Questions regarding Nikon Z8 Eye AF - animals

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).

Status
Not open for further replies.
Hi Steve, thanks for your explanation and for confirming to a degree what I am observing. The Z8 (Z9) does a heck of a job recognizing and tracking eyes on people, animals,and static birds (even long necked ones). In fact, I’ve found it gets fooled less than my R5/R3 on certain static birds such as a GBH, where the R5 frequently jumps to the neck. Nonetheless, it struggles with birds who are flying at obtuse angles or those flying directly at the observer. For the later, it is easier to place a small AF area over the face but how to you manage a Hawk or Eagle swooping out of a tree or a Sandhill landing at an angle where the head position is always changing in the frame? Again, my Canon and Sony bodies didn’t struggle in these scenarios even when they were operating in full arrays.

Do you think Nikon is working on improving these algorithms for BIF? The static ones seem really good and I have few problems with waterfowl or other birds swimming and the airplane/car settings are fantastic.
Of course they are working on it. Nikon has already done to major firmware updates to the Z9 which made the Z9 a significantly better camera. I’m sure we will see more firmware updates to the Z8 and Z9.

One of my favorite things about the Z8 and Z9 is the programmability of different buttons. If I’m not getting focus with the mode I am in I can easily go to wide s as Steve said or I can go to dynamic small and get the shot. I simply push a single button to go to either one. In my case it’s the function 2 button to get to dynamic area and AF on. Fn 1 button is set to single point.

Every situation is different and one focus mode may not do everything you need it to be. That’s why there are several modes to fit the specific needs to get the best shot available.
 
What I find particularly intriguing is how the system, which has world class recognition of the eyes on static/perched/swimming subjects, fails completely when the same structures take flight. Is Nikon truly unaware of these comparative deficiencies and at the same time it would be interesting to understand the engineering challenges.
 
Of course they are working on it. Nikon has already done to major firmware updates to the Z9 which made the Z9 a significantly better camera. I’m sure we will see more firmware updates to the Z8 and Z9.

One of my favorite things about the Z8 and Z9 is the programmability of different buttons. If I’m not getting focus with the mode I am in I can easily go to wide s as Steve said or I can go to dynamic small and get the shot. I simply push a single button to go to either one. In my case it’s the function 2 button to get to dynamic area and AF on. Fn 1 button is set to single point.

Every situation is different and one focus mode may not do everything you need it to be. That’s why there are several modes to fit the specific needs to get the best shot available.
That's the technique I employ for perched subjects namely switching AF zones (spot, dynamic L/S, custom, 3D) as the situation dictates. Now, please explain how to do this when the subject is rapidly moving, and head position is changing dynamically in the frame over the course of a second or two (as in the landing sequence)?

I appreciate that these are the same challenges that the engineers face though they are using brute force calculations to do things which the hand cannot and that's why we rely on automated systems. For me, the issue is both Canon and Sony have largely figured out how to accomplish this, it is apparent that Nikon has not (as of yet).
 
That's the technique I employ for perched subjects namely switching AF zones (spot, dynamic L/S, custom, 3D) as the situation dictates. Now, please explain how to do this when the subject is rapidly moving, and head position is changing dynamically in the frame over the course of a second or two (as in the landing sequence)?

I appreciate that these are the same challenges that the engineers face though they are using brute force calculations to do things which the hand cannot and that's why we rely on automated systems. For me, the issue is both Canon and Sony have largely figured out how to accomplish this, it is apparent that Nikon has not (as of yet).

A lot of us manage to get great in focus shots, so I think nikon has figured it out.

I can also point out Canon struggles with af pulsing on some models, and Sony (per reports on this forum) apparently can't always focus when a bird is over water?

No manufacturer is perfect but if you find nikon can't live up to your expectations (reasonable or otherwise), then swap systems to one that will.
 
That's the technique I employ for perched subjects namely switching AF zones (spot, dynamic L/S, custom, 3D) as the situation dictates. Now, please explain how to do this when the subject is rapidly moving, and head position is changing dynamically in the frame over the course of a second or two (as in the landing sequence)?

