Calibrating autofocus

Had a play with something interesting? Got something that we all covet? Found a real lemon? Write a few lines about it, and share your experiences.

Moderator: Moderators

Forum rules
Please ensure that you have a meaningful location included in your profile. Please refer to the FAQ for details of what "meaningful" is.

Calibrating autofocus

Postby Murray Foote on Thu Jul 05, 2012 3:31 am

There are two kinds of camera autofocus – phase detect and contrast detect. Cameras that use contrast detect (such as mirrorless cameras) do not require adjustment because they focus directly on the sensor. DSLRs that use phase detect have an autofocus sensor that is separate from the main sensor so the relative alignment of the two sensors is critical to autofocus accuracy. It is not peculiar to the camera body, either. Focusing adjustment will differ with different lenses and with zooms, at different focal lengths. Consequently many DSLRs, especially the higher models, have menu adjustments to calibrate the autofocus with each lens.

There are various methods to deal with this. One is to set your camera and lens on a thick piece of wood and administer three sharp taps to each side of the body with a hammer. Some may prefer to put a strip of rubber under the camera first. Then suspend the camera and lens in a watertight container in a solution 50% molasses and 50% cod liver oil. Cover with leaves, bury in six feet of moist soil and leave for eighty years or more. You will then not have any autofocus problems with this camera and lens but you will not be able to take any images either. So buy a new camera and lens.

This leads to the next method. It is well known that if you spend money on a camera in the right way, you become an excellent photographer and your equipment will be better than anyone else’s. There is no need to learn aesthetic sensitivity or technique or to test your equipment or even to take photographs because you will have become such an amazing photographer. And your images, if you take any, will be incredible whether they are in focus or not.

For those of us who do use DSLRs and take photographs, some will be in focus and some won’t. This will apply even if your camera and lens is calibrated, due to user error and shot-to-shot variation. It is good to have your equipment operating as well as it might, but with some methods of calibration you need to take care or you can make things worse than they were to start with. Continuing with the alternatives:

You can purchase the LensAlign system. This in essence gives you a target (placed square to the camera) and a rule (square with the camera but tilted) so you can focus on the target and determine what microfocus adjustment you may need from the rule.

You can also do this yourself at little or no cost, downloading or constructing an appropriate target and rule and using them either inside your house or in your back yard. You need to ensure that both the target and the camera are fixed and will not move or vibrate (including in the wind). You will need a very solid tripod for your camera and the camera and target will need to be precisely level and aligned. You will need to take many shots between each image to account for autofocus variation and manually defocus before each autofocus and shot. The focus on the tilted rule will then tell you what direction you need to adjust in but not how much.

You may see on the web methods using just a tilted rule but they are not recommended because autofocus may not be accurate.

There is another viable do-it-yourself approach using your monitor and a target that induces moire, as explained by Keith Cooper on Northlight Images.

There is also in my opinion a better approach than any of the above (even the molasses and cod liver oil one). I have recently purchased a copy of Reikan FoCal Pro and used that. The difference is that this is an automated or semi-automated process using a tethered camera that removes most of the guesswork. I’m not proposing to provide a review because Martin Bailey has already done that but just to relate a user experience.

FoCal currently supports seven Canon cameras and nine Nikon cameras. Most of the Canon cameras and one of the Nikons work in fully automatic mode. For the rest, you use semi-automatic mode and change the focus calibration setting in your camera’s menu when FoCal tells you to, which is still a lot easier than the manual methods above.

You need a strong lighting source, say natural light or halogen but not fluoro or LED because they flicker. I had some initial problems getting a lock on the target and worked out the main cure was ensuring the lighting was bright enough and moving closer to the target. I made some comments on providing user feedback through the interface when this happens and that will be incorporated in the next edition due soon.

The D3s and D3 columns headed 3 to 6 June are my initial testing results:

Image

Plus or minus 20 is the maximum amount of adjustment for my cameras (and probably for any Canon or Nikon). I discovered that both my D3s and the 85mm f1.4G were hitting that max and sent them back to Nikon for repair. Both of them had previously been in for autofocus repair. I have received the D3s back, repaired under warranty and the results are now much better as shown under the 21 to 22 June column. Nikon still have the 85mm f1.4 and I will have to pay for that as the warranty had expired. I took it to Japan with the D3 and didn’t notice any problems though I was mainly shooting for landscapes and stopping down.

So what this demonstrates is that this program isn’t just good for calibrating autofocus but can indicate when you need to send a body or lens in for repair. This can be particularly useful to check when a lens is repaired or when you get a new lens.

