Autofocus Accuracy, Autofocus Consistency and Manual Focus On A Digital Camera; How Does It Compare?

Nubble Light, Maine (click to enlarge)
X-T2, 16-55mm f/2.8 lens @ 22.7mm; 1 sec. @ f/3.6; ISO 200

Recently, when I was testing EVF, shutter and human lag in regards to precisely timing when I can expose an image on my Fuji X-T2, which you can read here, I got the impression that when I autofocused my camera, then switched over to manual focus before pressing the shutter button, the focus would slightly change.  It kind of appeared that way in the EVF, but I wasn't sure.  The resultant images look pretty sharp but were they still perfectly focused?

I did some preliminary tests and found that the focus did not change.  But then I started wondering if autofocus was more accurate and more consistent than manual focus on today's modern digital cameras and electronically controlled lenses, specifically on my gear.

I had three questions that I wanted to answer.  I had made assumptions that the answer to each of these three questions was "yes" and, anecdotally, I think "yes" but I had not done any actual testing to find out for sure.  Here are the three questions:

Does my Fuji X-T2 with various Fuji lenses autofocus accurately 100% of the time?

Does my Fuji X-T2 with various Fuji lenses stay in exact focus when I autofocus, then switch to manual focus, before making an exposure?

Does my Fuji X-T2 with various Fuji lenses manually focus those lenses accurately and consistently 100% of the time?  As a sub-question; Is Focus Peaking better than Magnification (Standard) for attaining accurate focus?

I think so, but I don't know for sure.  The answer I want is yes to all three questions.

So, I set up a small experiment to test answer those three questions.

These tests were made with my Fuji X-T2 and my Fuji lenses.  If you conducted the identical test, your results may differ.  So, this is not written in stone for all camera and all lenses.

First, I tightly taped a flat target with small detail on a vertical wall.  In this case an advertising circular page with very small, but distinct printing.  I planned to set a single focus point, medium size and focus on one small area of the target.  Then I would check focus accuracy and consistency in only the small area on which I focused and not halfway to the edges or at the edges of the image. If you don't understand if your lens is a flat field or a curved field lens and test it for focus and sharpness on a flat surface, you may get partially erroneous results.  Many who proclaim to test lenses make that error and end up with bad test results.  (If you don't understand to what I am referring, read my previous posts here, here, and here to get a full understanding of how lenses are designed and how it is easy to think the edges of lenses may not be sharp but really are.)  Most lenses are curved field and testing them using a flat target will show the center sharp and correctly focused but not so much for the edges and corners.  So, I placed my focusing square directly over the tiny typeface in the center of the image and checked that place only for focus accuracy.

Then, I placed my X-T2 on my very large and sturdy Induro CT414 carbon fiber tripod with a Really Right Stuff B55 ball head and placed the camera squared with my subject. I wanted the plane of the sensor to perfectly match the plane of the test target to ensure accuracy.  For consistency on the target specifically, I would light the target with a bright, incandescent bulb in a reflector and set white balance for that light.  I set the shutter speed to 1/160th of a second, self-timer for a 2-second delay and used the electronic shutter feature to eliminate any camera movement or shutter bounce. Also, to judge focusing accuracy as much as possible, I would set each lens I tested at its widest aperture and, in the case of zoom lenses, to its longest focal length, to minimize depth of field and eliminate any compensation depth of field may have in potentially compensating for inaccurate focus.

I decided to make 20 shots of each test to get a decent sized sample for comparison (will provide a 5% error rate minimum).  In between each exposure, I would de-focus the lens, then either let the camera and lens autofocus or allow me to manually focus. Additionally, I wanted to ascertain if focus peaking or image magnification gave me better assistance in attaining exact manual focus. 

In order to get a good cross section of data I would also try this experiment with more than one lens.  I thought two to three manual focus lenses (older ones as well as one of Fuji's newest) and two to three zoom lenses. Additionally, I thought one test with my 18-55mm f/2.8-4 lens would be interesting for comparison as this is known as a "kit lens" and is not quite up to the level of the pro lenses such as the 16-55mm f/2.8 lens or the 50-140mm f/2.8 lens.  Would there be a difference?  We shall see!

Here are the lenses and settings:

16-55mm f/2.8 lens @ 55mm f/2.8
18-55mm f/2.8-4 lens @ 55mm f/4 (remember it has variable aperture and this is max)
35mm f/2 lens @ f/2
56mm f/1.2 lens @ f/1.2
50-140mm f/2.8 lens @ 140mm f/2.8

For the 35mm lens, the camera was about 4 ft. from the target.  For the three intermediate zooms all tested at 55/56mm, the camera was about 6 ft. from the target and for the 50-140mm f/2.8 lens, the camera was about 10 ft. from the target.

Here are the results.

This is the entire frame.  The red square is the approximate area that fell under the single focus sensor (click to enlarge)

100% crop of sample images of one that is "precisely focused", left, versus the one on the right that is not quite as "precise."
(click to enlarge)
You may or may not be able to see a differnce on the web, but looking at these two in comparison at 100%, there
is a visible difference. The red boxes represnet the area on which the single focusing square in the X-T2 was positioned.  Look at the very center of these two images and see if you can see a difference.  Yes, there is a bit of luminance and contrast difference (I can't explain why as they were both shot and edited identially).  These were made with the terrific Fuji 16-55mm f/2.8 lens @2.8. 

