July 11, 2019

Please reload

Recent Posts

I'm busy working on my blog posts. Watch this space!

Please reload

Featured Posts

Does the D500 Automatic Focus Fine-Tune Calibration Work?

April 29, 2017

Short answer: yes.  But you know there’s a catch.  There’s always a catch.  So here it is: garbage in, garbage out.


The ‘secret’ to getting great results for focus calibration is multi-faceted.  You must use the proper focus target, your camera must be very stable, you need consistent technique, you need to decide on a zoom setting, you need to pick a target distance, you may need to pick an aperture, and you must use good lighting.


As much as people hate to hear this, it’s a fact of life that measurement and calibration always involve statistics.  Your gear isn’t perfect, so you are guaranteed to get some variability in measurement.


Focus calibration isn’t a “one and done” scenario.  Many measurements are required, and then you have to find the average value.  You might also have to throw out the measurement values of outliers.


The Nikon D500 “automatic” focus fine-tune calibration assumes that your live-view, contrast-detect focus is dead-on.  It compares the distance your phase-detect focus system decided to focus on a target to the distance that the contrast-detect focus system decided to focus on that same target, and then calculates the fine-tune value that would shift the lens phase-detect focus to match the contrast focus.


Depending on which direction (near-to-far versus far-to-near) your lens travelled to obtain focus, you may get a different calibration answer.  Depending on what your selected focus point ‘sees’ as being the subject, you may get a different answer.  Depending on your selected lens aperture (mostly high-speed lenses), the answer may vary.  Depending on the zoomed focal length, you can get a different answer (think parfocal optics).  Depending on the target distance, you can get a different answer.  In dim light, you can get a different answer.  The physical lens focus mechanics have some slop.


So don’t expect miracles here.  You need to decide on your favorite zoomed focal length, distance, and aperture to use while conducting the test (depending on the type of lens being measured, of course).  You need to use a proper focus target that has absolutely no ambiguity about what the “target” really is, from the standpoint of your focus sensor.


Sigma is savvy to these ugly facts of optics life, and provides their newer lenses with the ability to calibrate zooms at multiple focal lengths and multiple focus distances. Their (inexpensive) USB dock and free software lets users reprogram the lens firmware with this calibration information, in addition to letting users select focus algorithms, upgrade firmware, select anti-vibration modes, and other features.  Nikon is not savvy, but I digress.


I use a specific focus target, rotated to a 45-degree angle, to calibrate focus.  The target design, and its associated analysis software, was created by Frans van den Bergh.  I wrote an article to explain his software and focus target here.


The recipe to perform a single calibration measurement on a D500 goes as follows:
⦁    Lens VR OFF
⦁    Focus mode set to AF-S
⦁    Select the center focus point
⦁    Camera/lens on sturdy support pointing at the focus target
⦁    AF fine-tune ON in setup (wrench) menu
⦁    Live View mode
⦁    Normal-area AF
⦁    Focus on the target. Focus point centered on an edge illuminated by bright light.
⦁    Press the “AF Mode” button and “Movie Record” buttons simultaneously and wait patiently for about 3 seconds.  Don’t wiggle the camera…
⦁    Highlight “Yes” and press “OK” when prompted.


The calibration value will be written for the lens in the usual AF fine-tune menu location.
Rinse and repeat.

 

The focus target using the MTF Mapper program, showing measurement labels on squares

 


Note in the photo above that the camera focus point is centered on the right-hand vertical edge of the large black target.  There is no ambiguity about what it is focused on; the focus point can only see a single vertical high-contrast edge.  The MTF Mapper program measures the resolution of every black square it finds, and lets the user easily see where the optimal in-focus squares are located relative to the large black central target right-hand vertical edge.  Using this software, you can visually see where the sharpest edges are located with each test you make (the edge labels eliminate any judgement calls you’d have to make in an un-labelled photo).


After each AF fine-tune measurement is done, write down the value saved in the AF fine-tune menu for the lens being used.  Take several measurements and average their values.  You may have to throw out any wild readings first.  These readings will give you a feel for the natural focus variation of the lens.


I like to manually change the focus to alternate between near and far for each test, prior to initiating AF-S auto-focus.  This will let you explore any bias that the lens has for focus direction.
Manually enter the averaged fine-tune value for your lens in the AF fine-tune menu.  This will give you the best “typical” focus result.

 

 

An MTF Mapper “profile” plot of the focus chart photo.

 

 Note “AF Tune Value” is displayed in the plot above.  The D500 focus calibration feature automatically saves this value in your camera.  The MTF Mapper program extracts the value from the focus chart photo EXIF data to add it to the plot.


Conclusion


I used this automatic fine-tune calibration feature dozens of times, and the fine-tune value tracked the actual phase-detect focus error quite well (within about 1 or 2 counts).  The problem is that it can only calibrate against a moving target.  Your own lens natural focus variation will prevent you from ever getting “the” calibration answer.


In my own testing, I would get an auto-calibration tune-value variation of about plus/minus 3 for a typical lens. This is roughly the same as my own manual calibration best efforts; it would get within about 1 or 2 counts of what the software measurement software determined to be "best".  I have read reports that some users get terrible calibration repeatability, but I suspect that may be largely due to using a poor focus target and/or sloppy technique.


I don’t think that phase-detect can ever compete with Live View focus accuracy across the board, because it cannot address issues related to focus-shift-with-zooming, focus-shift-with-distance, or spherical aberration effects. None the less, I applaud Nikon for adding this feature.  Now if they could match Sigma lens firmware capabilities, they’d really be leader of the pack.  By the way, now even the Samyang (Rokinon) company has added the same firmware-programming features that Sigma offers.


Oh, and one thing I'd definitely change with their auto-calibration is the two-button-press thing.  There's no way to keep things rock-steady doing that.  They need to make it an operation you can do using a remote trigger or timer so that you don't have to jiggle the camera.


 

Tags:

Please reload

Follow Us
Please reload

Search By Tags