Struggling with Z8 focus tracking

If you would like to post, you'll need to register. Note that if you have a BCG store account, you'll need a new, separate account here (we keep the two sites separate for security purposes).

Using the 180-600 seems to be adding another variable into the mix. The 180-600 probably focuses slower due to the large focal range and design differences. Still, one would hope that once focus is acquired focus would be maintained.
 
One feature I would like to see Nikon add to the Z firmware is focus in DX but record in FX mode. It seems that the focus algorithm with SD works better with larger subjects when in DX mode all other things being the same. I understand that you would not be able to confirm composition BUT it might greatly improve keepers in certain situations. It might also allow the user to keep the subject under the focus point better. Anyone else think that would be useful?
 
Very strange, shooting a lot of sports I can see how frustrating this is and it seems like the camera is not performing as it should, at least not like my Z9 or even my z6iii is actually performing. So a few more random thoughts....I may be reaching here but you never know.

I would stop using BBF and just set the shutter release to focusing. My thought is that maybe there is some issue with the AF-On button or maybe even a conflict with the shutter release and the AF-On button.

I would check and clean all camera and lens contacts, maybe just a poor contact is causing issues.

I would try different batteries, maybe there is just a voltage issue that caused focus motor problems.

I know these are reaching, but stranger things have happened!

I have Reikan FoCal pro version and I would run it through a focus calibration process with the 70-200 and then run a focus consistency test with the 70-200. It would be interesting to see what type of focus graph the camera and lens combination would create.

I have used my 180-600 for outdoor sports a lot and it has performed very well. It does not focus as fast as the 70-100 but for most situations that has not been an issue. I am shooting a softball regional today and I will shoot most of it with the 180-60.
 
I just spent two hours with my Z8 and 70-200S testing the AF system outdoors, chasing birds along the nearby coastline. My experience was similar to what you (OP) initially described. Keep in mind, while I have considerable experience shooting theatre performances in low light conditions, I’m not an expert in wildlife, bird, or fast-paced sports photography.

All test shots were handheld, using BBF, AF-C, Matrix metering, 20 FPS continuous, Auto ISO, and shutter speeds of 1/3200, 1/4000, and 1/5000, with VR set to sport and off. Camera has the latest FW installed.

I tested a variety of AF modes, including handoff: wide small, wide large, custom, 3D, and full auto with SD set to 90% birds and 10% auto. I also used dynamic small, dynamic medium, and single point (obviously, no SD was involved in these). I experimented with short and long bursts, and spent time simply tracking subjects in the viewfinder to observe the AF system’s behaviour.

In a nutshell, the more control you give to the camera's brain, the more mistakes it makes—often losing the subject and wandering off target. The most success I had was with the 1x1 custom AF box with SD (bird) enabled, followed by single point, dynamic small and a custom 5x3 box with SD enabled (in that order) which is almost identical to my theatre performance shooting experience.

Dynamic small tends to wander off target easily if something with stronger contrast and/or closer in distance enters under the outer helper points. Also, the focus point/area is slightly larger than its borders displayed in the viewfinder, which can lead to an unexpected behaviour. For example, outer points can latch onto other objects with higher contrast, even if they barely touch the focus area. In theory, this shouldn't happen since the center point of the dynamic area is supposed to take priority, but real-life experience suggests otherwise. It seems only Nikon knows the exact algorithm that determines how much weight is given to the center point in each scenario.

Custom 5x3 behaves very similarly to Dynamic Small when it comes to latching onto something other than the intended target. While it can help keep the focus area on a moving subject, I found more success using this mode in combination with a handoff to the 1x1 custom area to improve accuracy.

Wide small, wide large, 3D, and auto AF (on its own) are hit and miss. Sometimes they lock on and track the subject well, but other times they struggle to achieve an initial lock, and they often lose the subject between frames (though this can be mitigated to some extent with 20 FPS). The best way to use them is in combination with handoff to other AF modes and longer bursts.

Overall, I believe there’s nothing wrong with your Z8 (or mine). The AF system is far from perfect for sure.
 
I would stop using BBF and just set the shutter release to focusing. My thought is that maybe there is some issue with the AF-On button or maybe even a conflict with the shutter release and the AF-On button.
Interesting idea, but what are the odds of seeing the same behaviour if Fn 1/2 (or other buttons) are programmed to AF-ON + AF Area? Without this setup, he wouldn’t be able to do the handoff.
I would check and clean all camera and lens contacts, maybe just a poor contact is causing issues.