As well as calibrating autofocus, FoCal offers methods to check for your sharpest aperture, find dust on your sensor and check focus accuracy from side sensor points. I haven’t undertaken these as yet. The program is under active development and more features will arrive in time. Low light autofocus consistency is one. I also suggested a method whereby they could check for focus shift and that will be in the release following the forthcoming one.
User avatar
Murray Foote
Senior Member
 
Posts: 1227
Joined: Sun Feb 10, 2008 1:31 pm
Location: Ainslie, Canberra

Re: Calibrating autofocus

Postby Remorhaz on Thu Jul 05, 2012 9:47 am

Thanks Murray - interesting writeup (and interesting to see your results) - I'd heard Martins review before but nice to hear from more locally.

I've only ever used the downloaded target and rule method before (and only bothered changing the micro cal for a couple of my lenses - the 50/1.4 and 70-200/2.8 - the rest were close enough to not bother) but would be interested to see how the FoCal works one day.

I was surprised how big an adjustment you needed to make and also how it varied at different focal lengths for the zooms (can you even save different micro adjusts for the same lens at different focal lengths - I don't think I can do this on my D7000?). I was surprised I even had to adjust for my 70-200 (because I considered it a "pro lens" and figured it wouldn't need it) but I notice you had to adjust many of your pro lenses as well (and a little research online showed that it was quite common to have to adjust the 70-200 - i.e. many reports of people having to send them in to Nikon to be recalibrated).
D600, D7000, Nikon/Sigma/Tamron Lenses, Nikon Flashes, Sirui/Manfrotto/Benro Sticks
Rodney - My Photo Blog
Want: Fast Wide (14|20|24)
User avatar
Remorhaz
Senior Member
 
Posts: 2522
Joined: Thu Apr 29, 2010 8:14 pm
Location: Sydney - Lower North Shore - D600

Re: Calibrating autofocus

Postby Murray Foote on Thu Jul 05, 2012 4:05 pm

Hi Rodney

I tried doing it manually before and gave up because I wasn't convinced my results were accurate enough.

It's common to have different calibration settings for different focal lengths in a zoom. Generally you base it on the telephoto end because that's where depth of field is more critical. The only camera to allow you different settings for each end of a zoom is the new (or forthcoming) Canon 1Dx.

The variation I get on my 17-35mm f2.8 seems extreme though it's obviously less critical with a wide angle lens.
User avatar
Murray Foote
Senior Member
 
Posts: 1227
Joined: Sun Feb 10, 2008 1:31 pm
Location: Ainslie, Canberra

Re: Calibrating autofocus

Postby Mr Darcy on Thu Jul 05, 2012 6:31 pm

Hmm. I may have to go for a drive...
Greg
It's easy to be good... when there is nothing else to do
User avatar
Mr Darcy
Senior Member
 
Posts: 3332
Joined: Thu Oct 26, 2006 11:35 pm
Location: The somewhat singed and blackened Blue Mountains

Re: Calibrating autofocus

Postby biggerry on Thu Jul 05, 2012 9:45 pm

Nice one Murray, i would be interested in this, however i baulked a bit at the 69 euro (i would go teh pro not the standard one), not a huge amount for someone with lotsa gear, but its a bit above my threshold.

I adjusted for just about all my lens and have no doubt it is better than them not being adjusted - but using a focus chart is tedious and there is margin for error in there, which the Reikan FoCal Pro would appear to remove altogether.

I would be interested in Nikons cost for recailbrating teh lens to the body also..
gerry's photography journey
No amount of processing will fix bad composition - trust me i have tried.
User avatar
biggerry
Senior Member
 
Posts: 5512
Joined: Tue May 13, 2008 12:40 am
Location: Under the flight path, Newtown, Sydney

Re: Calibrating autofocus

Postby Murray Foote on Thu Jul 05, 2012 10:57 pm

There's a discount link at the bottom of Martin's review which makes it $A68, which is still active.

I didn't ask Nikon to recalibrate the 85mm lens to the body, just to fix them both. I sent them both at the same time but the lens turned up some days later. The body I don't have to pay for but the lens will cost me about $400, which I wouldn't have had to pay if I had had FoCal a year earlier (though it wasn't available for Nikon then, if at all).
User avatar
Murray Foote
Senior Member
 
Posts: 1227
Joined: Sun Feb 10, 2008 1:31 pm
Location: Ainslie, Canberra

Calibrating autofocus Take 2

Postby Murray Foote on Sun Aug 05, 2012 5:34 am

When you run an autofocus microadjustment test in Reikan Focal, you get a multi-page report including a summary graph like this:

Image

=======================================================================================================================================

I got my D3s and 85mm f1.4G back from Nikon after repair and these were the test results:

Image

The 85mm f1.4G appears little changed after repair - the adjustment value remains at 20 or greater for both cameras.

The D3s has improved but still does not appear satisfactory. Apart from the 85mm f1.4G, one of the 180mm f2.8s is maxing out while the 50mm f1.4G and the 300mm f4 have very high adjustment values. Leaving aside the 85mm f1.4G, the average adjustment value for the primes is 3.8 with the D3 and 12.0 with the D3s.

I'm sending them both back to Nikon and suggested they adjust the autofocus for the D3s by a relative amount of -10 and repair the lens again.

