Camera and lens calibration

[elsa hoffmann]

I have (to date) always checked my camera and lens combinations myself, but from time to time doubt creeps in about how accurate I am.

I know there are other ways to check and tune - one being FoCAL calibration

Does anyone have any experience with this or any other software/device/method?

http://www.reikan.co.uk/focalweb/

[afx]

I use FoCal. Makes the process much easier and faster.

cheers
afx

[elsa hoffmann]

Afx - which version do you use?

Standard, Plus or Pro?

[Les Olson]

There is a method set out at http://www.northlight-images.co.uk/article_pages/cameras/1ds3_af_micoadjustment.html which I have found excellent - and which is free.  Its advantage is that the interference method makes the point of optimal focus unmistakable, so once you have the camera set up square to the screen you can do a series of lenses very quickly. 

[Andrea B.]

Free is always good. Thanks for the link, Les. Looks like there is a lot of info there.

[bobfriedman]

focal is a tethered system and can automatically adjust the AF.  Unfortunately, this works for Canon only as Nikon hasn't released the SDK to incorporate into the focal software. So it would be manual.

In that case might as well use lensalign pro. In either case you need to take enough measurements at each microfine adjustment to eliminate noise. I routine shoot 4-5 shots at each increment. Hardly statistically significant but allows the lensalign software to find minimum mean square error.

Along with distance (hard to find a room that is 100ft for big glass) it is a time consuming process. It takes me 4-5 hours for combinations cam body/lens/teleconverter.

[elsa hoffmann]

Thanks Bob!

[afx]

Afx - which version do you use?
Standard, Plus or Pro?

Sorry, forgot to check before I left, on the road right now.

focal is a tethered system and can automatically adjust the AF.  Unfortunately, this works for Canon only as Nikon hasn't released the SDK to incorporate into the focal software. So it would be manual.

In that case might as well use lensalign pro

Focal is still faster than a purely manual approach in my eyes.

cheers
afx

[bobfriedman]

my understanding was FOCAL does not have a tethered automated mode for Nikon equipment

[afx]

my understanding was FOCAL does not have a tethered automated mode for Nikon equipment

It can not set the adjustment value of the camera. But it can trigger AF and measure and then will tell you what to set.
So it is slower than fully automatic, but faster than full manual.

cheers
afx

[Bernard Delley]

a bit late to this party, but here I am.
Bob and Elsa touched the core of the matter in my view by saying B:"statistics" and E:"doubt".

Now I wanted to visualize the statistics, which is better done with enough data.  So needed to expedite and relax my procedures.
For today, I decided to use jpg images of the test target and simply use the jpg size to score image detail.
The target needs to be thinner than the DoF and no structured stuff around it should make into the jpg images. In my case I used the MacBookPro screen in an otherwise dark room.

To expedite my procedure I wrote a little program that looks at the bunch of files. It associates the file size with the AF Adj value that gets extracted from the file by invoking  exiftool from within my program. So I can shoot in rapid succession alternating defocus to far and close before AF-ON. I chose to  change the AF fine tune setting after six shots.  The association of file name , Adj and file-size, helps to look at images of interest for further judgement of what might have happened. It turned out that my images with the AFS 80-400mm at 200mm wide open VR=off, sometimes had camera shake. So I redid the series with radio and  MUP. The high up points relate to images showing moire.

Even without fitting it is clear for what settings the sharpest images are made.   It also makes uncomfortably clear how jittery AF is when shooting for top sharpness. 

Similar information is also shown by commercial programs. At least on, forgot which one, uses a flat target and grades the 'sharpness' . I do not think a commercial program could expedite significantly over my procedure. The slowest part is reasonably careful setting up and aligning.  Then there is the slow manual jacking of the AF Adj value from time to time.  Ingesting and analysing the bunch of jpg files is a breeze.

 I see, this site allows gif but not png, which is slightly preferable.

[bobfriedman]

below is actual data from a focustune session- and i might add, a steadier support was needed

[Bernard Delley]

Bob
the statistics of your graph look quite appealing. Made me poke around the internet to find out more about FocusTune.
Do I understand the procedure correctly: Step1, You set up the LenAlign target, which is slightly more tricky than just a basically planar target. Step 2, you shoot in rapid succession jacking the the AF Adjust value from time to time. Step 3, you feed the whole bunch of jpg to the program FocusTune which analyses each image and determines the amount of front and back focus automatically. That task is more challenging than just grading sharpness: bravo for Michael Tapes, its software designer ! The only minor problem I could see is the influence of  field curvature on finding  the  position of the zero line: the front/back focus measurement area is offset from the AF target.

I did not see a specific recommendation for the optimal Adjust value.  In any case its about -2.  Congrats for the lens - camera combo, you have to test very carefully, as you did, to realise such a small potential for improvement.

What you could do, is to reduce the number of  images for the outer values of AF Adj, also by bigger steps. But then gain better info on the zero line crossing by doing more shots in the range of interest. Suggestion: do 10 shots for each Adj value from -4 to +1 for this combo.
As out in the wild you may focus coming in from near, or from far mostly out of your control, you might be interested to find out with the improved statistics if there would be a different Adjust for the two sides of approach.

My feeling is that a contrast maximising procedure, like mine or FoCal, tends to be more disturbed by the statistics of the measurements.

[bobfriedman]

Bernard: i don't have to shoot in rapid succession.. i am not sure really what you are talking about here.. what i do is carefully aim with a supported cam/lens arrangement and take 3-5 shots at each AF correction value... and the variation is still high.. perhaps a laser and bolting the cam/lens to a marble slab would be better.

this case was very close.. i have other arrangements that were not that close. the S/W also provides numeric values to choose the best value. but it is probably better just to draw a straight line through the graph

[Bernard Delley]

Bob,

maybe, I came across overly dramatic by saying rapid succession.  My idea of your procedure is that the camera is mounted with proper aim at the target.  I do not assume any sort of haste. I guess it takes 2 - 3 seconds to go through these steps: a) manually defocus ( I would want to make sure that the AF does not just stay where it was - If I  do not do this, I do not hear/see it budge)  b) AF focus c) expose .