Lens Micro-Adjustment - Review/ tutorial

[Bob Keenan]

I just put together a short review of various methods of camera/lens micro-adjustment. All of these methods have been written about on the net. What I tried to do is put them all together in one blog post.

http://www.bobkeenanphoto.com/blog/

This article is more for people who have never done micro-adjustment or have only tried one method. Feel free to comment.

Bob Keenan
http://www.bobkeenanphoto.com

 

[Cem_Usakligil]

Hi Bob,

Welcome to OPF. We are happy to have you on board with us and thanks for the useful link to the lens-micro adjustment in your blog. I have a couple of issues with it.

Firstly, the original source of the micro-adjustment you and Northlight Images are referring to is right here by our own Bart van der Wolf, as also acknowledged by Keith Cooper in his website. So the question is, is your blog adding a value to what has already been written here? From a first scan of the contents, this seems to be the case up to a certain extent so I don't have a problem with it. Where I have a problem with is that you should give credits to Bart/OPF as the main source of your article and also include a backlink to the the original thread.

Secondly, we do not allow new members to advertise in any way for their own products and services. Even when your blog is non-commercial, this still constitutes a breach of our policies as you are posting here for the first time in order to draw traffic to your blog.

So now that we have this out of the way, please realize that this is not personal and I still am glad to have you on board and looking forward to seeing your contributions. Welcome.

Cheers,

 

[Doug Kerr]

Hi, Bob,

I just put together a short review of various methods of camera/lens micro-adjustment. All of these methods have been written about on the net. What I tried to do is put them all together in one blog post.

http://www.bobkeenanphoto.com/blog/

I really nice article. Thanks so much.

I was especially interested in this passage:

"It turns out that each push of the focus button [from the EOS Utility in tether mode] moves it the equivalent of one MA adjustment."

I have been doing a lot of work lately to understand many aspects of the Canon AF system.

One thing I have learned is that, after a focus error measurement is made (via the AF detector), the order to the lens to move to a different focus position is given in terms of what I will call "ticks", which are the unit of focus mechanism movement that are delivered by an optical encoder (such as a "chopper wheel") on the focusing cam barrel or in the drive for it. The lens reports to the body the "scaling" of the "ticks" it uses (in term of the number of ticks per mm of equivalent lens extension movement).

It may well be that the unit of AF MA, and thus (per your finding) the increment of EOS Utility focus movement, is in terms of a fixed number of ticks (perhaps even one). I will do some playing around with that when I get the chance.

I haven't been paying much attention to AF MA (I have no body that supports it, for one thing) so I may have missed some clues in this regard.

Thanks again for the nice review of MA technique.

Best regards,

Doug

 

[Bob Latham]

The "each push of the focus button [from the EOS Utility in tether mode] moves it the equivalent of one MA adjustment" statement can be expanded slightly further in that each unit of MA equates to 1/8 of the Depth of Focus.
I believe that this means (unless I'm missing something) that a wide, fast prime will have less available correction than something akin to a 600/4 given the +/-20 units of MA available on Canon bodies.

Bob

 

[Doug Kerr]

Hi, Bob,

The "each push of the focus button [from the EOS Utility in tether mode] moves it the equivalent of one MA adjustment" statement can be expanded slightly further in that each unit of MA equates to 1/8 of the Depth of Focus.

Interesting.

The reference there is probably the single-sided depth of focus (Canon usually uses that as a unit in these matters). Canon defines that as 0.035*N (in mm), where N is the f-number. (Note that the focal length of the lens does not enter into it at this stage.)

So for an f/2.0 lens, the unit "one depth of focus" is 0.0175 mm. 1/8 of that is about 0.0022 mm.

That change in focus in image space corresponds, in object space, to a change in the location of the object plane of focus of approximately:

(0.005*D^2*N)/f^2

where D is the distance to the starting plane of object focus (in mm), N is the f-number, and f is the focal length (in mm). (So here's where the focal length enters the picture.)

Thus. for a 50 mm f/2 lens and an initial focus distance of 3000 mm (3 m), the increment of focus change would be approximately 36 mm.

Now, for a 35 mm f/2 lens and that same initial focus distance, the increment of focus change in object space would be approximately 73.5 mm.

For a 600 mm f/4 lens and that same initial focus distance, the increment of focus change in object space would be approximately 0.5 mm.

Again, this is all based on the unit of focus adjustment being "1/8 the depth of focus" (i have no personal knowledge of that).

By way of comparison, in older Canon EF lenses, when adjusting the "best focus correction value" in the lens, by way of an electrical jumper setting, often the step of adjustment is 1/2 or 2/5 the depth of focus.

I believe that this means (unless I'm missing something) that a wide, fast prime will have less available correction than something akin to a 600/4 given the +/-20 units of MA available on Canon bodies.

I'm not sure exactly what you mean by that.

Keep in mind that focus errors in lenses are typically about in the same broad neighborhood in image space. Thus a range of ±20 MA units has about the same impact on lenses of any focal length.

Best regards,

Doug

 

[Bob Latham]

Hi, Bob,
The reference there is probably the single-sided depth of focus (Canon usually uses that as a unit in these matters).

Yes, single sided and it was remiss of me not to state that fact...sorry

I believe that this means (unless I'm missing something) that a wide, fast prime will have less available correction than something akin to a 600/4 given the +/-20 units of MA available on Canon bodies.

Hi, Bob,

I'm not sure exactly what you mean by that.