=======================================================================================================================================

There are other things you can test with FoCal. I present some summary captures below but more information is available in each report.

You can test autofocus consistency:

Image

It's a bit hard to know what is normal for a test like this but you should at least be able to confirm if your lens autofocus is seriously inconsistent.

=======================================================================================================================================

You can test for multipoint autofocus accuracy. These captures are for my 105mm f2 on the D3s, though the actual findings are irrelevant. I didn't realise when I posted this that the test failed to complete and there are only a few data points for each sensor cell. So this is an example of the format of summary output the test can provide rather than an example of the sort of results you might get.

Image

This requires some interpretation. The numbers are the indicated microfocus adjustment and they are black if positive and red if negative. The background colours go from dark green to light green to yellow to light red to dark red in decreasing order of confidence for statistical significance.

Image

The second chart shows quality of focus which ostensibly looks OK (or would if the results were significant).

I tested using 9 exposures for each cell on the sensor. With 51 sensor cells that gives 459 exposures for a test. So for each cell in the sensor array, there would be exposures at microfocus adjustments of -20, -15, -10, -5, 0, 5, 10, 15 and 20. That would take quite a while by itself. Problem is that the lens defocuses to infinity and then focuses back for each exposure. That can make this test quite gruelling. The test lets you specify how many sensor cells you want to measure and how many microadjustment values for each sensor cell. Nine exposures for a cell is not enough for accuracy but you also get graphs for each sensor cell which makes better assessment possible. I might be better, though, going for fewer sensor cells and more exposures.

Potentially this is a useful test that for example could pick up inaccurate side sensors in the D800. That's not to say all values should be uniform because field curvature in a lens could conceivably affect side values as well. This is a time-consuming test, though, and requires some experimentation to find the best settings.

=======================================================================================================================================

There's also a test for dust on the sensor which is the easiest way I've seen to do this. The D3s came out as nada, zip, nil, zero, not a sausage, clean as one of those things that you blow to make some noise. The D3 came out with some evidence of dust. No surprise there, I took nearly 9,000 shots with it in Japan in February and I've been processing the images and posting to my blog since the beginning of March and am now over 80% complete - I have encountered the odd indications of dust on some images during the processing.

Image

Image

The first chart shows how different types of dust can potentially have effect at different apertures. The second is an example of dust distribution at f11.

=======================================================================================================================================

Lastly, you can also test sharpness by aperture.

Image

Image

The interesting thing here is that I happen to have two 180mm f2.8D IF-EDs, which is what you are seeing above. One is 16 years old (top chart) and the other is less than a year old but they are the same model lens. The first lens is generally softer at wider apertures, peaks at smaller apertures and falls off sooner when stopped down. The second lens is generally better, especially wider apertures, apart from f2.8. I would not have thought they would show up so differently.

There are two possibilities here: Nikon could have modified the lens at some point and not told anyone or this is just natural variation in the same lens. I can find three review that show resolution by aperture (two in Photozone and one in SLR Gear) and all seem to me to look more like the second lens. I conclude that this is therefore likely to be due to variation in the two lenses, which makes such testing more significant to determin how your particular lenses perform.
Last edited by Murray Foote on Thu Aug 09, 2012 1:29 am, edited 1 time in total.
User avatar
Murray Foote
Senior Member
 
Posts: 1227
Joined: Sun Feb 10, 2008 1:31 pm
Location: Ainslie, Canberra

Re: Calibrating autofocus

Postby zafra52 on Sun Aug 05, 2012 10:52 am

Umm, I think the first option was easier
for the old brain.
User avatar
zafra52
Senior Member
 
Posts: 3231
Joined: Thu Dec 01, 2005 10:22 pm
Location: Brisbane

Re: Calibrating autofocus

Postby Murray Foote on Sun Aug 05, 2012 5:21 pm

zafra52 wrote:Umm, I think the first option was easier
for the old brain.

This one you mean?
I wrote:...Set your camera and lens on a thick piece of wood and administer three sharp taps to each side of the body with a hammer. Some may prefer to put a strip of rubber under the camera first. Then suspend the camera and lens in a watertight container in a solution 50% molasses and 50% cod liver oil. Cover with leaves, bury in six feet of moist soil and leave for eighty years or more.

Feel free if it works for you. Let us know how you go.
User avatar
Murray Foote
Senior Member
 
Posts: 1227
Joined: Sun Feb 10, 2008 1:31 pm
Location: Ainslie, Canberra

Re: Calibrating autofocus

Postby zafra52 on Sun Aug 05, 2012 5:38 pm

The very one, but not just yet. :D
User avatar
zafra52
Senior Member
 
Posts: 3231
Joined: Thu Dec 01, 2005 10:22 pm
Location: Brisbane

Re: Calibrating autofocus

Postby Murray Foote on Thu Aug 09, 2012 1:46 am

I have modified my previous post in this thread Calibrating autofocus Take 2. I have extensively modified my comments for the multipoint autofocus test (though the results in this case are not significant). I have also added some comments to the Sharpness by Aperture test.

