D5/D500 Auto Fine Tune procedure that works!

Much has been said about the convenience of the new Nikon auto AF Fine Tune function but its accuracy has also been called into question.  I have spent two days fully characterizing the new function and coming up with a procedure that will give you accurate AF fine tune values.  A big part of the problem is that there is as much as a plus or minus 5 point shot to shot variance in AF fine tune values due to the inherent inaccuracy of having the AF sensor in a completely different place than the sensor that the lens is supposed to be focusing the image on.  In a DSLR that is being shot in its normal reflex mode, that is NOT in live view mode, a small percentage of the light that passes through the lens is transmitted through the mirror.  This light is then bounced off of a secondary mirror to the autofocus sensor array, typically somewhere in the bottom of the camera.  This is where the AF data is taken and a signal is then sent to the lens to adjust its focus point to where this AF sensor array thinks the optimal focus setting is.  There are inherent inaccuracies in doing autofocus this way due to manufacturing tolerances in the light path to the AF sensor and the light path to the imaging sensor which are different.  Live view does not have this inherent inaccuracy since AF is done on the sensor itself.  So, when doing normal off sensor Phase Detection Auto Focus, which is what the normal AF module is doing, these tolerances plus moving parts in the whole camera/lens system will have shot to shot differences.

The fundamental flaw of the new Automatic AF Fine tuning, as Nikon has implemented it, is that it takes a single shot to determine the AF Fine Tune value which could be as much as 10 points off due to the plus or minus 5 point shot to shot variance.  When somebody that knows how to perform an accurate autofocus fine tune does a manual tuning procedure, they take several shots to obtain a median or average value based on the shots taken.  We can do the same with auto AF Fine tune. Here is the procedure - it should only be performed with the camera mounted on sturdy tripod and the head locked down:

1. Set up a high contrast focus target.  The downloadable and printable target for the FoCal AF software (which I don't currently recommend) is a great target for this purpose.  Print it, or another target and tape it to a perfectly vertical surface such as a wall.  Now perfectly align the camera both in height and tilt to the target so that the sensor is perfectly parallel to the camera's sensor.  You can also utilize the LensAlign II target if you own this as it has a simple to use target alignment system.  The camera should be about 25 times the focal length away from the target for general use or at the distance that you do your shooting at if you are in a controlled environment such as a studio.

2. Now put the camera in live view and focus the image with the focus target in the center.  Now initiate the auto AF Fine Tune system by hitting the Focus selection button on the lower left of the lens mount (with you standing behind the camera) and the movie record button (small red button next to the shutter button) simultaneously and hold until the screen asks you to confirm that the camera is steady.  Select yes and a value will be entered in the AF Fine Tune table.  Now go to the AF Fine Tune menu item at the bottom of the first page of the Tools Menu (wrench icon), make sure that AF Fine Tune is set to On and then read out the value that was entered in the Saved Value item.  Write this value down.

3. Repeat the same procedure as in number 2 above 9 more times for a total of 10 samples.  Write all of them down.

4. Now cross out the lowest value (the most negative if a negative number) and the highest number (the least negative number if the highest number is also negative) to get rid of any weird outliers.  Next, take an average of the remaining 8 values by adding them up and dividing by 8.  Round this to the nearest whole number and enter this value in the Saved Value line item of the AF Fine Tune menu.

5. Repeat the above for every lens and every lens/teleconverter option.  On zoom lenses, use a focal length at or near the longest focal length unless you do not use your lens at that focal length.

Here is an example of a 200-400 f/4G lens that I just fine tuned on a D500.  I first manually did it and came up with a value of -2.  The following AF Fine Tune values were obtained using the auto AF Fine Tune procedure 10 times:
-1, +3, -3, -2, -2, -3, -4, -1, -5, -1, -3

Throwing out the highest and lowest values (+3 and -5) and then averaging the rest gives a result of -2.37.  Rounding that to the nearest integer, you get a value of -2 which is identical to the value I came up with.  

I have repeated the above on 8 different lenses at this point and in every case the procedure and my manual method was either identical or within one point of each other.  One point difference is way inside the shot to shot error and is inconsequential.

Hope this helps

Still haven't pulled the trigger on a D500 but I have saved this to help when I do. Looks very workable. Thanks!

Sure is nice that Nikon did this auto fine tuning....so easy that a child could do it. (sarcasm)

Thanks EJ.
Your's and Steve Perry's info has certainly helped de-mystify for me what is happening with the AF system and how best to do a AF Fine Tune whether with a D5/D500 or other camera. Very useful information.

Thanks E.J. for taking the time to explain the Nikon Fine Tune procedure. 

One question I have is concerning holding the Focus Selection Button and the Movie Record Button Simultaneously until the screen asks you to confirm that the camera is steady.  I'm probably missing something, but by having both your hands on the camera while it takes a test shot, wouldn't this cause camera shake just by having your hands holding the buttons as the shot is taken? I thought to negate camera shake you need to employ the cable release and shutter delay option?

Scotty wrote:Thanks E.J. for taking the time to explain the Nikon Fine Tune procedure. 

One question I have is concerning holding the Focus Selection Button and the Movie Record Button Simultaneously until the screen asks you to confirm that the camera is steady.  I'm probably missing something, but by having both your hands on the camera while it takes a test shot, wouldn't this cause camera shake just by having your hands holding the buttons as the shot is taken? I thought to negate camera shake you need to employ the cable release and shutter delay option?

