What is your Micro Adjustment procedure?

[RC]

I know there are a lot of MA procedures out there so I wanted to share with everyone what I am doing and hoping I can improve on it based on what others learned and experienced. I fiddled with a few procedures several months ago but it wasn’t until now that I had to actually make an adjustment. My new 24-105 required an adjustment of +9 which seems like an awful lot! (I bought this lens for Christmas, my first copy had what appeared to be an air bubble in one of the elements, and now my replacement is way off focus. Seems like Canon QC took a dive this year with all the disasters, but that’s another topic for another time.)

I tried 3 other procedures before I found this one:

http://www.dgrin.com/showthread.php?t=116078
Credit goes to Scott Quier, thanks for sharing Scott if you are out there!


For the most part I follow Scott’s procedure with a couple of adjustments. I use a basketball for the subject on the first pass instead of a tree trunk and I use a chart with lines and text for the final pass.

I’ll summarize the procedure here:

1. Manual exposure, RAW, widest f stop, locked down on tripod using RC6 remote. Level camera with7D’s leveler, lens parallel (square) with target
2. Take 9 shots beginning with -20 MA in 5 step increments thru +20. I use a basketball as my subject and focus on specific point.
3. Remove camera (or flash card) without moving tripod
4. Download to PC, leave RAW as is, use DPP to export to jpg (this allows rapid cycling to find sharpest image)
5. Use ACDSee (cause it is easier) to crop center of images (same crop values for all images)
6. Using ACDSee, cycle thru images to find sharpest one.
7. When I find sharpest image, I identify what the MA value is.
8. Using the value in step 7, I take 7 more shots adding 3 to each side. For example if my sharpest image in step 7 was +10, then I will take test shots +7 thru +13 in one step increments. I use a test chart as my subject
9. Repeat steps 3 - 7 to find my sharpest image.

Why a basketball for the first pass? I focus about 45 degrees off the face of the ball on specific point like the tip of a letter. This allows me to see any front and back focus. Also, a BB has great texture, detail, and lettering making it easy to focus your eyes on the tiniest detail

Why a flat test chart for the second pass? Now that I am very close, I don’t want any curvature to throw me off. I want to focus on fine lines and text

Questions I have
1. What is the best distance to perform this test? I did my test at 16’
2. What focal range is best on a zoom lens? I did my test at 105mm (on crop body BTW). Is it best to test at mid zoom?
3 Best f stop wide open or the optimum aperture for that lens +/- f8?

Thanks for the feedback!

 

[Mt Spokane Photography]

The process outlied may work, but then again, there may be a problem. I've also tried lots of adjustment procedures.

The most critical thing is to know absolutely what your camera focused on, and to be absolutely level and parallel to the target. The focus point you see thru the viewfinder is smaller than the actual focus point, so if it lights up, the focus point could be substantially out of the area covered by the red box.

The culpret is the camera AF system. It can focus anywhere over a fairly large center area, and relies on detail and contrast as well as to determine just where it can focus. The AF system will focus on a horizontal line if it has a choice.

And, unless you focus on a flat surface very accurately aligned to the camera sensor, you may actually focus on a point that is not where you think it is. Tools like Lens align were developed with the input from many photographers who had a lot of insight into how the AF system works, and yields a much better result with less variation than many of the do it yourself methods I've used.

Of the DIY systems out there, Jeffrey Freidel has a good one. He is also one of those who gave a lot of help to the Lensalign design.

http://regex.info/blog/photo-tech/focus-chart

Good Luck

Here is a shot of my Lens align system. You adjust the camera up and down, and back and forth until the white circle is in the center (its res around the edges.) Even a inch off makes a obvious difference, but once adjusted, you are parallel to the flat target and your camera AF will lock onto it. Then, the slant ruler on the right can be used to adjust forward or rear AF. Distance used varies from 25X to 50X the lens focal length. I check zooms at two or three focal lengths, there will be a slight difference, but if its not objectionable, your ok.

The first image is using liveview and contrast detect, so AF is right-on. The second image is one of several I took after adjustment. about one of 4 will have a very slight difference.

 

[Kahuna]

That is an absolutely beautiful system. Had to check it out on your website. I will have to purchase one soon.

 

[neuroanatomist]

Here are my tips:

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

 

[epsiloneri]

I use the procedure outlined in this thread. It is admittedly somewhat intricate, but it gives great quantifiable results that don't depend so much on your ability to visually decide what MA gives the best results. The quicker way is to do as neuroanatomist suggests in his very nice write-up above.

 

[briansquibb]

Here are my tips:

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

+1

I have been lucky so far - only my 50D has needed micro adjustment

 

[Mt Spokane Photography]

I've placed a pre-order for the Reikan FoCal Pro automatic adjustment software. so I will be comparing it to the lens align system. If it works well, life for those who use MA could get a lot simpler.