In summary:
    Autofocus microadjustment test: Very useful, much better than manual and semi-manual methods
    Autofocus accuracy test: Quick and probably useful to identify a serious problem
    Multipoint autofocus accuracy: Time consuming and puzzling but addresses a real potential problem known particularly for the D800
    Sensor dust Test: Quick and informative
    Sharpness by Aperture Test: Quick and useful.

Comments?
User avatar
Murray Foote
Senior Member
 
Posts: 1227
Joined: Sun Feb 10, 2008 1:31 pm
Location: Ainslie, Canberra

Re: Calibrating autofocus

Postby ATJ on Mon Aug 13, 2012 5:00 pm

I purchased the software yesterday and wasted around 2 hours today and didn't get very far at all.

I got very inconsistent results with my Micro 60mm f/2.8D to the point that FoCal gives up on it. That was with both my D300 and D7000. I managed to complete the analysis for the 18-55mm f/3.5-5.6G VR on the D300 and the 70-300mm f/4.5-5.6G VR @ 200m on the D7000. The 70-300mm gave inconsistent results on the D300.

At one point the app kept crashing and I had to reboot to be able to complete an analysis.

At this point I think I'm regretting spending the money as I have no confidence in the results.
User avatar
ATJ
Senior Member
 
Posts: 3982
Joined: Fri Feb 18, 2005 10:44 am
Location: Blue Mountains, NSW

Re: Calibrating autofocus

Postby Murray Foote on Mon Aug 13, 2012 6:46 pm

Well, you can always ask for a refund. FoCal has definitely been useful for me to identify my D3s and 85mm f1.4G needed repair. Most of my results are fairly consistent though I may have an advantage in that I am mainly using primes and my zooms are ultrawides for which profiling is less essential.

One thing I noticed with the resolution by aperture test is that at one point my two 180mm lenses had very different results in image quality level. I quickly realised that was due to alignment to the target (I moved the tripod between tests). FoCal still performs the test when slightly tilted with respect to the camera. So to ensure correct alignment I measured the distance from the floor to the centre of the lens and set the centre of the target at that height. That should eliminate vertical tilt. The target was on a holder above a bench and I put a book hard up against against the holder and visually aligned the camera and tripod against the side edges of the book. That should have eliminated side tilt.

The quality of the results is going to be most reliable where you have the best quality target image. This should be well defined and high contrast but sometimes is pale and low contrast. I found the solution was generally to move the camera closer to the target. Rich of FoCal suggested using exposure compensation but I found this didn't make much difference. Even if you get good quality target iages, the constraint is that you are generating compensation values at the subject distance you will be using the lens. I must admit I haven't checked this yet for my equipment. Roger Clark has an interesting article on calibrating microadjustment in situ.

Testing is time-consuming. I've done 129 microadjustment tests so far though I have the advantage with the D3s that it's fully automatic and only 26 of those are with the D3 where I need to change the microadjustment values when indicated (as for the D300 and D7000). You need to look at the graphs as well as the indicated value. Autofocus is always variable so sometimes abberant results might mislead the graphs. Quite often the graphs are fairly flat and the value the algorithm picks may not be the most appropriate one. In these cases you can get somewhat different predicted values even though the graphs are very similar. In any case, a difference in adjustment amount of a few points will not make much difference.

Another possible issue is autofocus consistency. Perhaps this might be an issue for the 60mm micro because it is a non-AFS lens on a consumer body. It might be an idea to check your lenses with the autofocus consistency test. For what it's worth (WNTBA?), here are my autofocus accuracy results:

Image

The red figure for the 85mm f1.4G with the D3s (an average of three quite different results) is where both lens and camera needed repair. I assume all other figures are "normal".

FoCal does crash from time to time and I've found the multipoint test particularly exacting for needing to often adjust the lens focus. However, I haven't had much trouble with the microadjustment test since I worked out how to align the target and establish better camera-to-target distances. I don't recall needing to reboot the computer; usually you just need to close FoCal down and reopen it and sometimes just go back to the initial camera selection screen.
Last edited by Murray Foote on Mon Oct 22, 2012 11:03 pm, edited 1 time in total.
User avatar
Murray Foote
Senior Member
 
Posts: 1227
Joined: Sun Feb 10, 2008 1:31 pm
Location: Ainslie, Canberra

Re: Calibrating autofocus

Postby Murray Foote on Thu Oct 04, 2012 2:43 am

I got the camera and lens back from readjustment again (actually I've had it for a while).

Image

I had asked for the D3s to be adjusted by a net amount of -10 but it doesn't seem much changed. The 85mm f1.4 is now within adjustment range on both cameras although very high on both and 19 on the D3s. Still, it seems to work at that and focuses quite accurately at that setting. It also doesn't seem to show any evidence of focus shift, which can be an issue with fast lenses.

