AF micro adjust -- Sigma 150-600 etc

I have Reikan Focal that I just updated to test release 2.0.4 but I have to admit I've not yet used it.  I looked into it a year ago but at the time I was on the road and away from home 90% of the time so I just put it off.  So, now I'm looking to do both my D800E's and if I add a D7200 that would be a third camera.  Now add to that I have just today received the new Sigma 150-600 Sport lens with USB dock and also downloaded the Sigma Optimization Pro software that permits additional AF micro adjustment beyond the single point that Nikon has in camera.

So, what comes first, the camera AF adjustment or the lens adjustment of the new Sigma?  I have about 10 lenses I plan to AF adjust and that would be done with 2 or 3 bodies so that's quite a bit of work.  The Sigma optimization software appears to offer 16 points on a zoom but that actually works out to 4X focal lengths and 4X working distances at each of those 4 focal lengths.  Finally, with the Sigma being used on 2 or 3 different bodies it would seem to me a lens would be adjusted just once and then each body would be adjusted at one focal length.  If that's so what focal length would be best?  Someplace in the middle of the range, say 400mm?  Or if 600mm is the FL most likely to be used would that be the best FL to calibrate?


Brian

How are you going to incorporate the Sigma dock and the Reikan software? The Reikan software simply tells you a value to put in the camera table but then you still have to go through with the dock and do the calibration at all of the shooting distances and focal lengths. If you are going to do those with Reikan it means that you will likely have to run through the Reikan software a 100 times or more as you play high low at each of those 16 calibration points. use the Reikan on the other lenses, on the Sigma I would recommend a more manual AF Fine tune method.

If you read my review, I go through how I calibrated the Sigma and some of the problems in getting it calibrated.

Yeah, I went back over your report and perhaps I misunderstood but I got the sense you used the values from Reikan to get you a starting point for the Sigma calibration.

One more question ...

If I plan to use the Sigma on more than one camera body it would seem to me you'd need to tweak the camera(s) to the lens.  In your report you indicated that you couldn't get the Sigma cal as good as you wanted until you changed the camera value for it.  That's the chicken and egg thing.


Brian

Yes but the lens portion should stay pretty constant so once you ahve acceptable values written into the lens, a second body should be able to adjust to it via it's AFFT table.

I attempted to run Focal 2 numerous times but it just kept failing to work for me.  I had the target brightly lit and I tried several different light sources in every combination but it just wouldn't work.  I even tried using targets printed on premium presentation paper (matte) as well a glossy photo paper -- no help.  So, instead, I took a series of images at the following AF settings: -20, -15, -10, -5, 0, 5, 10, 15, 20.  I then compared the file size of the compressed images and found where the largest file sizes were and selected that as the AF value. Actually, I interpolated between the two largest file sizes.  I did that with 7 lenses and both bodies.  I do hope to figure out the issue with Focal but I have what I think are better numbers than before.

Now with the Sigma 150-600 Sport and the dock there are 16 points of adjustment but I'm still at a loss to figure out the best approach for performing this.  The setting changes need to be done when the lens is connected to the dock, but that can't be done when the lens is connected to the camera -- obviously.  In addition, I wonder if the way to determine a value is to start at one point, say, 150mm @ 2.6m, and use the AF setting adjustment in camera to determine the best value and then move to the next point, say, 150mm @ 6m and do the same.  Then complete the remaining values logging each AF value at each point.  Then, input those values into the lens via the dock.  

If as EJ discovered, one or more of the points is beyond the max +/- 20 you will then repeat the 16 tests with the lens values updated from the first test.  With the lens table updated the second round of tests should have enough adjustment range to permit the camera to find best values for all 16 points.  Then, add the new offsets to the lens table.  Finally, decide what focal length and distance is most valuable and retest at that point and set the camera value to the measured value.

Well, that's the best I can come up with and I wish Sigma had a formal procedure to do this.  I looked but did not find.


Brian

It's basically a game of high low and can be extremely time confusing.

Not sure if +1 in camera is equal to +1 through the dock (sorry if EJ has already confirmed this).
It would imply that +1 on a Canon body is equal to +1 on a Nikon body, and historically that would surprise me a bit ;o) (or would sigma have implemented different compensation units depending on the lens mount ?)

If compensation units are different between dock and body, then I would think that:
1 - you cannot use in-camera figures to update the lens table, thus have to use dock only and mount and unmount between every setting change.
2 - you cannot add offsets as easily as Brian suggested.