I would try different batteries, maybe there is just a voltage issue that caused focus motor problems.
Even though I believe there’s nothing wrong with the camera, an old-school hard reset wouldn’t hurt. Step 1: reset the camera to factory settings and remove the battery for several hours or overnight.
 
Using the 180-600 seems to be adding another variable into the mix. The 180-600 probably focuses slower due to the large focal range and design differences. Still, one would hope that once focus is acquired focus would be maintained.
In my experience, the 180-600 focal range has little or no bearing. Tracking is not a problem for it, even though the AF motor is a bit slower than the VCM in, say, a 400 f2.8. The f6.3 aperture between 500 and 600 mm doesn't seem to effect tracking much, if at all, but initial focus acquisition on fast birds can be more problematic. This is why I am so desperate for that Z9 firmware update we were expecting.
On dslrs, dynamic 9 was my goto BiF AF mode, but I can't for the life of me get comparable results on the Z9.
The outer points we always better performers when the background had a very low detail frequency ie: sky. So contrast was a big contributor to helping maintain focus distance and tracking acquired with the central AF point.
D9 was the same as AF point surround on Canon, and both functioned very well in pretty much all light scenarios.

But with the Z9, and presumably the Z8, everything has been turned on its head to some degree or other. SD can sometimes help, and be a pain in the bum at other times.

Again, IMO, focal length plays a big part from the PoV of 'how big the subject is in the frame'. The subject really needs to occupy 20-30% of the frame minimum, especially when allowing the camera to have a big choice of AF points. Certain AF modes have their own locked in a3 variations, so if you play with those settings, you might not always be getting what you think!
 
Very strange, shooting a lot of sports I can see how frustrating this is and it seems like the camera is not performing as it should, at least not like my Z9 or even my z6iii is actually performing. So a few more random thoughts....I may be reaching here but you never know.

I would stop using BBF and just set the shutter release to focusing. My thought is that maybe there is some issue with the AF-On button or maybe even a conflict with the shutter release and the AF-On button.

I would check and clean all camera and lens contacts, maybe just a poor contact is causing issues.

I would try different batteries, maybe there is just a voltage issue that caused focus motor problems.

I know these are reaching, but stranger things have happened!

I have Reikan FoCal pro version and I would run it through a focus calibration process with the 70-200 and then run a focus consistency test with the 70-200. It would be interesting to see what type of focus graph the camera and lens combination would create.

I have used my 180-600 for outdoor sports a lot and it has performed very well. It does not focus as fast as the 70-100 but for most situations that has not been an issue. I am shooting a softball regional today and I will shoot most of it with the 180-60.
Reikan is a whole other issue for me right now. When I sign into the app it says I have a free license only and won't let me do sny tests. Yet elsewhere the app itself acknowledges I have a pro license and if I try to buy a license it says I can't because I already have one.

It's been like this for a couple weeks. I've contacted support twice and haven't gotten any reply. I was honestly beginning to wonder if the company went out of business or something.
 
One feature I would like to see Nikon add to the Z firmware is focus in DX but record in FX mode. It seems that the focus algorithm with SD works better with larger subjects when in DX mode all other things being the same. I understand that you would not be able to confirm composition BUT it might greatly improve keepers in certain situations. It might also allow the user to keep the subject under the focus point better. Anyone else think that would be useful?
Now THAT update is something I'd actually pay money for!
 
A lot of suggestions are flying around and the more that I think about this, particularly in context of my long tenure in photography, the more I am reminded of a similar experience that I used to have with my Canon equipment, namely the 7d and its successor the 7dii. For some reason, I struggled mightily with that camera in terms of af consistency. I recall receiving all kinds of advice (none of which worked), sent the camera back to CPS twice without any changes in performance and ultimately sold the bodies. Interestingly, I had performed some extensive testing with the bodies, namely set up targets in good lighting, and demonstrated the af inconsistencies. I provided this documentation with images and video to CPS, who simply looked at the bodies, information I had provided, and returned them as being "in spec". Perhaps, it is time for you to perform some static testing by setting up a manequin in a well lit, real life setting and perform some experiments. First, shoot it in AFS with SD on the eye. Defocus the lens each time and let it reacquire focus. Look at the images and see if it captures the eye or some other part on the face (mine tends to wander sometimes capturing the lashes, brow, etc.). Next, try the same thing in AFC with the various AF modes, dynamic small, medium, AA, Wide area, SD on/off. Let us know what you find. I may do the same thing with my bodies.
 