So I think I'll leave it at that. The autofocus adjustment values are rather high for the 85mm but at least everything works OK.

I found an interesting article by Falk Lumo recently. (I include the link but it's probably only worth reading if you have FoCal). There are two interesting points here. One is that Nikon AF adjustment values are much more sensitive than Canon's - so that for an equivalent amount of adjustment on lenses with same maximum aperture, the Nikon value is twice or more higher than the Canon. Also, the amount of adjustment may vary according to the maximum aperture of your lens: "I currently believe that AFMA values determined for an f/1.4 lens are twice the values as determined for an f/2.8 lens".
User avatar
Murray Foote
Senior Member
 
Posts: 1227
Joined: Sun Feb 10, 2008 1:31 pm
Location: Ainslie, Canberra

Re: Calibrating autofocus

Postby Murray Foote on Tue Oct 16, 2012 4:21 pm

Reading more on the Lumo article I cited in the previous post, I realised that the article implies maximum calibration settings as follows:

Image

For example, the 10 for Nikon FF at f2.8 implies that a lens calibration is satisfactory if the value is between plus and minus 10. Otherwise, you should send it off for adjustment.

Figures are likely to be accurate for Nikon Full Frame; other columns are more speculative due to insufficient data. This also implies that there is no bias due to the camera which you can only determine by checking other lenses and multiple cameras.
User avatar
Murray Foote
Senior Member
 
Posts: 1227
Joined: Sun Feb 10, 2008 1:31 pm
Location: Ainslie, Canberra

Re: Calibrating autofocus

Postby Murray Foote on Sat Oct 20, 2012 6:15 pm

MIchael Tapes, who produces the manual system LensAlign is about to announce a new computer-based system FocusTune, initially available for $US30.

This is not a tethered system, so you blat away with your camera and then download the card onto your PC. I expect to get it to see what it can do and to compare the results with FoCal. However, not being tethered is likely to be a mixed blessing. It means you'll have to change autofocus adjustment settings manually and also manually defocus for each shot. Inherently you need to take a fair few shots to rule out autofocus variation so this might get tedious.

Conversely, I understand that FoCal are working on a method to bypass the SDK and fully automate testing for all Nikon cameras. They expect to have this ready by the end of 2012 or the beginning of 2013.
User avatar
Murray Foote
Senior Member
 
Posts: 1227
Joined: Sun Feb 10, 2008 1:31 pm
Location: Ainslie, Canberra

Re: Calibrating autofocus

Postby Remorhaz on Sat Oct 20, 2012 8:11 pm

Thanks for the update Murray.

I bought Focal Pro yesterday (he indicated he'd have D600 support working in about a week) - I havn't installed or used it yet - might try/practice with the D7000 next week sometime.

Any gotcha tips for using this (NB: with D7000 and D600 so for now no fully automatic mode)?

I noticed in a post somewhere the suggestion for printing a bunch or targets and then laying them out in a large grid (maybe 6 x 4 targets or something like that) to make a huge target
D600, D7000, Nikon/Sigma/Tamron Lenses, Nikon Flashes, Sirui/Manfrotto/Benro Sticks
Rodney - My Photo Blog
Want: Fast Wide (14|20|24)
User avatar
Remorhaz
Senior Member
 
Posts: 2522
Joined: Thu Apr 29, 2010 8:14 pm
Location: Sydney - Lower North Shore - D600

Re: Calibrating autofocus

Postby Murray Foote on Sun Oct 21, 2012 12:38 am

Hi Rodney.

I am fortunate to be able to use automatic mode on the D3s but still had to use semi-automatic mode on the D3. That's still quite viable though it does slow you down a bit.

Not much in the way of gotchas. Good light levels help. I went for artificial light which means halogens. Fluoros and LCDs flicker and probably won't work.

I had some problems early on (failure to validate target or crashing) that I solved by increasing light levels or moving closer to the target. I think they were problems of earlier versions, though.

Correct alignment is likely to be important. I measured the centre of the target from the floor with a tape measure and set the camera on the tripod so the centre of the lens was at that height. I then set a book in front of the target and sighted along it to ensure the camera was square on. Putting a small mirror on the front of the target to check the alignment might be another way.

I haven't tried printing a bunch of targets, I just use one at A4 and one at A3+. If I recall correctly, that suggestion was for the Multi Point Focus test rather than the Semi Automatic AF Microadjustment. You probably won't get to that one for a while.