Vertigo wrote:Not sure if +1 in camera is equal to +1 through the dock (sorry if EJ has already confirmed this).
It would imply that +1 on a Canon body is equal to +1 on a Nikon body, and historically that would surprise me a bit ;o) (or would sigma have implemented different compensation units depending on the lens mount ?)

If compensation units are different between dock and body, then I would think that:
1 - you cannot use in-camera figures to update the lens table, thus have to use dock only and mount and unmount between every setting change.
2 - you cannot add offsets as easily as Brian suggested.

It does not.

The process for calibrating this lens is as follows:
1. calibrate the lens with the dock at each of the 16 points.  As long as every point is within the calibratable range, then you are good to go.  If it is not, then you have to dial an offset into the camera to compensate in the opposite direction of the direction that the lens could not be calibrated in and redo the 16 points.
2.  For a second body, find an in camera setting that gets the longest focal length at infinity on target.  The other 15 points will be very close if not exactly on target.

But, if there's no correlation between the camera AF values and the lens dock values I have to ask where the numbers come from that are put into the dock. I mean, it can not be the case that you make a guess, attach lens to dock, test again on camera, then repeat a dozen times for each of the 16 points -- that would be insane and would involve perhaps hundreds of lens removals from camera and dock.

As I said I looked around for an actual procedure from Sigma but could not find one.  Sometimes the Japanese are anal to an extreme degree with there procedures, but at other times they just can't come up with anything.


Brian

As I understand it you indeed have to mount a lot of times, at least if you plan to have an exhaustive approach to it (all possible combinations).
Think that some bodies still do not have AFMA at all, and the dock was probably partly designed for these users.

Maybe sigma thought that people would have a more partial approach, e.g. correct only at the fl/distances where they observe massive misfocus, and neglected to publish the complete procedure because of that(or purposedly did not publish it, because it is too complex for most consumers)

Even using dock-only, I imagine you can decrease the number of mounting by:
- setting +1 for all fl/distances through the dock
- then shoot the 16 fl/distances combinations with the lens on body
- back to the dock, +2 for all points
- ...
- search for the sharpest shots afterwards.
You would have to set 4 targets at various distances, though.

I must say that I would be happy to test the C version as an alternative to the 400/5.6 (x1.4 also). If I see encouraging side by side real-world IQ comparisons, and also reports that the customizable focus limiter can be useful for BIF (e.g. very fast AF when limited to 10-30m), I might pull the trigger.

Brian Stirling wrote:But, if there's no correlation between the camera AF values and the lens dock values I have to ask where the numbers come from that are put into the dock. I mean, it can not be the case that you make a guess, attach lens to dock, test again on camera, then repeat a dozen times for each of the 16 points -- that would be insane and would involve perhaps hundreds of lens removals from camera and dock.

As I said I looked around for an actual procedure from Sigma but could not find one.  Sometimes the Japanese are anal to an extreme degree with there procedures, but at other times they just can't come up with anything.


Brian

That is exactly how it's done!  But it generally doesn't take a dozen per point but can easily take 4 or 5.  It's a very long and drawn out process.  it's no problem on a prime but on a zoom it is very complex.

Well I'm thinking the huge number of lens swapping required with that approach just can't be the best answer.  OK, it is likely that -2 with a Nikon body is not exactly the same as -2 on a Canon body or -2 on the lens itself.  Frankly, a -2 on one Nikon body is not likely to be exactly the same as -2 on another Nikon body of the same model.  However, a negative value is a negative value.

My thought is to do this:

With the lens connected to the camera and the lens at default zero values I'd shot 9 shots at each of the 16 zoom and distance settings then determine the AF values for each of those 16 points then attach lens to dock and put those numbers in.  Then repeat the same tests with the lens updated with the first set of values.  At each point, if the cameras AF values is, say, -2 then I'd update the lens in the dock with new values.  So, if the first lens value was, say, -12 and the second camera AF value for that point is -2, then the adjusted value in the lens would be -12 + (-2) = -14.  With this approach you shouldn't need to swap the lens more than a few times.  I would hope 3 passes would be all that's needed.  Even then, that's a crap ton of testing and lots of shutter releases.


Brian

There are lots of ways to do it but be warned that adjustments to one setting can have a bit of an affect on others - there is enough overlap in the ranges.  So once you go through your whole procedure, go back and check all 16 points one more time as there might have been a slight amount of shift induced by an adjacent setting.  When I do a 150-600 it takes me about 4 hours to do it so that it is completely right.  This is what the table looks like when calibrated to my D7200 with the D7200 set to +11:

With the difficulty and time required to do a complete adjustment of a zoom lens I have to believe there's an opportunity for someone to automate the process to greatly speed up the process and reduce manual involvement.  Additionally, given that these values are digital parameters within either the camera or lens or both there's no reason that the camera couldn't have a more comprehensive adjustment table with many more points of FL and distance for each lens. But, getting that done on 2 or 3 camera bodies and perhaps 10 lens would strain even the most determined pros ambitions.  