I appreciate that these are the same challenges that the engineers face though they are using brute force calculations to do things which the hand cannot and that's why we rely on automated systems. For me, the issue is both Canon and Sony have largely figured out how to accomplish this, it is apparent that Nikon has not (as of yet).

It seems that most users of the Z8/9 (at least for action/wildlife) have multiple buttons programmed to activate different AF modes for instant switching. For example, some people have the DISP button - rarely used for its default purpose - set to AF-ON single point, meaning that pressing this button will instantly start AFing with single point AF. Others have the Fn1 button set to do the same. Lots of people have the Fn1 button set to activate 3D tracking, so regardless of what AF mode the camera is in it will immediately start 3D tracking if pressed. I have my movie record button set to turn of subject detection so if the camera is grabbing the neck of a heron or something as a subject even though my AF point is on its head, I can press this and it will stop doing that and revert to focusing on the head/eye where my AF point is.

These are very easy, efficient, and useful and most people would be using them regardless of how well they thought the AF worked. In fact, users of higher end Nikon DSLRs have long programmed buttons for similar purposes because the cameras allowed it and it is very helpful to be able to instantly start using a different AF mode depending on circumstances.
 
It takes practice to keep a bird in flight in the center of the frame. Ive had my Z9 since January of 2022 and have a lot of practice. The same principles apply with the Z8 except you don't have quite as many programmable buttons. I'm still learning the Z8 as I have only had it a couple of months and just havn't shot that many BIF with it. With practice you can switch between focus modes and toggle from Fx to Dx modes very quickly and yes in 2 seconds. However, rarely do i stay on a BIF for just a second or two. Usually it is for 5 to 10 seconds with short bursts while I am tracking. I use 3d in some cases and have it programed to my joy stick while I keep the camera in Wide L for my BBAF. If a bird is flying towards me I nearly always hand off from Wide L to 3D.

Frankly, I didn't know the part about long necked birds being harder to focus on but it makes a lot of sense to me with the experiences I've had shooting both GBH's and Cranes.

These pictures below are all from my Z9 since I purchased it. Most over the past year as these were from April of 2022 thru current. I know on a couple of these shots the camera focused on the eye because I remember looking at the picture on the camera to see where the camera focused and was amazed the box was right on the eye. The Turkey Vulture and the Nighthawk getting a drink. I haven't shot enough with the Z8 to know if it is as good as the Z9 for BIF but my trip to Pensacola made me more confident in the camera. Only took the Z8 on this trip.

DSC_3209.jpg
You can only see EXIF info for this image if you are logged in.
DSC_3018-Enhanced-NR.jpg
You can only see EXIF info for this image if you are logged in.
Z9W_4680.jpg
You can only see EXIF info for this image if you are logged in.
Z9Z_9032.jpg
You can only see EXIF info for this image if you are logged in.
Z9Z_3360-2.jpg
You can only see EXIF info for this image if you are logged in.
 
DON'T TRUST THE RED SQUARE !! Your third image makes that obvious. If that square was accurate your entire bird would be out of focus. Trust your eye in looking at the image at 100%. I haven't used the red square since my last DSLR. The first and time I used it was the last time. Your first images to me look relatively sharp, but grainy and soft because of the lighting.
 
Hi Steve, thanks for your explanation and for confirming to a degree what I am observing. The Z8 (Z9) does a heck of a job recognizing and tracking eyes on people, animals,and static birds (even long necked ones). In fact, I’ve found it gets fooled less than my R5/R3 on certain static birds such as a GBH, where the R5 frequently jumps to the neck. Nonetheless, it struggles with birds who are flying at obtuse angles or those flying directly at the observer. For the later, it is easier to place a small AF area over the face but how to you manage a Hawk or Eagle swooping out of a tree or a Sandhill landing at an angle where the head position is always changing in the frame? Again, my Canon and Sony bodies didn’t struggle in these scenarios even when they were operating in full arrays.

Do you think Nikon is working on improving these algorithms for BIF? The static ones seem really good and I have few problems with waterfowl or other birds swimming and the airplane/car settings are fantastic.
Nikon is always working on firmware :)

Look how far the Z9 has come since 1.0.
 
