z9 z8 AF won't stick to bird head.

[Johan Schmidt]

Improvements to the Auto AF area since I first had a Z9 way back have been nothing but remarkable. My hit rate on kingfishers, in the wild, in flight with Auto AF has been remarkably good even with quite busy backgrounds. Almost as good as an A1 where I would mainly be using zone AF. Given how poor Auto was back in the day Im amazed how much trust I put in it nowadays.

If there is one thing that makes the biggest difference in hit rate with Kingfishers (in flight) is that I put the Z800 PF away and put the Z600 PF on. With the Z800 I find that operator error creeps in more easily and gets exagerrated on small birds. The tracking (of subject in the frame) will be more erratic (for most normal human beings trying to keep the bird steady in the narrower fov) wth the 800 v 600 so the camera could well be made to work harder to maintain focus throughout any burst hence a little bit more AF drift on some frames.

I agree 100% - these are the best cameras that photographers have ever had access to, yet people still complain. It's almost as if they want to skill to be taken out photography - aim and point like an iPhone and it's all done 'automatically' - sure, these tools should assist as much as possible, but there's still the craft that the operator needs to develop. Looking fwd to a new update on the Z9 firmware

 

[Skye.prf]

I agree 100% - these are the best cameras that photographers have ever had access to, yet people still complain. It's almost as if they want to skill to be taken out photography - aim and point like an iPhone and it's all done 'automatically' - sure, these tools should assist as much as possible, but there's still the craft that the operator needs to develop. Looking fwd to a new update on the Z9 firmware

That is true… but not my point… the Z8 is a great camera… but if marketing and some reviewers pretend that Is it on par (AF) with the Z9… it is not the case… very good, of course, but an edge to the Z9.

 

[Kjellis]

I have seen some say here on the forum and on Youtube that the Z9 has an edge to the Z8 when it comes to AF speed and acuracy. I can’t really say I’ve noticed any difference between these cameras when it comes to this. But then I haven’t shoot them side by side to compare them either, so I’m not saying that there isen’t any difference. I’m just curious to learn.

As I shoot only Z-lenses, mostly the 600 TC and 100-400, I wonder is this difference in AF more noticeble with F-mount lenses?

Maybe you have a video in the works about this topic @Steve ?

 

[SoCalRick]

I haven't noticed any difference in AF performance between my Z8 and Z9.

If there is a difference between the Z8 and Z9 in AF performance, what would the cause of that difference be attributed to (since the electronic internals are essentially identical)?

 

[Cameron T]

I haven't noticed any difference in AF performance between my Z8 and Z9.

If there is a difference between the Z8 and Z9 in AF performance, what would the cause of that difference be attributed to (since the electronic internals are essentially identical)?

Different software updates, with different af machine learning amounts/types/etc.

 

[ajrmd]

I have seen some say here on the forum and on Youtube that the Z9 has an edge to the Z8 when it comes to AF speed and acuracy. I can’t really say I’ve noticed any difference between these cameras when it comes to this. But then I haven’t shoot them side by side to compare them either, so I’m not saying that there isen’t any difference. I’m just curious to learn.

As I shoot only Z-lenses, mostly the 600 TC and 100-400, I wonder is this difference in AF more noticeble with F-mount lenses?

Maybe you have a video in the works about this topic @Steve ?

When I shot Canon gear, as I did for many years, the "pro" or flagship bodies always had superior AF to the "prosumer" or secondary tier cameras. It didn't matter if it were the comparison between the 1dmk(?) series and the 5dmk(?) series or even with the respective MILC's R5(?) and R3 (now R1). In those situations, it was always attributable to a greater number of cross style AF points, greater AF sensitivity, independent or unique processors on the flagship bodies, etc. Canon never tried to hide the distinctions though the situation with the Z8 and Z9 are less clear. Do they share the same AF components, logic boards, etc. or are there distinctions? For one, we know the FW is different but beyond that, Nikon's own published comparison suggests that they are similar: https://www.nikonusa.com/p/z-8/1695/overview. It's not even clear to me that there are any distinctions in performance apart from battery life and card slots.

 

[Kjellis]

Different software updates, with different af machine learning amounts/types/etc.

I can see that this is a plausible cause. And makes a possibility that the two cameras performs more or less the same with the subjects I’m shooting, which is nordic wildlife and birds.
if this is the case this may differ from the different FW updates i reckon.

Interesting topic for sure

 

[JAJohnson]

or one, we know the FW is different but beyond that, Nikon's own published comparison suggests that they are similar: https://www.nikonusa.com/p/z-8/1695/overview. It's not even clear to me that there are any distinctions in performance apart from battery life and card slots.

The Z8 and Z9 have identical sensors, circuitry, and software logic which controls autofocus and processes stills and video to a card. The primary differences are in body structure, heat dissipation ability, GPS antenna and receiver, button layout, battery life and card slots. There may be some isolated differences periodically with capability enabled by firmware (i.e., pixel shift), but I can discern no difference in AF capability.