If you do get to the Multi Point Focus test, though, be careful not to try for too much. I specified such an exhaustive test at one point that I ran out of battery, even with the D3s battery and the automated test, which meant that after several hours I was left with nothing. A good place to start may be Falk Lumo's suggestion that I forgot to mention earlier: Select five points. Full range from -20 to +20. 6 AFMA points. Validate focus consistency to 2%.
User avatar
Murray Foote
Senior Member
 
Posts: 1227
Joined: Sun Feb 10, 2008 1:31 pm
Location: Ainslie, Canberra

Re: Calibrating autofocus

Postby Remorhaz on Mon Oct 22, 2012 6:20 pm

Good news - the latest beta was released and has both D600 and MacOSX (10.6/10.7 & 10.8) support - it installs and runs so when I get a chance for a good bit of time (perhaps friday or this weekend) I'll give it a shot.
D600, D7000, Nikon/Sigma/Tamron Lenses, Nikon Flashes, Sirui/Manfrotto/Benro Sticks
Rodney - My Photo Blog
Want: Fast Wide (14|20|24)
User avatar
Remorhaz
Senior Member
 
Posts: 2522
Joined: Thu Apr 29, 2010 8:14 pm
Location: Sydney - Lower North Shore - D600

Re: Calibrating autofocus

Postby Murray Foote on Mon Oct 22, 2012 6:39 pm

I'll be interested to hear how you go. It's always worth checking when you come to use it because they have functional updates quite often.
User avatar
Murray Foote
Senior Member
 
Posts: 1227
Joined: Sun Feb 10, 2008 1:31 pm
Location: Ainslie, Canberra

Re: Calibrating autofocus

Postby Remorhaz on Mon Oct 22, 2012 9:26 pm

I just did a quick run through (auto testing only (with manual AF tune adjustments on camera)) with three lenses (50/1.4, 85/1.8 and the 70-200/2.8 ) on my D600 and I feel like I got results that gelled with what I was expecting.

Notes:
- I had manually AF tuned the 50/1.4 and 70-200/2.8 on my D7000 some time ago and they both required quite a bit of negative AF tune
- I did it in what I thought was pretty crappy indoor light (night time) - I actually wasn't expecting it to work - the room has some LED MR16 globes on the ceiling with one spreading onto the wall where I mounted the A4 target - but it seemed to be enough to run the tests
- my targets were laser printed (Reikan actually suggests not to use laser but to use inkjet but hey... :) )
- the new Mac software ran quickly and smoothly - no issues
- target setup/alignment using the software was relatively easy
- the overall process was actually easier than I was expecting (even if I did have to manually set the AF adjustments on the camera when prompted - but I knew that)

Results:
- the Sigma 50/1.4 @ 3.2m needed -11 (result confidence of Excellent) after 22 shots
- the Nikon 85/1.8G @ 2.9m needed -6 (result confidence of Good) after 30 shots
- the Nikon 70-200/2.8 @ 200mm @ 4.2m needed -10 (result confidence of Excellent) after 21 shots
- the Nikon 70-200/2.8 @ 135mm @ 4.1m needed -17 (result confidence of Excellent) after 23 shots
- the Nikon 70-200/2.8 @ 70mm @ 4.1m needed -25! (not possible to set) (result confidence of Good?) after 32 shots

I'll do a few more lenses and I'll redo the 70-200 in better light (daylight) another time - I'll try to do the long lenses (70-200, 70-300 & 50-500) from further away as well (recommendation seems to be 50x focal length (so at least 10m))
D600, D7000, Nikon/Sigma/Tamron Lenses, Nikon Flashes, Sirui/Manfrotto/Benro Sticks
Rodney - My Photo Blog
Want: Fast Wide (14|20|24)
User avatar
Remorhaz
Senior Member
 
Posts: 2522
Joined: Thu Apr 29, 2010 8:14 pm
Location: Sydney - Lower North Shore - D600

Re: Calibrating autofocus

Postby Murray Foote on Mon Oct 22, 2012 11:11 pm

Sounds like your D7000 might be at something like -10. The next thing is probably to go out in the garden (or whatever your equivalent may be if you don't have one) and take some shots at the recommended settings to confim that those settings work for you. And with the 70-200mm if further tesing confirms those settings, you're going to have to decide whether to go with the long telephoto setting of -10 or a compromise such as -15. What the negative values that increase from 200mm to 70mm mean is increasing back-focus.

I haven't greatly worried about the 50x rule though I would have been close with at least the 105mm and the 85mm. It's definitely something worth checking though I haven't done so as yet.
User avatar
Murray Foote
Senior Member
 
Posts: 1227
Joined: Sun Feb 10, 2008 1:31 pm
Location: Ainslie, Canberra

Re: Calibrating autofocus

Postby ATJ on Tue Oct 23, 2012 7:40 am

It is interesting that they recommend against LED lighting for the testing. I (and most other UW photographers) use LED focusing lights when photographing underwater. I've never had a problem and have not heard of others having issues.