A bit on and off topic. Today I spent some time getting to know Fast Raw Viewer and there are a few features that would help with this topic. You can see both fine and edges details - similar to focus peaking. There are other reasons for using this program for culling as well, including showing your over or under highlights and shadows.

I haven’t put any bird images in it today as I was just learning the navigation and features.

Lightroom doesn’t offer these features. So I am going to continue to experiment with it. You can do Lightroom ratings from within it that will carry over to your import.
 
i think the zf is probably the writing on the wall for the expeed6. my take is if they thought it was worthwhile/possible to get a major upgrade in the 6, they probably would have done it there. of course never say never
I think so too…there are dual chips in the II models but the speed of the sensor and writing hardware is likely the limiting factor and the AF in the Expeed 7 models requires the more powerful chip. They could put a 7 in the Z7III…but it costs more and all the other hardware would need the upgrading as well and then it doesn’t meet the price point and profit margin requirement.
 
Whew! Just scanned this whole thread to date.

My observations based on data from the first page by the OP and that no definitive explanation has been so far identified.

I am presuming that all pictures were taken handheld.

In the case of stationary subjects, the subject image will remain stationary at a fixed position in the viewfinder. So will the background.

For a moving subject, well panned, the subject image will remain stationary at a fixed position in the viewfinder. The background will be moving.

The degree of "stationaryness" will depend upon the focal lens's focal length, with longer lengths being more difficult to hold steady. My experience is that moving subjects are harder to hold steady in the viewfinder than stationary ones.

What is the difference between a BIF against a featureless background, well panned, and a stationary bird? Is it reasonable to assume that if the camera is focused on the eye, there should be no difference. After all, the eye, in the limit, should not be moving (within the viewfinder) in either case. Of course, the bird's shape will be changing in the latter case.

I think that your success with stationary bird shots relative to BIF is a big clue.

Ideas for BIF:
-Track bird as early as possible to allow more time for eye detection and lock-on.
-Experiment with "Erratic" focus setting (A3) and AF response time
-Experiment with VR modes, including VR off
-Shrink the focus area as small as your "stationaryness" will allow. For me, this takes a lot of practice.
-Try transitioning to 3D once the eye is detected
-Go for smooth backgrounds through composition or depth-of-focus

I hope this gels some ideas. If not, at least we have added a new word for something all photographers need...Stationaryness.
 
After going through this thread I did a test with the Z9 and Z100-400 with 1.4 TC at max zoom (560mm). I would agree to ignore where the focus square is located on the subject. I do not use Nikon NX so I downloaded it and will removed it. The following images show the focus on the body but clearly the eye is truly focused more than the body.

Untitled-1.jpg
You can only see EXIF info for this image if you are logged in.
2023_09_30_Test-0843-Enhanced-NR-Edit.jpg
You can only see EXIF info for this image if you are logged in.
 
But @ajrmd had shown some example that he made FP stay on BIF from other cameras (Sony?) Does that say more about the issue on the camera, not the user? Can we expect other cameras for this issue too?
Sony and Canon will keep a box on the eye for BIF way more than Nikon will. I'm not even sure if I have a single BIF shot from Z9 that shows eye-detect active for BIF. I don't really have time to go through them all and double check.

I have swallow in flight shots with eye-detect active from the A1. I have many other birds with eye-detect active from R5 and R3. All cameras struggle with long-necked birds, especially one coming towards you so close that the entire bird isn't even in the frame anymore (as seen in the SHC earlier in this thread.

In general the Z9/Z8 require the bird to be way larger in the frame to have subject detect or even AF in general work compared to Sony and Canon.

All brands do better when the photographer can pan smoothly and keep the bird centred in the frame.

The real question I see is whether Nikon's lack of showing eye-af for BIF really affects the final keeper rate or not? I think for most of the time it doesn't really matter and keepers will be sufficient and probably comparable to Sony and Canon. The further the bird is away from you the less it will matter as DOF will cover you. It is when the birds are close that BEAF makes the bigger difference and can easily ruin the shot if the camera chooses the neck, body or near wing to focus on. But even on an A1 (which IMO still has noticeably superior AF to the competition) it will often focus on the near wing of a large, long-necked bird unless you use a very small AF area (like Expand Flex Spot) to place the point on the head of the bird and are able to keep it there.

