hasn't done, I notice) how a lens can be set up wrongly to cause a camera to deliver wrong focus. Is he saying that the lens elements can be misaligned/mismanufactured to cause wrong focus? This can certainly happen in the case of a lens that has "differential focus" or where the lens elements are misaligned. Or is he alluding to inaccuracies in the lens ROM data? If the lens ROM data is wrong (e.g. it tells the camera the wrong gearing or wrong direction of gearing, etc) then this can cause reproducible FF or BF because, when the camera makes the final correction it will adjust the focus by the wrong amount. (BTW, backlash in focus drive won't produce reproducible FF or BF, just random errors).

"Mike" made an exceedingly glib statement saying "lenses are made within a tolerance so if your lens is at one end of the tolerance range and the body is at the other end blah blah". I'd be interested if he can expand on what that exactly means in the case of a lens.

As for a camera's "tolerances", I don't think that the classic form of DSLR BF and FF is caused by manufacturing tolerances. David Kilpatrick wrote an article about this on photoclubalpha some years ago, but it seems to have gone. The gist of it was that it's too expensive to mass produce consumer DSLRs accurately enough to focus within reasonable tolerances, so they test each camera after production to determine the discrepancy in the light path lengths to the AF sensor and the image sensor. This data is then fed into the firmware and is used as a final correction to achieve accurate focus. Presumably if the camera is hopelessly out of line they reject it. However, provided this is done correctly the focus should be OK assuming that a fully compatible lens is used. The manufacturing "tolerance" here seems to be in the accuracy of the measurement of this data. I guess when people send their camera back for FF/BF they recheck this. The way DSLRs achieve focus is so complicated that there may be other ways for FF/BF to occur, though - I posted a link to another DK article about this on another thread.


Quote:I am certain you will agree that if you can know where in the lens range the focusing motor is and can tell how out of focus it is then you have a better chance of moving the lens to the correct point quickly. If you look at the latest Canon lens patents they have items for precisely knowing the rotations of the motor.


Maybe, but I don't buy your argument that the process of AF is always done by a single movement of the lens focus elements because I can hear the AF motors iterating to achieve correct focus. It's possible that Sony AF using a different approach to Canon, but I doubt it.
LenShepherd 9 3.2k United Kingdom
14 Sep 2012 10:03AM
One of the challenges of using fine tune is using it often produces spurious results if a photographer does not have a first class test target.
This Nikon link and quote from it help clarify why AF is not always good
https://nikoneurope-en.custhelp.com/app/answers/detail/a_id/48334/kw/Auto%20focus
Quote “Cameras AF systems require a difference in contrast to achieve focus. Plain subjects or subjects without much detail or a pattern may mean that the camera is unable to focus accurately.
Targets most likely to cause spurious AF accuracy are in Nikon instruction books or at
https://nikoneurope-en.custhelp.com/app/answers/detail/a_id/4585/kw/Getting%20good%20results%20with%20Auto%20focus
It is extremely rare to find some-one reporting they need fine tune based on a likely 100% reliable for calibration AF target Wink
Unfortunately "garbage in" by way of a less than ideal subject often results in "garbage out" with a false indication bodies and/or lenses working good have a focus issue.
LenShepherd 9 3.2k United Kingdom
14 Sep 2012 10:10AM

Quote:It's actually not called "manufacturing tolerance" in that manufacturers don't tolerate it


There is a lot of truth in this.
See my last reply for some detail - in my experience less than ideal AF targets result in far more poor AF accuracy issues than camera or lens tolerances Smile
31 Jan 2013 1:16PM
The Pentax K-r has AF fine tuning. It should be added to the list.
31 Jan 2013 2:46PM
I have a Pentax k-x. Officially it does not have AF fine tuning (i.e. there is not a menu item for it). However I found on the internet a procedure which allows the user to place the camera in "debug mode" and then access the AF fine tuning.

In short AF fine tuning makes a world of difference and has become a "must have" for me. I know users of the Nikon D5000, and the Nikon D5100 and before doing AF on my Pentax, their camera's were taking more sharply focused pictures then mine. After doing the adjustment on mine, my pictures blow away their pictures in terms of sharpness! I tested their cameras for focusing accuracy and sure enough their cameras are slightly back focused.
Unfortunately there is no AF fine tuning for their cameras so they are stuck with this problem.

There has been a lot of discussion as to the cause and remedies to this AF fine tuning problem. In my experience all the points made above are right and wrong at the same time! AF fine tuning can make a world of difference, but of course it has to be done properly.

I have tested four cameras; my k-x, a d5000 and 2 d5100, with various lenses. The focusing was at least slightly off on all of them. This would lead me to believe that most cameras probably settle for a compromised focus and that they could all benefit from AF fine tuning.

Here is some specific data for your consideration;

1 - Pentax K-x, with a pentax - F 35-70mm lens required an offset of +140
2 - Pentax K-x with a pentax - f 70-210mm lens required an offset of +140
3 - Pentax K-x with a pentax DAL 18-55mm lens required an offset of +60
4 - Pentax K-x with a pentax - A 50mm f1.7 required an offset of +60
All of the above were slightly back focusing trough the whole zoom range but the amount of back focus varied slightly as the zoom setting was changed. The settings above were a compromise which places the "sweet" spot near the middle of the DOF.
5 - The D5000 with the nikon 18-105mm lens was back focusing to the point that one could see it easily by naked eye on a laptop monitor. I tested it and found that it was back focusing more on the lower end of the zoom and much less on the higher end of the zoom. I sent a sample picture to nikon and they asked me to sent it in for adjustment i.e the confirmed the back focus problem.

6 - This one is the weirdest of all. I tested a demo model of the d5100 with the nikon 18-55mm kit lens. This combo showed back focus at 18mm, front focus at 35 mm, and back focus at 55mm!!

Before all you nay sayers jump on me as to my testing methods and accuracy etc. etc . I used the center focus spot only, a shutter speed of at least 180, turned off VR, and in the case of the nikons I used view nx2 to confirm the focal point.

As you can see, from the above, AF problems do exist. They are caused by the camera body as well as the lenses. AF fine tuning can offset the problem in most cases, but there are cases such as item 6 above that fine tuning would not be able to solve.

Hope this helps. I wish there was some way to get Nikon to issue new firmware for the D5000 and the D5100 to allow AF fine tuning.
31 Jan 2013 8:49PM
Pete, how does one update the list of cameras with AF fine tuning? I'd like to add the Pentax K-r to that list.
Thanks
31 Jan 2013 8:53PM
Pentax k-30 also has AF fine tuning.

Sign In

You must be a member to leave a comment.

ePHOTOzine, the web's friendliest photography community.

Join For Free

Upload photos, chat with photographers, win prizes and much more.