Yes that is a good point but if you truly have the sensor plane and the focus target plane perfectly parallel this should not matter very much.  Also this can be minimized by locking things down firmly.  I should have said this procedure must be done with the camera on a tripod or some other very solid support!  I have edited the text above to include that.

Wow EJ, thanks a bunch for this info you da man!!! I am off for a nice 3 day weekend and may try this with my 200-400VR. Hope it doesn't take all 3 days! LOL.

I have to figure out where I can get a nice level run out back I guess (wait I have a pool so on the pavers!). My calc would be 400mm x 25 = 10000mm = 33 ft (rounded). Then do the same for the x1.4 since that is the only one I use on that lens. Question becomes with the TC on what focal length to use. I guess all the way to 400 again....that can be the issue since not really sure how often I go all the way to the max with a x1.4 but we will try a couple and see if any real differences (say 300mm with TC and 400mm with TC)

Anyway, thank you for giving us this info and for taking the time to actually try it first! VERY much appreciated!

Come to NY one day I owe you a drink or 2!

ps: Besides that how did the camera seem to you?

George

George DeCamp wrote:Wow EJ, thanks a bunch for this info you da man!!! I am off for a nice 3 day weekend and may try this with my 200-400VR. Hope it doesn't take all 3 days! LOL.

I have to figure out where I can get a nice level run out back I guess (wait I have a pool so on the pavers!). My calc would be 400mm x 25 = 10000mm = 33 ft (rounded). Then do the same for the x1.4 since that is the only one I use on that lens. Question becomes with the TC on what focal length to use. I guess all the way to 400 again....that can be the issue since not really sure how often I go all the way to the max with a x1.4 but we will try a couple and see if any real differences (say 300mm with TC and 400mm with TC)

Anyway, thank you for giving us this info and for taking the time to actually try it first! VERY much appreciated!

Come to NY one day I owe you a drink or 2!

ps: Besides that how did the camera seem to you?

George

EJ is spoiled now after using a hundred megapixel Medium Format.

George DeCamp wrote:Wow EJ, thanks a bunch for this info you da man!!! I am off for a nice 3 day weekend and may try this with my 200-400VR. Hope it doesn't take all 3 days! LOL.

I have to figure out where I can get a nice level run out back I guess (wait I have a pool so on the pavers!). My calc would be 400mm x 25 = 10000mm = 33 ft (rounded). Then do the same for the x1.4 since that is the only one I use on that lens. Question becomes with the TC on what focal length to use. I guess all the way to 400 again....that can be the issue since not really sure how often I go all the way to the max with a x1.4 but we will try a couple and see if any real differences (say 300mm with TC and 400mm with TC)

Anyway, thank you for giving us this info and for taking the time to actually try it first! VERY much appreciated!

Come to NY one day I owe you a drink or 2!

ps: Besides that how did the camera seem to you?

George

You would use the actual focal length of 560mm and then calculate the distance based on that.

Mike in O wrote:EJ is spoiled now after using a hundred megapixel Medium Format.

That makes sense, I guess it depends on where you're coming from!

E.J. Peiker wrote:

You would use the actual focal length of 560mm and then calculate the distance based on that.

Great! Thanks again!

I have given it a try and is weeks to delivered reasonably close figures. It has set my 500mm f4 at +2 and with a 1.4 tc at +4

Don't know if these are correct will test it tonight. I used a dottune Af afterwards and that gave results of +4 and +6

E.J.:  Thanks so much for your process of mathematically removing reducing outlier values.  Simple and clever.    

I am one who also thinks my D500 needs tweaking here.   It is perfect at fairly close distances but I think it is a little off  out there a ways.  

I definitely will use your technique soon, and not rely on just one or two adjustment series.

Blck-shouldered Kite wrote:E.J.:  Thanks so much for your process of mathematically removing reducing outlier values.  Simple and clever.    

I am one who also thinks my D500 needs tweaking here.   It is perfect at fairly close distances but I think it is a little off  out there a ways.  

I definitely will use your technique soon, and not rely on just one or two adjustment series.

Robert, could you clarify what it is that is "perfect at fairly close distances ?



Based on E.J.'s founding, it seems to me it's a sample-size issue. Perhaps what we were expecting was for Nikon to give us THE value one test image. When in reality, just like LensAligh and others for example, we should be taking several test images to find that value. Seems to me it's nothing that a firmware update cannot fix? Right, Nikon? 

I'll try, but there are so many variables, so that often.....what I think I see, could really just be bias based on several other factors.

After thousands of images with my 80-400/D500 combo (and also my former D7200/80-400 combo), it seems to me that I am either "AF fine tuning" this lens with bias toward the closer distances, or the lens is inherently sharper at closer distances. I fine tuned the AF on the D500 at about 30 feet distance. I did take many, many test images, trying to get that sample size way up there. Am not home now so I do not know what my default setting is ended at. I need to read more on this and do more testing....no doubt.

At this point I am suspicious that my 80-400 AFS/D500 combo is "right on the money" at closer distances and front focusing at farther distances....say 50 feet and out, but more surely at maybe 80 feet and out. And I really am not certain about all of this. But that is best answer I can give you DChan....,:)