Final thing I will say is that you can probably find as many images with the indicated focus box in the "correct" spot where the correct spot isn't in focus as you can find images with the indicated focus box in the "wrong" spot but with the correct spot in focus. I don't find that indicated focus spot on any camera to be in sync with reality consistently enough to be worth looking at in post to work out AF problems.
 
Last edited:
The real question I see is whether Nikon's lack of showing eye-af for BIF really affects the final keeper rate or not? I think for most of the time it doesn't really matter and keepers will be sufficient and probably comparable to Sony and Canon. The further the bird is away from you the less it will matter as DOF will cover you. It is when the birds are close that BEAF makes the bigger difference and can easily ruin the shot if the camera chooses the neck, body or near wing to focus on. But even on an A1 (which IMO still has noticeably superior AF to the competition) it will often focus on the near wing of a large, long-necked bird unless you use a very small AF area (like Expand Flex Spot) to place the point on the head of the bird and are able to keep it there.
i do think that nikon and sony have a different _approach_ to how they are drawing the boxes based on their philosophies and how their systems work. i get the feeling that sony kind of does this "this right _here_ is the eye and it's what i'll try to focus on" even if from a dof perspective that level of precision isn't really meaningful, where nikon is more willing to take that into account and basically draw you a larger box so you see that you _don't_ _have_ that level of precision.

maybe that's rationalization, but i've never had the issue when shooting dogs where at comparible distances that i had a problem with focus.

in reality the problems with focus i have are when we get knife-fighting close and dof becomes very thin and the dog is moving in lots of funky ways and then "perfect" focus is hit or miss for all the cameras (based on my experience with the z9 and a1, and from what i've seen of evf footage from the r3).
 
I just red this entire thread.
My take…

1) The OP gets dismissed while he has proof. I understand that reality is hard to face. Instead of pushing back with lectures and blaming the OP, put your self in his shoes, or don’t respond. The pushing back on the OP literally triggered me PTSD from DPR… Everyone deserves the right to express their findings.

2) I have allot of BIF practice under my belt, beginning with the Z6ii with a 70-200+2xTC. Nikon is clearly not there yet with BIF and a box on the eye. Can you get perfect BIF shots? Absolutely! Does the ‘subject eye’ detect work on BIF as advertised? Or as we expect? Hmmmm No… The challenge of so many birds, perhaps the technology isn’t there yet.

3) Nikon will prioritize ‘Perfection’ vs. ‘Close enough’. And this is a big difference between Nikon and Sony. Thom wrote it in his Z9 vs. A1 article. Therefore when you select AF-C release to focus priority, the shutter will give you a run for it… Nikon’s engineer who wrote the code, wants a Japanese perfection before releasing the shutter. Steve’s BIF book makes it clear to set it ‘Release’ otherwise the camera won’t fire on a perfectly focused bird, because it isn’t 100% sure it is a focus… you get my point…

4) I encourage the OP to do what I did with my issue of v4 with sleeping babies. I sent to Nikon all data and findings, RAW files, screenshots, etc. as much as possible, they sent it to Japan, and they fixed it in v4.1

Hope Nikon can figure out a way to get BIF eye detection better. Because all brands are cutting corners on this particular challenging task.
 
I just red this entire thread.
My take…

1) The OP gets dismissed while he has proof. I understand that reality is hard to face. Instead of pushing back with lectures and blaming the OP, put your self in his shoes, or don’t respond. The pushing back on the OP literally triggered me PTSD from DPR… Everyone deserves the right to express their findings.
I thought we all brought up solid points, and didn't 'blame' him for anything, instead asking questions. You might be reading too far into things. And if everyone has the right to express their feelings, then so do we for disagreeing, asking clarifying questions, and providing contradicting evidence.