16-55mm f/2.8 lens @ 55mm f/2.8

Autofocus and expose—20 shots.  Does the lens autofocus accurately and consistently?

3 out of the 20 images were not quite as in precise focus as the other 17.  I wouldn't say they were out of focus, but there was a slight visible difference in the sharpness of the typeface.  That being said, I would hesitate to say that any of us could see a difference in the real world.  This is really nitpicking but there was very slight, indeed, but still it was visible.

Autofocus, then switch to manual focus, then expose—20 shots.  After autofocusing, does the lens maintain correct focus after switching to manual focus for exposure?

1 out of the 20 shots was similar to the above.  It was ever so slightly less precise in focus than the other 19.

Manual focus, half with Focus peaking, half with Image Magnification—20 shots.  Can accurate manual focus be consistently achieved and is focus peaking or image magnification more accurate?

All images were in precise focus.  There was no difference between using Focus Peaking and image Magnification (Standard setting in the menu).


18-55mm f/2.8-4 lens @ 55mm f/4 (remember it has variable aperture and this is max)

Autofocus and expose—20 shots.  Does the lens autofocus accurately and consistently?

As in the above autofocus only test, this lens produced 5 out of the 20 that were slightly different than the other 15.  Again, I would characterize this as not quite being in precise focus, yet still focused extremely well.  I'm not sure, looking at a regular image, that one would be able to say it was less than accurately focused, but there was a visible difference.

Autofocus, then switch to manual focus, then expose—20 shots.  After autofocusing, does the lens maintain correct focus after switching to manual focus for exposure?

In this test, I saw differences, as described above, in 3 of the 20 images.  Same situation.

Manual focus, half with Focus peaking, half with Image Magnification—20 shots.  Can accurate manual focus be consistently achieved and is focus peaking or image magnification more accurate?

All images were in precise focus when manually focusing.  There was one anomaly that I didn't see in any other lens.  When manually focusing this lens, rocking the focusing ring back and forth to get precise focus, the image in the EVF would shift in the direction of 10 o'clock to 4 o'clock when turning the ring in one direction and from 4 o'clock to 10 o'clock when turning the ring in the opposite direction.  The IS was turned off so I'm not sure why the lens elements were shifting.  However, this did not affect precise focusing of the lens.

35mm f/2 lens @ f/2

Autofocus and expose—20 shots.  Does the lens autofocus accurately and consistently?

All 20 images in precise focus.

Autofocus, then switch to manual focus, then expose—20 shots.  After autofocusing, does the lens maintain correct focus after switching to manual focus for exposure?

All 20 images in precise focus.


Manual focus, half with Focus peaking, half with Image Magnification—20 shots.  Can accurate manual focus be consistently achieved and is focus peaking or image magnification more accurate?

All 20 images in precise focus.  No difference when using either the Focus Peaking or image Magnification.

56mm f/1.2 lens @ f/1.2

Autofocus and expose—20 shots.  Does the lens autofocus accurately and consistently?

All 20 images in precise focus.

Autofocus, then switch to manual focus, then expose—20 shots.  After autofocusing, does the lens maintain correct focus after switching to manual focus for exposure?

All 20 images in precise focus.


Manual focus, half with Focus peaking, half with Image Magnification—20 shots.  Can accurate manual focus be consistently achieved and is focus peaking or image magnification more accurate?

19 of 20 images in precise focus.  There was one image that was slightly off precise focus. That is a human error, not a camera or lens error.  No difference when using either the Focus Peaking or image Magnification.

50-140mm f/2.8 lens @ 140mm f/2.8

Autofocus and expose—20 shots.  Does the lens autofocus accurately and consistently?

All 20 images in precise focus.

Autofocus, then switch to manual focus, then expose—20 shots.  After autofocusing, does the lens maintain correct focus after switching to manual focus for exposure?

All 20 images in precise focus.


Manual focus, half with Focus peaking, half with Image Magnification—20 shots.  Can accurate manual focus be consistently achieved and is focus peaking or image magnification more accurate?

All 20 images in precise focus.  No difference when using either the Focus Peaking or image Magnification.

There you have it.  I tested some of Fuji's best lenses and a kit lens.  I tested primes and zooms.

It is clear to me, and this is just me, that manual focusing was accurate 99% (1 total images) of the time, but autofocus, over all, was accurate 94% (12 total images) of the time.

This test has given me some valuable information about the focus performance of my gear.  I hope it provided you with the same.

Again, this applies only to me and my particular camera and lenses.  Yours results may vary.

On another note, I found a troubling anomaly that concerns my camera during the testing process.  I will highlight and write about it in the next week or so.

Thanks for looking. Enjoy!

Dennis A. Mook 

All content on this blog is © 2013-2017 Dennis A. Mook. All Rights Reserved. Feel free to point to this blog from your website with full attribution. Permission may be granted for commercial use. Please contact Mr. Mook to discuss permission to reproduce the blog posts and/or images.