 

[fcotterill]

The topic of Nikon Mirrorless Autofocus has been covered in threads over and over here and other forums recently.....

https://www.fredmiranda.com/forum/topic/1880200/4

To repost:

the Z9 probably has the edge in AF performance... This is assuming its more recent firmware update (v5.0 13-iii-2024) incorporated the latest standard of Deep-Learning training of the subject detection.

The Z8 is on Fw v2.01 (24-iv-2023)

However, only Nikon know the details about this stuff, whether any differences in AF performance, stickiness etc represent ~11 months advantage to the Z9 of training on superclusters (?)

 

[SCoombs]

The Z8 and Z9 have identical sensors, circuitry, and software logic which controls autofocus and processes stills and video to a card. The primary differences are in body structure, heat dissipation ability, GPS antenna and receiver, button layout, battery life and card slots. There may be some isolated differences periodically with capability enabled by firmware (i.e., pixel shift), but I can discern no difference in AF capability.

Well, we can say for sure that the circuitry is not identical, for if it were then one could load Z9 firmware on a Z8 since the Z8 hardware wouldn't be able to identify itself as a Z8 and restrict the firmware based on that - so there must be some difference.

The question then becomes how much of a difference? I think it must be somewhat significant because Nikon has two different pipelines for the Z8 vs Z9 firmware. If it were just a question of locking some features out on the Z8 firmware, they would still have to put Z8 firmware through its own testing to make nature nothing was broken when flagging the lockouts in the code, but that wouldn't take two entirely different pipelines, which would be very inefficient from a business perspective.

I think then that there must be differences in the circuitry and hardware logic between the two. The question would be how much difference?

 

[Larry Shuman]

Also if your hand holding the lens your shutter speed needs to be 2 1/2 times the lens focal length. Using a tripod is a good idea really.

 

[Cameron T]

Also if your hand holding the lens your shutter speed needs to be 2 1/2 times the lens focal length. Using a tripod is a good idea really.

That's not the rule of thumb. It's 1/x. And even then, it really depends on how steady you are. I can hand hold the 180-600 down to 1/80th comfortably

 

[JAJohnson]

Well, we can say for sure that the circuitry is not identical, for if it were then one could load Z9 firmware on a Z8 since the Z8 hardware wouldn't be able to identify itself as a Z8 and restrict the firmware based on that - so there must be some difference.

The question then becomes how much of a difference? I think it must be somewhat significant because Nikon has two different pipelines for the Z8 vs Z9 firmware. If it were just a question of locking some features out on the Z8 firmware, they would still have to put Z8 firmware through its own testing to make nature nothing was broken when flagging the lockouts in the code, but that wouldn't take two entirely different pipelines, which would be very inefficient from a business perspective.

I think then that there must be differences in the circuitry and hardware logic between the two. The question would be how much difference?

UPDATE: The statement to the right is incorrect. You can’t cross-load firmware between the two because the menus aren’t identical.

I agree that having two development environments (and development teams) for the two cameras isn’t ideal, but that’s what Nikon has done.

My comments in my prior post stand.

 

[Lee R]

Well, we can say for sure that the circuitry is not identical, for if it were then one could load Z9 firmware on a Z8 since the Z8 hardware wouldn't be able to identify itself as a Z8 and restrict the firmware based on that - so there must be some difference.

The question then becomes how much of a difference? I think it must be somewhat significant because Nikon has two different pipelines for the Z8 vs Z9 firmware. If it were just a question of locking some features out on the Z8 firmware, they would still have to put Z8 firmware through its own testing to make nature nothing was broken when flagging the lockouts in the code, but that wouldn't take two entirely different pipelines, which would be very inefficient from a business perspective.

I think then that there must be differences in the circuitry and hardware logic between the two. The question would be how much difference?

From a business perspective I would want the exact same components shared across as many models as possible to enable economy of scale in purchasing those internal components. A bunch of lower rate custom internals is a horrible choice from that perspective.

Any difference is most likely only menu features and something like the menu for the FN3 button on the front of the Z9 and the button diagrams for that shell. A unique firmware is probably just far faster to troubleshoot on a single body rather than making a universal firmware for all the Z9 derivative bodies. That way one issue doesn't hold up the entire model line, instead just one model.

Any difference if any exists in AF performance is most certainly just the AF machine learning difference on the latest build which may go to the Z9 first and trickle down as it's the expensive flagship after all. Nikon likes to keep the flagship software leading the pack as best it can vs a Sony model where they sell "as is" and expect you to update the firmware by purchasing the next version.

Internally other than the sensor and buttons I would think the Z50/Zf/Z6iii and Z8 have the exact same internal components. A shell and buttons will be different.

Nikon has a pretty clear strategy of selling the Z9 hardware (internals and software) over and over as a different less expensive model attached to a paid for sensor down to APS-C. The Z8 certainly is simply the Z9 in a shell that's smaller with a few less buttons and that alone means it needs its own firmware version, likely they just aren't synced perfectly for firmware AF build on release.