People even posted images as OP demanded they do (even though the focus point means little to nothing in nx studio, as shown in yet more images). Those posts were ignored largely.
2) I have allot of BIF practice under my belt, beginning with the Z6ii with a 70-200+2xTC. Nikon is clearly not there yet with BIF and a box on the eye. Can you get perfect BIF shots? Absolutely! Does the ‘subject eye’ detect work on BIF as advertised? Or as we expect? Hmmmm No… The challenge of so many birds, perhaps the technology isn’t there yet.
Or people can have unreasonable expectations, especially with smaller birds (ex claiming sony/canon lock on to smaller birds, but does it matter if you're never going to actually use those images). My z9 functions as expected for me at sizes where I will actually use the image. Not perfect, but very good. I also don't rely on 3d as much as others, or auto area. I prefer wide area modes by far.
3) Nikon will prioritize ‘Perfection’ vs. ‘Close enough’. And this is a big difference between Nikon and Sony. Thom wrote it in his Z9 vs. A1 article. Therefore when you select AF-C release to focus priority, the shutter will give you a run for it… Nikon’s engineer who wrote the code, wants a Japanese perfection before releasing the shutter. Steve’s BIF book makes it clear to set it ‘Release’ otherwise the camera won’t fire on a perfectly focused bird, because it isn’t 100% sure it is a focus… you get my point…
I don't understand what your point is here at all. Obviously when you set the option to focus priority it'll only take the image when the camera believes (correctly, or otherwise) it's in focus. That's why it's an option. You can manage to take many great images with focus priority, though I agree release is just the better choice for most of us. So after all that, there really is no difference.

For bonus fun, Thom also notes that Sony eye af is often 'close' but not on the nose (or more accurately... It is). He recently noted it about DPRs a6700 review (https://www.bythom.com/newsviews/autofocus-is-improved.html), but it's come up before. So I'm also not sure this is the road you want to go down re accurate AF.
4) I encourage the OP to do what I did with my issue of v4 with sleeping babies. I sent to Nikon all data and findings, RAW files, screenshots, etc. as much as possible, they sent it to Japan, and they fixed it in v4.1
Your main issue, as I recall, was using 3d instead of a more appropriate mode (custom wide area, etc), but I'm not going to rehash that entire topic as it strays from this one too far. Suffice it to say using the proper area for the job makes your life easier, instead of fighting the system. I'm also not sure they actually changed anything in 4.
1, as my camera acts exactly the same as it did prior in v4. It's also not mentioned in the update notes anywhere. If you can provide more information on what this bug actually was, and how it was fixed, I'd be glad to revise this part of my post.

Hope Nikon can figure out a way to get BIF eye detection better. Because all brands are cutting corners on this particular challenging task.
They're always working on it, but there are always limits to what they can do quickly, and not cause other problems. The key (as many brought up) is if the picture is actually focused correctly, vs what a program claims is the focus point, which is why most of us don't treat this complaint as a serious one. If the picture is in focus correctly, the box doesn't matter because it can be wrong, as noted by many in this topic and shown by others (the hummingbird image is particularly handy for it). Thom also noted this at some point, I believe in a forum post somewhere though I can't nail down a source.

As said by someone else here, the above are just my opinions on this one
 
Last edited:
Sony and Canon will keep a box on the eye for BIF way more than Nikon will. I'm not even sure if I have a single BIF shot from Z9 that shows eye-detect active for BIF. I don't really have time to go through them all and double check.

I have swallow in flight shots with eye-detect active from the A1. I have many other birds with eye-detect active from R5 and R3. All cameras struggle with long-necked birds, especially one coming towards you so close that the entire bird isn't even in the frame anymore (as seen in the SHC earlier in this thread.

In general the Z9/Z8 require the bird to be way larger in the frame to have subject detect or even AF in general work compared to Sony and Canon.

All brands do better when the photographer can pan smoothly and keep the bird centred in the frame.

The real question I see is whether Nikon's lack of showing eye-af for BIF really affects the final keeper rate or not? I think for most of the time it doesn't really matter and keepers will be sufficient and probably comparable to Sony and Canon. The further the bird is away from you the less it will matter as DOF will cover you. It is when the birds are close that BEAF makes the bigger difference and can easily ruin the shot if the camera chooses the neck, body or near wing to focus on. But even on an A1 (which IMO still has noticeably superior AF to the competition) it will often focus on the near wing of a large, long-necked bird unless you use a very small AF area (like Expand Flex Spot) to place the point on the head of the bird and are able to keep it there.