A lot of suggestions are flying around and the more that I think about this, particularly in context of my long tenure in photography, the more I am reminded of a similar experience that I used to have with my Canon equipment, namely the 7d and its successor the 7dii. For some reason, I struggled mightily with that camera in terms of af consistency. I recall receiving all kinds of advice (none of which worked), sent the camera back to CPS twice without any changes in performance and ultimately sold the bodies. Interestingly, I had performed some extensive testing with the bodies, namely set up targets in good lighting, and demonstrated the af inconsistencies. I provided this documentation with images and video to CPS, who simply looked at the bodies, information I had provided, and returned them as being "in spec". Perhaps, it is time for you to perform some static testing by setting up a manequin in a well lit, real life setting and perform some experiments. First, shoot it in AFS with SD on the eye. Defocus the lens each time and let it reacquire focus. Look at the images and see if it captures the eye or some other part on the face (mine tends to wander sometimes capturing the lashes, brow, etc.). Next, try the same thing in AFC with the various AF modes, dynamic small, medium, AA, Wide area, SD on/off. Let us know what you find. I may do the same thing with my bodies.

Oh, I've done this extensively. Of the 275,000 shots on this camera, I wouldn't be surprised if 1/10 to 1/6 of them were from doing these kinds of tests trying to figure out what the heck was going on. There were four threads about this on DPReview last April/May about this where I was trying to sort out issues similar to that fashion photographer posting over there right now and one of the more technically inclined users set up a bunch of tests of his own trying to do the same thing. He largely got similar results.

What I found was single point AF in AF-S would be generally fairly consistently accurate - not perfectly, but enough that I would never have complained or noticed a problem if that's all I used. Subject detect in these modes was a downgrade in reliability, but was still pretty good.

In AF-C, subject detect (using ANY of the subject detect modes/box sizes/3D/etc.) would at a relatively high rate (maybe 20%) say it had focus with the green indicator on the eye when it did not. I'm not even counting "eyelash focus" here, which I considered close enough to count as a hit, but focus on literally some other focal plane inches in front or behind the eye. Most of the time, it was actually that it would focus on the other eye than the one it said it was. So, if someone was turned slightly, the focus indicator would be on the front eye but the focus plane would be on the rear eye, making the photo look quite out of focus. Sometimes, though, it wouldn't even be on the rear eye but would be on some other focal plane which didn't correspond to anything in the photo.

As light went down, this problem became more pronounced. At or above EV 7 or 8, I got the ~20% focus on random planes. At EV 5 or 6, it happened much more, and at EV 3 or 4 it was happening most of the time.

This whole ordeal was one reason I started using dynamic area, because in those same tests dynamic area would usually be focused on what it said it was. It wasn't perfect (maybe 85 - 90%) but far, far more reliable than subject detect, especially as light went down. I did a bunch of videos demonstrating it at the time. Note that some of these videos were originally accompanied by a link to all the raw files and so the videos themselves may not be as meaningful without that context, but a lot of them will definitely show the behavior clearly.






https://www.youtube.com/watch?v=SRB0boQ5uCQ

https://www.youtube.com/watch?v=nzNVcI6isvE

https://www.youtube.com/watch?v=tswwqAtQBQ0

https://www.youtube.com/watch?v=B3YihxCYVvI

https://www.youtube.com/watch?v=MYLC2-m6cFk

https://www.youtube.com/watch?v=sGaAZXdGZxg

https://www.youtube.com/watch?v=zOXsnKFE5SI
 
Last edited:
Oh, I've done this extensively. Of the 275,000 shots on this camera, I wouldn't be surprised if 1/10 to 1/6 of them were from doing these kinds of tests trying to figure out what the heck was going on. There were four threads about this on DPReview last April/May about this where I was trying to sort out issues similar to that fashion photographer posting over there right now and one of the more technically inclined users set up a bunch of tests of his own trying to do the same thing. He largely got similar results.