Edit: Ah... it is likely that only AC powered LEDs might flicker (if they have poor power supplies). Battery operated ones shouldn't be a problem.
User avatar
ATJ
Senior Member
 
Posts: 3982
Joined: Fri Feb 18, 2005 10:44 am
Location: Blue Mountains, NSW

Re: Calibrating autofocus

Postby Remorhaz on Tue Oct 23, 2012 10:59 am

Murray Foote wrote:Sounds like your D7000 might be at something like -10


this was my D600 I did but yes I believe both my D7000 and D600 are similar - I'll test both with the same lenses to see what pans out eventually.

I believe I know why the dim lighting "worked" - I'm pretty sure I had AutoISO on so I expect the tests weren't done at ISO 100 at all but rather something much higher - with the 70-200 it was probably well over ISO 2000 I'm guessing since it would have been a relatively slower lens (2.8 vs 1.4 or 1.8) and a faster shutter (1/200th rather than 1/50th). When I test in good light I'll see if there's much of a difference (see if using ISO 100 as Rich recommends really does matter :)).
D600, D7000, Nikon/Sigma/Tamron Lenses, Nikon Flashes, Sirui/Manfrotto/Benro Sticks
Rodney - My Photo Blog
Want: Fast Wide (14|20|24)
User avatar
Remorhaz
Senior Member
 
Posts: 2522
Joined: Thu Apr 29, 2010 8:14 pm
Location: Sydney - Lower North Shore - D600

Re: Calibrating autofocus

Postby Murray Foote on Tue Oct 23, 2012 2:43 pm

I've never tried a higher ISO. Obviously the resolution of the image captures would be less and the curves would probably be flatter and less accurate. What practical difference that would make is hard to say. In those terms, Auto ISO may work OK but it would be dangerous to use it outside where the light level might be changing because where that triggered ISO changes it might deliver spurious changes in resolution readings.
User avatar
Murray Foote
Senior Member
 
Posts: 1227
Joined: Sun Feb 10, 2008 1:31 pm
Location: Ainslie, Canberra

Re: Calibrating autofocus

Postby Murray Foote on Thu Oct 25, 2012 4:59 pm

Michael Tapes has now released FocusTune. I've had a look at the documentation and downloaded it but not used it as yet.

Its main feature is equivalent to manual mode in FoCal Plus and Focal Pro. You take test shots on a camera in a specified way and then import them into the software which gives you a microadjustment curve and a recommended microadjustment value.

It also has Autofocus Consistency and Aperture Comparison tests which I presume are equivalent to AF Consistency and Aperture Sharpness tests in Focal Pro.

FocusTune lacks the Semiautomatic Microadjustment of all versions of Focal, the Fully Automatic Microadjustment of Focal Plus and Focal Pro, the Sensor Dust Analysis of Focal Pro (which is useful but perhaps incidental) and the Multi Point Focus test of FoCal Pro (which tests for sensor misalignment, important especially for D800 owners). The Focal Semiautomatic and Fully Automatic Microadjustment tests require a tethered camera but they don't require that you defocus the camera for each shot and actually take the shots, a considerable reduction in tedium and likely to produce more accuracy because you are likely to take more shots and less likely to bump the camera.

I purchased FocusTune because it gives me an independent check on the values I obtained with Focal. Mind you, casual testing indicates that those values are OK so I don't really expect any significant discrepancies.

FocusTune is a useful way of improving your autofocus accuracy, perhaps especially if you are already a LensAlign user. Pricewise, it's equivalent to FoCal Standard. I think I would prefer Focal Standard in that comparison due to ease of use in its Semiautomatic Microadjustment test, even though it requires a tethered camera and does not have the AF Consistency and Aperture Sharpness tests. This is debatable. Other views may differ. However, Focal Pro is clearly much more powerful and will remain my preferred choice. Both products are likely to improve and add new features so there might be a case for revising this opinion at some time in the future.

I haven't actually used FocusTune as yet and probably won't find the time until some time next week.
User avatar
Murray Foote
Senior Member
 
Posts: 1227
Joined: Sun Feb 10, 2008 1:31 pm
Location: Ainslie, Canberra

Re: Calibrating autofocus

Postby Remorhaz on Thu Oct 25, 2012 8:00 pm

Murray Foote wrote:Michael Tapes has now released FocusTune.
FocusTune lacks the Semiautomatic Microadjustment of all versions of Focal, the Fully Automatic Microadjustment of Focal Plus and Focal Pro


Murray I reckon one other big win for the semi or fully auto tethered mode of Focal Plus/Pro on top of it controlling your camera and lens defocusing, etc is that instead of having to take a shot at every AF tune point, because the software gets data continuously from each shot it zeroes in on the AF tune value quicker and skips you having to do a full set. e.g. instead of manually shooting -20 to +20 in all 1 step increments offline and then feeding it into FocusTune - with Focal Pro it does 0, -20, -10, +10 and then +20 and by that stage it has estimated approx where it thinks the tune value is because of the overall curve it has so far and then just tells you to go to a few points around where it thinks it is going to be to fine tune the most accurate value - e.g. -13, -14, -15, -12 and then tells you -13 is the one.

