- Ed Dozier
Autofocus Fine-Tune on Nikon Z Cameras: Useless or Not?
I was comparing a DSLR (the D850) against a mirrorless Nikon Z8 using the same lens. After careful autofocus fine-tune adjustments, I was able to improve the focus results for the D850 camera. Since both cameras have the same sensor resolution, the Z8 (or Z9) should be able to achieve at least the equivalent focus accuracy (e.g. equal lens resolution measurements) of the D850.
I have never messed with the autofocus fine-tune feature on either my Z8 or Z9. I have read that it’s completely unnecessary, and until now I had believed this narrative. It now seemed like an appropriate time to go there. All of the testing and measurement associated with focus calibration isn’t exactly enjoyable, but there’s no way I was going to let my D850 beat my Z8 or Z9 without a fight.
I should mention that I used AF-C with 3D-tracking focus on both cameras. This is my favorite focus mode, so it’s how I like to test my lenses for real-world resolution results. The best way to determine the highest resolution a lens can get is to forget focusing the lens entirely, and instead move the camera/lens combination on an adjustable rail in small (sub-millimeter) amounts between test shots. But that's not how photographers actually shoot.
Focus Fine-tune in the Nikon Z8 Setup Menu
By default, the Nikon Z camera autofocus fine-tuning is turned off. The fine-tuning features are located in the Setup menu.
AF fine-tuning menu
To adjust the (attached) lens, scroll down to the “Fine-tune and save lens” option.
Set the desired setting
As shown above, I set the 500mm f/5.6 PF fine-tune value to -2, which will then force the lens to focus slightly nearer to the camera. Press “OK” to save the new setting.
The Nikon D850 resolution testing results
As shown above, my Nikon D850 was focus fine-tuned to -2 when I conducted focus testing. Many of the resolution readings hovered around MTF50 80 lp/mm for the center of this Nikkor 500mm f/5.6 PF lens. This was the best that I was able to get out of that camera/lens combination. I use the free MTFMapper program to measure lens resolution.
By the way, the horizontal axis values above are just the frame number of each shot.
The data plotted above had an average MTF50 resolution of 71.7 lp/mm, with a peak of 84.8 lp/mm and a standard deviation of 10.3. Notice that many of the resolution readings hovered around 80 lp/mm.
There’s a huge spread of resolution results, which means that the camera was quite variable in how it focused on the resolution target. This is unfortunately completely normal for a DSLR using phase-detect focus. This spread (variation) of resolution results is reflected in the standard deviation value of 10.3.
The Nikon Z8 resolution testing results
My Nikon Z8 testing results (with NO AF-tune adjustments) are shown above. Their consistency is superior to the D850, but many of the resolution readings seem a bit lower than what the D850 had occasionally achieved.
The Z8 resolution testing results had an average MTF50 resolution of 74, with a peak of 98.2 and a standard deviation of 6.7.
While the Z8 on average got better resolution than the D850, it looked like there was room for improvement. I tried various AF-tune values on the Z8, and it turns out that the same AF-tune value of -2 got the best results out of this lens. Normally, different cameras will have very different fine-tune settings with the same lens.
The Nikon Z8 resolution testing results with AF tune -2
The resolution results from the data plotted above got an average MTF50 resolution of 77.1. The peak reading was 84.1, with a standard deviation of 5.3.
Although not huge, the lens resolution increased from an average of 74 to 77.1 lp/mm after implementing the fine-tune compensation.
Note that the D850 standard deviation was 10.3, while the Nikon Z8 standard deviation was 5.3. This demonstrates significantly less focus variation with the mirrorless camera.
A word of caution is in order. Don’t use ‘pinpoint’ or ‘starlight’ focus mode when investigating autofocus fine-tune. These two modes use contrast-detect instead of phase-detect, and therefore ignore the fine-tune calibration. If your target is quite small, you might want to use the single-point focus mode, which is very selective and still uses phase-detect and allows AF-C focus.
Another caution: if you’re using F-mount lenses and you have more than one FTZ adapter, you’ll have to pay attention to which adapter is in use. Teleconverters can also mess up the fine-tune calibration.
The Nyquist Limit
According to Edmund Optics , the absolute limiting resolution of a sensor is the Nyquist Limit which is half of the sampling frequency (pixels/mm). For the Nikon Z9/Z8, for instance, the Nyquist limit is (8280/35.9)/2 pixels per millimeter (231/2 pix/mm) horizontal or (5520/23.9)/2 pixels/mm vertical, or again (231/2) pix/mm.
The Nyquist limit of (231/2) lp/mm is then 115.5 lp/mm.
The typical actual frequency response includes what’s called the “Kell Factor” named after Raymond Kell, which is conservatively set to 0.7. This value helps compensate for the physical space between the light-sensitive portions of the pixels, and to avoid patterns similar to the moire effect. This means the practical actual sensor frequency response limit is now (115.5 * 0.7) or 80.85 lp/mm.
Given this information, any measured resolution above 80.85 lp/mm is unreliable. In any event, the Z8 average of 77.1 lp/mm with this 500mm f/5.6 PF Nikkor is darned near the Nyquist limit.
Summary
It turns out that Nikon included the AF-tune feature in their Z cameras for good reason. It’s not useless and ignorable after all. I was able to squeeze out a little bit better focus accuracy (and resolution) from my lens. The resolution increase was only about 4 percent, but I’ll take it.
The Nikon Z8 won the focus fight over the D850 after all. It’s no longer feeling embarrassed by the D850.
Comments