What I found was single point AF in AF-S would be generally fairly consistently accurate - not perfectly, but enough that I would never have complained or noticed a problem if that's all I used. Subject detect in these modes was a downgrade in reliability, but was still pretty good.

In AF-C, subject detect (using ANY of the subject detect modes/box sizes/3D/etc.) would at a relatively high rate (maybe 20%) say it had focus with the green indicator on the eye when it did not. I'm not even counting "eyelash focus" here, which I considered close enough to count as a hit, but focus on literally some other focal plane inches in front or behind the eye. Most of the time, it was actually that it would focus on the other eye than the one it said it was. So, if someone was turned slightly, the focus indicator would be on the front eye but the focus plane would be on the rear eye, making the photo look quite out of focus. Sometimes, though, it wouldn't even be on the rear eye but would be on some other focal plane which didn't correspond to anything in the photo.

As light went down, this problem became more pronounced. At or above EV 7 or 8, I got the ~20% focus on random planes. At EV 5 or 6, it happened much more, and at EV 3 or 4 it was happening most of the time.

This whole ordeal was one reason I started using dynamic area, because in those same tests dynamic area would usually be focused on what it said it was. It wasn't perfect (maybe 85 - 90%) but far, far more reliable than subject detect, especially as light went down. I did a bunch of videos demonstrating it at the time:
Yes, I remember those tests well and ideally one would like to take the variables of camera and subject movement out of the equation. That's why I recommended static subjects with the system mounted on a tripod. In many respects, I wished that I had kept my Canon gear so I could perform a side-by-side comparison. It's not that I didn't experience af challenges with those cameras but when they locked on a subject, they stayed on the subject and tracked accurately (R3/R5). They didn't lose the subject, be it a person or animal, mid-action, and for no apparent reason. Likewise, they had no difficulty taking in focus portraits with SD, whereas I am having to take redundant shots with my Nikon gear.

BTW, you are using the latest FW update with the 70-200? I think it is 1.22?
 
Yes, I remember those tests well and ideally one would like to take the variables of camera and subject movement out of the equation. That's why I recommended static subjects with the system mounted on a tripod. In many respects, I wished that I had kept my Canon gear so I could perform a side-by-side comparison. It's not that I didn't experience af challenges with those cameras but when they locked on a subject, they stayed on the subject and tracked accurately (R3/R5). They didn't lose the subject, be it a person or animal, mid-action, and for no apparent reason. Likewise, they had no difficulty taking in focus portraits with SD, whereas I am having to take redundant shots with my Nikon gear.

BTW, you are using the latest FW update with the 70-200? I think it is 1.22?
Actually, I hadn't been aware of the 70-200 firmware update until last week. I did update it to the latest before game the photos from which prompted me to post this thread. Before that it was on the original launch firmware. The camera has long been on the latest firmware.

I can say that between the original lens firmware and the 1.22 I DID notice a difference, but not to what's being discussed here. The 1.22 firmware was supposed to fix issues with focus accuracy for photos taken while the zoom ring was being operated and I can say that I did find that from before vs after that there was slightly more definition on some of the photos taken while zooming. Nothing changed as far as the overall topic of this thread though.
 
Actually, I hadn't been aware of the 70-200 firmware update until last week. I did update it to the latest before game the photos from which prompted me to post this thread. Before that it was on the original launch firmware. The camera has long been on the latest firmware.

I can say that between the original lens firmware and the 1.22 I DID notice a difference, but not to what's being discussed here. The 1.22 firmware was supposed to fix issues with focus accuracy for photos taken while the zoom ring was being operated and I can say that I did find that from before vs after that there was slightly more definition on some of the photos taken while zooming. Nothing changed as far as the overall topic of this thread though.
Just as an off handed test, I validated your observations with the low light level AF challenges involving people. Thanks to my spouse who volunteered for the test, I had her stand slightly obliquely to the camera and using my 70-180 f/2.8 wide open tried AF with SD in both AFS and AFC. The green subject detect box was squarely over the one eye, though the white boxes with the arrows would dance across both. At the time of capture, the green AF box was clearly over the intended eye (which was the far one) and when I reviewed the images, the actual "in focus" area was the contralateral eye even though the EVF and NX studio agree that the AF point was the intended eye. For S&G, I decided to employ my cat, Nali for a similar test under similar conditions. I did not observe the problem. In the past, I've written that the Z8/(Z9?) af has some attendant issues with human eyes that I didn't observe with other animals (not ungulates - that's a different story). This seems to confirm my assertion as well as the Z8's behavior with humans. Also, it seems to cast doubt on the "it must be your body" refrain. Both my Z8's reproduce the exact same results as yours in this test.
 