I can envision an automated approach that would reduce the human involvement to just a few minutes per camera such that a pro or series amateur might send all there gear and for perhaps $15 per camera/lens combo (fixed) and $25 per camera/lens combo (zoom).  The Focal system isn't practical for this given the multiple distances involved and as I said, I haven't been able to get it to work for me at all.

Sadly, Nikon in particular isn't real interested in permitting automatic AF settings changes via PC connection and to fully automate the process you'd need that.  I just don't get why the camera makers are dragging there heals in this respect.

I should receive the D7200 tomorrow so I'll be looking to cal the Sigma 150-600 particularly with the 7200.  As I said I'm planning to hit Yellowstone in September though the smoke from the wildfires might postpone or cancel it for this year.


Brian

(I suggest to add sigma 150-600 in the thread title)

Going from -17 to +20 for the same lens, depending on the distance/FL ?!! Out of the box images must be awfully soft !
Sorry sigma, but for me it is a testimony of very poor in-factory AF calibration.
Or even worse: would it be possible that (reverse-)engineers at sigma were not able to fully understand how a Nikon body communicates with a lens for AF ?  If they did not fully crack the signal coding for distance, it could result in such large non-linear error.

EJ, these zooms are a goldmine for you !

Vertigo wrote: EJ, these zooms are a goldmine for you !

Not even close   I have to charge way too much for most people to do a Sigma zoom due to the complexity and time required to do the work properly with the dock; add in shipping and insurance and you are at almost $300 to calibrate a single lens.

Vertigo wrote: Going from -17 to +20 for the same lens, depending on the distance/FL ?!! Out of the box images must be awfully soft !
Sorry sigma, but for me it is a testimony of very poor in-factory AF calibration.
Or even worse: would it be possible that (reverse-)engineers at sigma were not able to fully understand how a Nikon body communicates with a lens for AF ?  If they did not fully crack the signal coding for distance, it could result in such large non-linear error.

BTW, most Sigma lenses have always required extreme adjustments on Nikon cameras.  You are often right up against the rails.

E.J. Peiker wrote:

Vertigo wrote: Going from -17 to +20 for the same lens, depending on the distance/FL ?!! Out of the box images must be awfully soft !
Sorry sigma, but for me it is a testimony of very poor in-factory AF calibration.
Or even worse: would it be possible that (reverse-)engineers at sigma were not able to fully understand how a Nikon body communicates with a lens for AF ?  If they did not fully crack the signal coding for distance, it could result in such large non-linear error.

BTW, most Sigma lenses have always required extreme adjustments on Nikon cameras.  You are often right up against the rails.

Interesting about Sigma and Nikon...I have 2 Sigmas, a 50/150 and a 35 1.4 and have never noticed any deviation from native Sony lenses on FF or cropped cameras.

E.J. Peiker wrote:BTW, most Sigma lenses have always required extreme adjustments on Nikon cameras.  You are often right up against the rails.

Yes, I had a sigma 300/4 APO for Nikon that front-focused so badly that you could not compensate. I tried a second copy that had the same problem. I did not care much because it was a proxy lens for me (MFD 1.2m, one of the copies was razor-sharp and the bokeh was fantastic).
Also the 100-300/4 was (and still is) a great lens, but had focus inconsistencies: if you AFMAd at 300mm, it was way off at 100mm. Not a new problem it seems !

My feeling is that +19 on a prime is an offset, that you can attribute to mount thickness tolerances.
-17 to +20 on a zoom suggests a worse problem, with the image plane moving all over the place.

Some other examples:
Sigma 150 f/2.8 OS Macro is -19
Sigma 35 1.4 is +15

Then again my Nikon 50/1.4 is -19 and any Nikon 500 + 1.4x you have about a 40% chance of not being able to calibrate in the range of adjustments available to you.

Sigmas do tend to behave better on Canon.  The conspiracy theorists on the web claim that Nikon does something in firmware to make non-Nikon lenses be all over the place  In fact, in general, Canon, especially newer bodies and lenses rarely need more than low single digit adjustments.

But, Sigma on Nikon, once you get them calibrated, there are a number of truly exceptional Sigma lenses.