Keep in mind that focus errors in lenses are typically about in the same broad neighborhood in image space. Thus a range of ±20 MA units has about the same impact on lenses of any focal length.

Doug,
I'll quote a portion Chuck Westfall's entry in the Digital Journalist Tech Tips Dec 2008
"The amount of focus adjustment per step is proportional to the maximum aperture of the lens"
No mention was made of focal length being factored in so I took the statement at face value and hence my belief that fast (large aperture) primes have less latitude available for focus errors to be corrected.

Bob

 

[Doug Kerr]

Hi, Bob,

Yes, single sided and it was remiss of me not to state that fact...sorry

I'll quote a portion Chuck Westfall's entry in the Digital Journalist Tech Tips Dec 2008
"The amount of focus adjustment per step is proportional to the maximum aperture of the lens"
No mention was made of focal length being factored in so I took the statement at face value and hence my belief that fast (large aperture) primes have less latitude available for focus errors to be corrected.

As I understand it, the adjustment is applied in a way so as to change the position of the image with respect to the focal plane (it works in "image space"). The unit of adjustment is an amount of that change. Its actual size does varies with f-number but indeed not with focal length.

The discussion of the size of the unit in terms of "depth of focus" (rather than depth of field) is consistent with that outlook, helping me believe that my understanding is generally correct.​

But we perceive focus error not in image space but rather in object space (as the location of the plane of object focus not being at the target - "front or back focus" as we speak of it).

A certain shift of the location of the image corresponds to a certain shift of the location of the plane of object focus. The relationship between the amounts of those two shifts depends on the focal length of the lens and the initial distance of the plane of object focus (that is, "where is the camera roughly focused").

Accordingly, the amount of change in the focus error we notice (in front or back focus) for one unit of adjustment depends on:

• The f-number
• The focal length
• The basic distance at which the camera is focused

Going at it in the other direction, if we find that we consistently have a back focus of about 11 mm using a certain lens, with a certain focal length, when focused on a target at about 3 m distance, that will suggest a certain shift in the MA, but that would have to be calculated, the calculation involving the focal length, the target distance, and the f-number.

********

I haven't really followed all the guides on the effective exploitation of the MA capability (I'll try and read some as soon as I can). Maybe we don't do it just as I said.

Maybe if we experience a back focus of about 11 mm with some lens when focused at some distance, the 11 mm is enough by itself to tell us what change in MA will most likely cure the error.

If that's so, let me know, and I'll try to find out where my understanding has gone wrong.

Best regards,

Doug

 

[Doug Kerr]

Let me review a related matter.

The AF operation revolves around the camera being able to measure a "focus error" by "phase comparison". Basically, that part of the system examines the relative positions of two copies of a part of the image as observed on two "AF subdetectors". It is quite equivalent to the comparison of the alignment of the two images on the two halves of a split-image focusing aid.

But here, the ideal result is not actually perfect alignment of the two images, mainly because the two subdetectors are offset from each other (they can't be in the same place!).

Hopefully, the working of the two little optical systems is such that the perfect result is when each image is the same distance from a reference point on its subdetector. But of course that doesn't always happen.

Thus, the ideal result is defined in terms of a certain difference between the locations of the two images from the reference points on their respective subdetectors. That reference difference is determined by precise measurement at the factory (or later at a Service Center) and stored in non-volatile memory in the camera.

If in fact life at the AF subdetectors was a precise analog to life at the image detector, that would be the end of the story. But it isn't.

Recall that each of the subdetectors works from rays through a little aperture, one near each edge of the actual aperture.

Various phenomena in the lens, most prominently spherical aberration, mean that rays passing through the center of the "real" aperture do not focus at the same point as rays near its outer edge. This means that the"best focus" setting for the actual shot (involving all the rays through whatever aperture we have chosen) will not necessarily be the same as the focus that will seem ideal to the AF detector system (working only with rays through two outboard parts of the aperture).

The difference is small, but not negligible.

Note that this phenomenon will be different for the use of a "standard" AF detector or an "enlarged baseline" AF detector (the ones that require an aperture of f/2.8 or better), since their little apertures take rays from different places across the whole (possible) aperture.​

To help overcome this matter, each lens transmits to the body, at the commencement of an AF operation, a best focus correction value (BFCV). The body applies this to the focus error reported by the phase comparison itself. The objective is that the corrected reported focus error tracks with the effect of the focus setting on the actual image.

The BFCV depends on:

• The lens model
• Small peculiarities of the individual copy of the lens
• The current zoom setting (for a zoom lens)
• The current focus position of the lens optics
• Whether AF is planned to be with a "standard" AF detector or a "enlarged baseline" AF detectors

The value given to the body will typically be based on all of those.

The value is stated in terms of an offset of the focus error in terms of the location of the image with respect to the focal plane (the "image space" focus error).

The unit of this value is some fraction of the single-sided depth of focus, always defined by Canon as 0.035*N mm, where N is the f-number.

Now, my understanding is that the MA value (as stored for the particular lens and its current focus position) is just added to the BFCV received from the lens before being used by the body to adjust the interpretation of the focus error as indicated by the phase comparison operation. Thus, it works in just the same way.

My discussions earlier in this thread are based on that understanding (which of course could be incorrect).

My thanks to my colleague "Wilba" from ProPhoto Home forum (and elsewhere) for first calling clearly to my attention the role of spherical aberration in this whole chain.

Best regards,

Doug