Just as an off handed test, I validated your observations with the low light level AF challenges involving people. Thanks to my spouse who volunteered for the test, I had her stand slightly obliquely to the camera and using my 70-180 f/2.8 wide open tried AF with SD in both AFS and AFC. The green subject detect box was squarely over the one eye, though the white boxes with the arrows would dance across both. At the time of capture, the green AF box was clearly over the intended eye (which was the far one) and when I reviewed the images, the actual "in focus" area was the contralateral eye even though the EVF and NX studio agree that the AF point was the intended eye. For S&G, I decided to employ my cat, Nali for a similar test under similar conditions. I did not observe the problem. In the past, I've written that the Z8/(Z9?) af has some attendant issues with human eyes that I didn't observe with other animals (not ungulates - that's a different story). This seems to confirm my assertion as well as the Z8's behavior with humans. Also, it seems to cast doubt on the "it must be your body" refrain. Both my Z8's reproduce the exact same results as yours in this test.

Interesting. How to explain all the people who are very insistent that they never have any of this happen?
 
Something I have discovered in some testing this afternoon:

If the subject moves out from under the main point slowly or if the camera moves slowly so that the helper points stay on the subject, it seems to keep focus. If the subject moves quickly, even if it stays under the helper points, it refocuses on the background.

So, if I follow a walking player and briefly lose him, the helper points keep focus on him. A goalkeeper or other player quickly darting, even if I keep the helper points on in the very same way, causes a loss of focus. I now also wonder if there is any way that this related to ajrmd's issues with diving birds.
Have you played around with the focus tracking settings, like change it to erratic mode? This video section (cued to start there) describes the issues you're encountering. Also, keep in mind that subject detection is not a "lock on a specific target", which sounds to me is what you're looking for. It could happen soon with so many AI advances, but for now, it has it's limitations and you'll need to either use a fixed point and pan to track (the old fashion way) or use a custom area AF that's small enough to stay on one individual. As long as your panning stays close on the player you'll help the tracking system.
 
Interesting. How to explain all the people who are very insistent that they never have any of this happen?
I'm not sure how many people actually subject their systems to test it? When I look at the image at normal magnification, it appears in focus. Now pixel peep at 100% and it's all too apparent that the observable focus point was different than the one indicated by the EVF green square and the one recorded in exif.

Quite candidly, I am still trying to better understand Nikon's AF system and Thom Hogan admits that his questions to Nikon yet to be answered. I do recall an older article that he published on his website which indicated that Nikon (Sony) sensors use the blue pixels in the arrays for AF, see: https://www.sansmirror.com/articles/choosing-a-mirrorless-camer/autofocus-sensors.html. Getting back to the portrait issue, if I recall correctly your subjects had blue eyes as does my spouse. I haven't dived into this further to determine whether the OOF issues occur with other eye colors? This would be interesting and might explain why I can't duplicate this phenomenon on my cat (or dog) or why others may not be encountering the same issue. Unfortunately, it doesn't address the (unrelated?) problems you're encountering in sports.
 
Last edited:
Have you played around with the focus tracking settings, like change it to erratic mode? This video section (cued to start there) describes the issues you're encountering. Also, keep in mind that subject detection is not a "lock on a specific target", which sounds to me is what you're looking for. It could happen soon with so many AI advances, but for now, it has it's limitations and you'll need to either use a fixed point and pan to track (the old fashion way) or use a custom area AF that's small enough to stay on one individual. As long as your panning stays close on the player you'll help the tracking system.

I've experimented with all combinations of the a3 setting. One of the major points under discussion here is that even in controlled testing it is very, very difficult to get any detectable difference in behavior with any of those settings.