If I understand the hype correctly, it will even go beyond the +/- 20 limitations now in the camera. If it does what I need, my lens align may be up for sale. There is a pre-order discount, and its also a risk that it might be all that wonderful, but several beta users have liked it, I just do not know their knowledge level of micro adjustment.

Eventually, Canon may offer automatic calibration as a feature, but more likely a $600 option.

http://www.reikan.co.uk/focal/focal.html

 

[Edwin Herdman]

I tried the screen pattern method (the Northlight Studios one) and found it impossible to judge.

Right now I'm planning on trying out this one:
http://photography-on-the.net/forum/showthread.php?t=750736 Free version!
http://photography-on-the.net/forum/showthread.php?t=1066028 $14.99 version!

It's essentially a lens align.

 

[Mt Spokane Photography]

I tried the screen pattern method (the Northlight Studios one) and found it impossible to judge.

Right now I'm planning on trying out this one:
http://photography-on-the.net/forum/showthread.php?t=750736 Free version!
http://photography-on-the.net/forum/showthread.php?t=1066028 $14.99 version!

It's essentially a lens align.

It is lacking the essential point of the lens align, and thats to have the target very accurately parallel to the camera sensor. You will also need some very accurate construction to get the angles perfect.

Let us know how it works out, it should at least improve things.

 

[Edwin Herdman]

I've read all those "not perfectly aligned" arguments before and I'm not concerned by them. A good 90 degree wall corner will take care of the problem if you're really concerned, though; otherwise, a slight sideways offset of the target won't prevent the slope from matching up with the flat target. Just looking at the device, I don't see how any but a rather severe degree of convergence between the sensor and parallel target face planes would make more than a +1 or at most a +2 difference between the lens align and any other method.

I noticed something odd in Neuro's MFA guide:

The problem with this is that the focus target (line) and the DoF scale are in the same 45° plane, so the target is not orthogonal to the sensor.

If the focust target or DoF scale were "orthogonal" to the sensor (plane), they wouldn't be visible at all as orthogonality requires a perpendicular (i.e., right-angle) relationship.

 

[Mt Spokane Photography]

I wasn't convinced that alignment was all that critical either until I found that moving the camera to one side just a tiny amount had a large effect on the MA adjustment and accuracy. I was not able to accurately determine the difference it made until after I got the lens align this last summer.

I'm sure there is a technical explanation, but for me, I get better results with the accurate alignment.

 

[epsiloneri]

I've placed a pre-order for the Reikan FoCal Pro automatic adjustment software. so I will be comparing it to the lens align system.

Seems like a very helpful piece of software! Let us know how it works out for you.

I've read all those "not perfectly aligned" arguments before and I'm not concerned by them.

I also think the emphasis on having the target very accurately orthogonally aligned to the optically axis is a bit exaggeratet. As long as the nearest and the most distant part of the target are much closer to each other than a fraction of the depth of field, you should be fine. Using a corner is an interesting idea, I have used a flat mirror with good results. Just put the flat mirror flat on the target and make sure you see the mirror image of the lens centered in the viewfinder. (as a flat mirror I've used a CD, taped tight to the tagret during alignment and then removed)

I wasn't convinced that alignment was all that critical either until I found that moving the camera to one side just a tiny amount had a large effect on the MA adjustment and accuracy.

That's indeed odd, I will check this next time.

I noticed something odd in Neuro's MFA guide

He just means orthogonal to the optical axis = parallel to the sensor.

 

[RC]

Thanks everyone for all the great info. I'm going to try a more scientific approach and see how it compares to my somewhat eccentric method.

I'm studying the pros and cons between the Lens Align and the Spyder Lens Cal--leaning towards LA, was leaning towards SLC. Ahh, the sighting method, thanks N. I tried some DIY units but I want something sturdy and less pesky.

Lens Align video tutorial
http://videos.lensalign.com/LensAlign/LensAlign-Video-User-Guide/6830310_jbdjqB
Lens Align general info
http://michaeltapesdesign.com/lensalign.html

 

[ddl_cgy]

I have a LensAlign Pro (long ruler setup), Spyder LensCal and am beta testing the latest version of Reiken Focal.

My Focal results agree for 3 of 4 prime lens with results I've derived from using the other two tools plus real life usage. My Sigma 85 f/1.4 went from +2 to +16 MA so I'm not certain what changed (distance maybe - lighting as well) - need to investigate that further.

If Focal has a few more improvements and programming bug corrections then it would definately be my calibration tool of choice for repeatability and documentation (graph of focus quantity versus MA setting, etc.) possible.

 

[Harley]

Thanks everyone for all the great info. I'm going to try a more scientific approach and see how it compares to my somewhat eccentric method.

Agreed. Thanks for sharing the info. +1 to you guys!

 

[Mt Spokane Photography]

Happy New Year, +1 to all as well.