Final thing I will say is that you can probably find as many images with the indicated focus box in the "correct" spot where the correct spot isn't in focus as you can find images with the indicated focus box in the "wrong" spot but with the correct spot in focus. I don't find that indicated focus spot on any camera to be in sync with reality consistently enough to be worth looking at in post to work out AF problems.
"The further the bird is away from you the less it will matter as DOF will cover you. It is when the birds are close that BEAF makes the bigger difference and can easily ruin the shot if the camera chooses the neck, body or near wing to focus on. But even on an A1 (which IMO still has noticeably superior AF to the competition) it will often focus on the near wing of a large, long-necked bird unless you use a very small AF area (like Expand Flex Spot) to place the point on the head of the bird and are able to keep it there."

Precisely. I was flummoxed by the apparent dualism that the camera seemed to have really good eye detect for static subjects and superior algorithms for airplanes/cars, and struggled with BIF compared to my other systems. As you (and others) have observed, for more distant subjects moving 90 degrees in relation to the photographer it matters little what part of the body the AF nails as usually DOF will provide sharp eyes. However, when the subject is moving directly or tangentially towards the observer, eye detect becomes important. When I observed difficulty with achieving eye detect under those circumstances, and after scouring references, and thousands of my own images, I raised the question as to whether the Z8/Z9 uses eye detect for BIF. I gain the impression that some perceived this as a "flame war", though I was trying to understand the AF system and determine why I wasn't being as successful as I was with my respective Canon and Sony systems. For example, my Canon gear's eye detect worked effectively in face+tracking (whole frame) or zone AF in similar situations. What I've learned from various sources is the Nikon system prioritizes the subject > body > head > eye and that using smaller "windows" for AF, namely wide array small versus large may provide better results - it probably forces the camera more rapidly to move through the sequence. Given that Nikon has targeted the mid-price range for high quality, super-telephoto glass (WL), my hope is they will continue to refine their AF system for these cases.

FWIW, I continue to be amazed at the comparative ability of the Z8 on static subjects, particularly ungulates (which both my Canon and Sony systems struggled with). Now, I have to spend some time out west looking for Moose, Elk,...
 
What I find particularly intriguing is how the system, which has world class recognition of the eyes on static/perched/swimming subjects, fails completely when the same structures take flight. Is Nikon truly unaware of these comparative deficiencies and at the same time it would be interesting to understand the engineering challenges.
I think you'll find that Nikon considers it a matter of technique. There are plenty of people who are able to make great images of BIF with the Z8 and Z9. But there are a lot of variables - subject distance, AF Area mode, AF Lock-on settings, pre-focusing, lighting, shutter speed, panning technique, use of IBIS, composition, metering mode, etc. The AF box is independent of actual focusing, so whether in the EVF before the shot, or on the screen after the shot, the only true measure is whether the shot is in focus. The AF box is close, but it's based on contrast in the image just before the shot is taken - effectively an AF-S perspective on a moving subject.
 
I think you'll find that Nikon considers it a matter of technique. There are plenty of people who are able to make great images of BIF with the Z8 and Z9. But there are a lot of variables - subject distance, AF Area mode, AF Lock-on settings, pre-focusing, lighting, shutter speed, panning technique, use of IBIS, composition, metering mode, etc.
See my post #75. The simple setup I describe there essentially rules out each and every of the variables you listed above except for AF Area mode.
And a flying swan fails dramatically compared to a swimming one, for example. If the cam isn't able to find a flying swan eye in a static image, no subject distance, AF Lock-on settings, pre-focusing, lighting, shutter speed, panning technique, use of IBIS, composition, metering mode, etc. in the world are going to help here.
The simple fact is: the thread opener clearly and by all means has a point in his criticism of Z8/9's AF regarding flying birds with long necks (like swans).
Sorry but after 100 posts I find it hard to take even more red herrings thrown around here..
I'm writing this as a more than happy Z8 user!
 
Status
Not open for further replies.
Back
Top