Let me try to explain what I am looking for in this way: I am looking to be able to acquire focus on a target and then have at least a very small margin for error so that the camera will not immediately refocus if the target moves and the AF box drifts off of the target for something like 1/20 to 1/10 of a second - in other words, a frame or two - before I get it back on. The reason that I have put so much attention on dynamic area mode in this thread is because what I am looking for is literally the entire purpose for dynamic area AF. It's the reason it exists, and what it is supposed to do. The helper points are supposed to temporarily keep focus on the thing you had focus on if it moves from under the center point for a brief moment. The issues has been that in practice, they are not keeping focus for even a tiny fraction of a second. The very instant the center point moves from the target, it is refocusing.

The reason subject detect has come up is that people have said not to use dynamic area and that subject detect is better, but it is also dropping focus in the same way.

3D tracking in particular is supposed to track a particular target, and this is why handoff to 3D tracking was for a while quite popular. I don't expect it to be perfect, either, but it's been if anything even less reliable than the other options, which is saying something!
 
I'm not sure how many people actually subject their systems to test it? When I look at the image at normal magnification, it appears in focus. Now pixel peep at 100% and it's all too apparent that the observable focus point was different than the one indicated by the EVF green square and the one recorded in exif.
I wouldn't say it takes pixel peeping to see the inaccuracy in focus for most of these. The only reason I was zooming in in those video is because it can be difficult to see at the especially small size of the viewfinder I was using when recording the videos. If it were a very minor softness that wasn't noticeable at normal viewing sizes I wouldn't have started most of these discussions.
 
Interesting subject. I haven’t done sports in years and no longer own a 70-200mm lens. Steve’s video about a short f2.8 vs a longer f6.3 is an eye opener. I did notice the two photos of the tennis layer were at different distances and the 180-600 was much closer. I would imagine if you were to crop the 70-200 to that distance the noise would be closer. Also any shot can vary a lot in iso from one moment to the next Due to various reasons.
personally I don’t use the dynamic modes for wildlife. I find them unreliable. As for SD I have a button to turn it on and off. If it were me I would use a 100-400f5.6 and use one of the tighter AF modes with SD off unless very close and then maybe try with SD on. Unless in really low light, then I would want to get closer.
but as I said, I haven’t photographed sports in a while.
 
Interesting subject. I haven’t done sports in years and no longer own a 70-200mm lens. Steve’s video about a short f2.8 vs a longer f6.3 is an eye opener. I did notice the two photos of the tennis layer were at different distances and the 180-600 was much closer. I would imagine if you were to crop the 70-200 to that distance the noise would be closer.

In the case of the two particular photos you're talking about, that's true. These were taken in decentish light. The thing that I've tried to point out a few times, but stuff naturally gets lost in long threads like this, is that for the HS sports I am working on right now the lighting is often very poor and when in the more poor light I have done the 70-200 side by side with the 180-600, cropping the 70-200 shots to match the 180-600, the 70-200 have looked a lot better even though cropping does make noise more apparent.

I think that what it comes down to is that even though perceived noise increases as we crop in on an image, there is still the basic fact of how much original detail was in a shot in the first place. Denoising can clean up noise, but it can't really restore detail that has was never there in the first place. A lot of these autumn HS events are starting at 4:30 or 5:00 so that the first half hour or hour has decent light and I can use the longer lens, but once it hits 5:30 or so it's dark enough that the 2.8 aperture is allowing more detail to be captured vs. the overall noise level so that even when I crop in I am winding up with something that looks better than what the 180-600 gives me.

That tennis shot was taken when the light was still okay. Later ones from the same day I did side by side with the 180-600 and the 70-200 and the 70-200 cropped in looked a lot better.

In any event, the primary concern was AF performance and I was able yesterday to test and found that using the 180-600 and getting tighter frames on the subjects didn't improve the AF stickiness at all.

Also any shot can vary a lot in iso from one moment to the next Due to various reasons.
personally I don’t use the dynamic modes for wildlife. I find them unreliable. As for SD I have a button to turn it on and off. If it were me I would use a 100-400f5.6 and use one of the tighter AF modes with SD off unless very close and then maybe try with SD on. Unless in really low light, then I would want to get closer.
but as I said, I haven’t photographed sports in a while.

I only started using the dynamic mode because I found the subject detection modes unreliable, though I will say that I find them a lot more reliable for wildlife than people. I think ajrmd said something similar in his recent posts. I also have a button to deactivate subject detection but the problem for sports is that even without subject detection the closest subject priority of those area boxes causes problems with so many different people out there on the field.
 