FYI I tried the sensor dust one with my D600 and it was easy and worked nicely - showed me that I shouldn't see dust spots at all up to f/14 (at least with the lens I was using) and then I had a few at f/16 and it was reasonably consistent with a little growth in number up to f/25 but then at f/32 there were 7 visible :)
D600, D7000, Nikon/Sigma/Tamron Lenses, Nikon Flashes, Sirui/Manfrotto/Benro Sticks
Rodney - My Photo Blog
Want: Fast Wide (14|20|24)
User avatar
Remorhaz
Senior Member
 
Posts: 2522
Joined: Thu Apr 29, 2010 8:14 pm
Location: Sydney - Lower North Shore - D600

Re: Calibrating autofocus

Postby Murray Foote on Fri Oct 26, 2012 1:46 am

Well, you can take whatever shots you like with FocusTune, including a small range or various intervals, and then get the software to process them. This is why Michael Tapes' suggestion is to use it in conjunction with LensAlign, testing a small range around the indicated value. But I wonder about testing a small range in that way because the curve can easily be misleading. I also think it's good to run these tests multiple times to make sure you're not getting a spurious statistical result and you're much more likely to do that with FoCal since it's much more automated.

Yes the dust test surprised me. It's the easiest and most graphic method I've seen. I bet you didn't get anywhere near the amount of dust I reported on my D3 earlier in this thread. My D3s, though, with a self-vibrating sensor, showed up as entirely clean.
User avatar
Murray Foote
Senior Member
 
Posts: 1227
Joined: Sun Feb 10, 2008 1:31 pm
Location: Ainslie, Canberra

Re: Calibrating autofocus

Postby Remorhaz on Fri Oct 26, 2012 6:59 pm

OK I retested in appropriate light - I think we can pretty much ignore my previous results since they were shot in crap light indoors at very high ISO's (1000-4000+)

I tested 4 lenses today at various focal lengths and distances (outdoors under cover) and with good light used ISO 200 or 100 depending on what sort of shutter speed I'd get. I also used the larger A3 target for all these. I also turned on the Optimise Target setting (and adjusted the exact target size value - presumably so the distance calc would be more accurate?).

Nikon 70-200/2.8 VR on D600 @ ISO200
70mm -10 Good 4.5m
80mm -12 Good 4.5m
105mm -10 Good 4.5m
135mm -11 Excellent 4.5m
160mm -7 Good 4.5m
200mm -6 Good 4.5m

You can see the 70-200 got much more consistent results across the focal range this time :)

Sigma 50/1.4 on D600 @ ISO100
50mm 6 Good 1.5m
50mm 7 Excellent 2.5m
50mm 0 Good 3.9m
50mm 0 Acceptable 4.1m

I actually did the 50/1.4 from the bottom up (distances) so I was happy with the 0 results I was getting at the start - it looks like as you get closer to the MFD (note this is well under the recommended 50x focal length for testing and Reikan does say things get unreliable if you get too close) the AF tune appears to get further off

Nikon 85/1.8G on D600 @ ISO100
85mm 11 Excellent 1.3m
85mm 3 Good 1.9m
85mm 3 Good 2.9m
85mm 1 Excellent 4.6m

Similar results for the 85/1.8G - getting closer to MFD gets wackier

The following is Chris's lens (he's very kindly lent it to me for a couple weeks to try it out before I decide if I want a 24-120, 24-70 or no midrange zoom (UWA + primes + telezooms) - only did two quick tests at each end at different distances

Nikon 24-120/4 VR G on D600 @ ISO200
24mm -1 Excellent 2.0m
120mm 6 Excellent 4.7m

I also ran a few Aperture Sharpness Graphs for some lenses - Chris here is one for your 24-120 (@120mm) - you can see it's very good from about f/4.5 to f/13

Image
D600, D7000, Nikon/Sigma/Tamron Lenses, Nikon Flashes, Sirui/Manfrotto/Benro Sticks
Rodney - My Photo Blog
Want: Fast Wide (14|20|24)
User avatar
Remorhaz
Senior Member
 
Posts: 2522
Joined: Thu Apr 29, 2010 8:14 pm
Location: Sydney - Lower North Shore - D600

Re: Calibrating autofocus

Postby Murray Foote on Mon Oct 29, 2012 5:08 pm

I think that especially when you get results that FoCal does not describe as "Excellent", you need to also look at the graph. Often there can be large variations and you can see that the line of the graph could be drawn in a different way.

p.s. I'm not sure now whether I'll be testing FocusTune any time soon. I just came back from the Sydney Blues Festival and have a huge number of images to process.
User avatar
Murray Foote
Senior Member
 
Posts: 1227
Joined: Sun Feb 10, 2008 1:31 pm
Location: Ainslie, Canberra


Return to Equipment Reviews

Who is online

Users browsing this forum: Google Adsense [Bot] and 2 guests