Well, this afternoon after watching some CFB (unfortunately, I've lost my credentials and am no longer shooting on the sidelines) I walked down to my local pond with one of my z8's and 70-180 f/2.8. There are a pair of Trumpeter swans who I befriended years ago when they took up residence and used to nest in the area behind my house before the local developers drained it (my apologies for the sidebar). Anyhow, the Trumpeters know me and I know them so they were willing to pose close up for a nice shoot. Anyhow, I started in AA, SD-birds, AF-C with the swans about 6-10 feet away and observed something very interesting. The little white tracking box in the EVF recognized the swan's eye right away though when I pressed the BBF, a large green box would appear on most of the abdomen. If I released the BBF, the small white box immediately appeared again back on the eye. We played this game a couple of times with exactly the same result. Next, I switched to Wide S and moved the red box near to the eye. The little white box jumped to the eye and the green box remained there as well, most of the time, though occasionally, the large green box would appear again over the abdomen. I wish that I had a recorder with me to demonstrate the behavior, though it is very strange to say the least.

Unfortunately, Harold and Sophie decided they had enough of the portrait session and headed back to the water, so I wasn't able to test the performance/behavior of the other AF modes.

It's not that one can't make incredible and sharp images with this camera. I happen to like this one where she was doing some head bobs and the light was solid. If I remember correctly, this was shot at 85mm and she was perhaps 6-8ft away (I can't confirm in exiftool because for some reason it is no longer reading distances since the FW upgrade). I shot this in AF-C, SD-bird, Wide small because in AA (and yes, this close, the evil green box would appear over the abdomen even though the SD white box was following over the eye). YES, EVEN THIS CLOSE WITH THE SUBJECT THIS LARGE IN THE FRAME (apologies for shouting).

As I mentioned, for these subjects and this setting, AA was useless and Wide S seemed to generally work well, though as I mentioned it too would occasionally lose capture and revert back to the abdomen - see the highlighted screenshot of the series below.

There is something not right with this AF system on so many levels, and I have never seen anything like this with either Canon or Sony.

TS0000.jpg
You can only see EXIF info for this image if you are logged in.

Screenshot 2024-10-19 213403.jpg
You can only see EXIF info for this image if you are logged in.
 
Last edited:
3) I'd love to invest in better equipment, but there are financial considerations here and I think a lot of it can come down to the "tier" of work being done. I forget if I mentioned it earlier in this thread or if it was on another forum, but I'm the only one I ever see at these games that has even brought a lens longer than 70-200. I bring my 180-600 even if I use it much less frequently than the 70-200. The other professionals at these high school games just come with a 70-200. It's very much the norm. I'm not saying that this means I shouldn't use better equipment just because others are not: I'm rather simply saying that the equipment I'm using seems to be the norm, at least around here, for people doing this type of work (vs. professional or college level stuff) and dealing with the financial considerations involved.

I shoot high school football regularly and in our area (southern California) many of the traditional pro's still use the 400 mm f2.8. One of the guys who shoots Nikon uses the 400 mm with the built in teleconverter. In addition to football I shoot girls flag football, both at night as well as basketball in very poorly lit gyms with the Z8 and the 70-200 f 2.8 and unfortunately my experience with the auto focus is similar to yours.
 
I shoot high school football regularly and in our area (southern California) many of the traditional pro's still use the 400 mm f2.8. One of the guys who shoots Nikon uses the 400 mm with the built in teleconverter. In addition to football I shoot girls flag football, both at night as well as basketball in very poorly lit gyms with the Z8 and the 70-200 f 2.8 and unfortunately my experience with the auto focus is similar to yours.
What have you found to be the most successful?
 
I encounter this problem with wildlife frequently, but doing more sports now it's really becoming a huge problem in costing me a lot of key shots I should be getting and it's giving me a lot of clear examples that I can use to demonstrate and ask about this.

There are at least three ways of tracking a subject with the Z system: Wide area with subject tracking, 3D AF, and dynamic area AF.

Wide area with subject tracking has proven to be the most reliable when it works, but doesn't always work and sometimes isn't right for a situation. If the wide area AF finds a subject, it seems to track it reasonably well. However, I also find it to be the most difficult to use to track a subject in a chaotic environment like sports and it's the mode that most takes control out of my hands and makes me rely on the algorithm identifying a subject, which doesn't always work.

3D AF is supposed to, when used with subject tracking turned off, hang onto a subject based in part on color. Seems like it should be a good option for situations like this, where I want to follow a very distinctly colored subject in an environment with lots of distractions. It's not. It almost always tracks for a few frames and then goes to the background. Note that in this case the 3D tracking had a more solid lock initially then you see here. Just in case anyone is unaware, 3D tracking overrides the camera's A3 setting for focus tracking with lock on.


Dynamic area modes are in theory a tried and true method. They give a central AF point which is essentially a single point AF and should function as such if you keep the subject under it. There are then a series of "helper points" around the outside so that if the subject briefly moves from the central point, the camera is supposed to maintain focus as long as you it remains at least under the helper points.

In practice, I have found that this doesn't work almost at all, either in real-world-use or in testing. I am shooting in AF-C at 20FPS most of the time and always the very first frame when the central point moves off of the subject the camera immediately refocuses on the background. I have tried every combination of settings in A3: from a quick response to a delated response, steady or erratic, and it behaves exactly the same way regardless: it immediately swaps focus. There is no delay even with A3 set to 5.

Some examples of this follow. Before looking, it's worth a quick note that some people say that the AF point displayed by NX studio is not always the real point that was used, but to that I would say 1) We only have to go off of what we have to go off of, 2) I have been paying very close attention to this lately when in the field and the reported points agree exactly with my first-hand, live experience, 3) this is in AF-C with old-fashioned, non subject-detect AF modes which really should eliminate the vast majority of that concern.


These are a few examples of many, many more where the same thing happens: I'll have the dynamic area AF point on a subject with focus, then the subject moves and the AF point moves off of them for as little as 1/20 of a second up to around 1/2 of a second - and sometimes not even off of them completely, but only partially off of them - and the camera immediately focuses on the background or some other thing. I've also done a lot of this with more controlled testing where I have some subject moving according to my choice rather than trying to follow a bird or a player or something and see the same thing.

Does the setting A3 for adjusting the focus t racking delay make a difference? I already noted that it does not, but I will add some nuance to that now: if I get a subject very large in the frame, there is a difference. If I get something that fills the entire height of the frame, for instance, I can put the main AF point on them and then move so that they're now under a helper point and with A3 set to 1 (quick) it will refocus almost immediately whereas with A3 set to 5 (delayed) there is a very nice, healthy delay before it refocuses. In fact when doing this there is even a nice little indication: the AF point turns briefly from green to red right before it refocuses.

This suggests to me that the problem here is that in some or most of these cases my subject isn't being identified as a subject by the AF which is why it's not operating as expected. I could understand this in some extreme cases with a very small subject, but in a lot of these cases we're talking about subjects which are a pretty reasonable size in the frame or a composition which a photographer might even be specifically trying to achieve. For instance, I have a hard time thinking that something like the below is just not good enough for the system to work right.

What am I missing here?


View attachment 99430
You hit the nail on the head with this comment: Your quote: " if I get a subject very large in the frame, there is a difference. If I get something that fills the entire height of the frame, for instance, I can put the main AF point on them and then move so that they're now under a helper point and with A3 set to 1 (quick) it will refocus almost immediately whereas with A3 set to 5 (delayed) there is a very nice, healthy delay before it refocuses. In fact when doing this there is even a nice little indication: the AF point turns briefly from green to red right before it refocuses".......you must remember that the camera does NOT know what the subject is and unless you fill the frame with as large of a subject as you can, then the AF sensor may think some other object in the frame is the subject. I have shot thousands of sports images with my Z6 and Z6ii and I use dynamic area AF exclusively. So long as I can keep the center focus box on my target, my keeper rate is >90%. But as I am getting older now, in the past 6-8 months, it is more difficult to track a moving subject when they are jumping around. Any and all of the oof images are 100% MY fault because I can see where my focus target box ended up on the background. This is strictly my fault for not following the subject continuously, even when not actually capturing the shots.

So long as you keep practicing with dynamic area AF with longer focal lengths to fill the focus box, you'll nail this technique down like I did--seriously. I would have no trouble whatsoever tracking either of those 2 guys above no matter where they went
 
Back
Top