I expect them to repeat the cycle on fresh expeed 8 mainboard and internals within another year or two as the next Z9 and following cameras.

 

[SCoombs]

You can’t cross-load firmware between the two because the menus aren’t identical.

I agree that having two development environments (and development teams) for the two cameras isn’t ideal, but that’s what Nikon has done.

My comments in my prior post stand.

From a business perspective I would want the exact same components shared across as many models as possible to enable economy of scale in purchasing those internal components. A bunch of lower rate custom internals is a horrible choice from that perspective.

Any difference is most likely only menu features and something like the menu for the FN3 button on the front of the Z9 and the button diagrams for that shell. A unique firmware is probably just far faster to troubleshoot on a single body rather than making a universal firmware for all the Z9 derivative bodies. That way one issue doesn't hold up the entire model line, instead just one model.

Any difference if any exists in AF performance is most certainly just the AF machine learning difference on the latest build which may go to the Z9 first and trickle down as it's the expensive flagship after all. Nikon likes to keep the flagship software leading the pack as best it can vs a Sony model where they sell "as is" and expect you to update the firmware by purchasing the next version.

Internally other than the sensor and buttons I would think the Z50/Zf/Z6iii and Z8 have the exact same internal components. A shell and buttons will be different.

Nikon has a pretty clear strategy of selling the Z9 hardware (internals and software) over and over as a different less expensive model attached to a paid for sensor down to APS-C. The Z8 certainly is simply the Z9 in a shell that's smaller with a few less buttons and that alone means it needs its own firmware version, likely they just aren't synced perfectly for firmware AF build on release.

I expect them to repeat the cycle on fresh expeed 8 mainboard and internals within another year or two as the next Z9 and following cameras.

You're missing the key point here: yes, Nikon wants the two cameras to behave differently with menus for instance, but those means the hardware MUST be different or else the firmware would not be able to tell what kind of camera it was dealing with in order to make sure a given firmware version can only load onto the right kind of camera.

If it were literally just Z9 parts put in a different body, any Z9 firmware would just see the Z9 parts and think it was a Z9 and load in just fine.

Likewise, we also know even apart from this that the circuitry is different by the different form factor, the lack of the Fn3 button, etc. Taking circuitry - especially logic circuitry - and trying to make it all fit on a smaller board as is the case with the Z9 vs Z8 is not nearly as simple as you might think. You can't just take the same exact components and rearrange them - a redesign is always part of this kind of downsizing and it can make a big difference in how the circuitry functions even if it's intended to do the same thing.
The long story short is that even though they've designed the Z8 to be identical in function, there is no chance the circuits are physically identical and this can have unintended consequences.

 

[Calson]

With the Nikon DSLR cameras the D5/D6 had different motors and could focus faster than the D850 with both using the EN-EL18 battery. It was quite noticeable with a fast approaching bird where the D5/D6 could nail focus nearly 100% of the time but the D850 nearly always failed to lock focus.

With the Z9 if the head direction provides two eyes then eye detection works great. With one eye the detection is not nearly as good. I have also had eye detection fail at 300mm but succeed at 400mm from the same location. The additional image magnification provided more data for the camera's AF system.

 

[JAJohnson]

You're missing the key point here: yes, Nikon wants the two cameras to behave differently with menus for instance, but those means the hardware MUST be different or else the firmware would not be able to tell what kind of camera it was dealing with in order to make sure a given firmware version can only load onto the right kind of camera.

If it were literally just Z9 parts put in a different body, any Z9 firmware would just see the Z9 parts and think it was a Z9 and load in just fine.

Likewise, we also know even apart from this that the circuitry is different by the different form factor, the lack of the Fn3 button, etc. Taking circuitry - especially logic circuitry - and trying to make it all fit on a smaller board as is the case with the Z9 vs Z8 is not nearly as simple as you might think. You can't just take the same exact components and rearrange them - a redesign is always part of this kind of downsizing and it can make a big difference in how the circuitry functions even if it's intended to do the same thing.
The long story short is that even though they've designed the Z8 to be identical in function, there is no chance the circuits are physically identical and this can have unintended consequences.

I misspoke earlier about firmware and menus, I got replies to two different threads mixed up because I’m multitasking.

I don’t think we’re all that far apart. My point earlier is this: regarding autofocus, I can’t tell the difference in performance. I took both cameras out yesterday to photograph ducks on a lake near my home. With the same lens, AF on each camera performed similarly… some things done well, and some frustrations. They’re using the same sensor, and AF system, so I expect nothing different.

 

[Skye.prf]

I misspoke earlier about firmware and menus, I got replies to two different threads mixed up because I’m multitasking.

I don’t think we’re all that far apart. My point earlier is this: regarding autofocus, I can’t tell the difference in performance. I took both cameras out yesterday to photograph ducks on a lake near my home. With the same lens, AF on each camera performed similarly… some things done well, and some frustrations. They’re using the same sensor, and AF system, so I expect nothing different.

For ducks… probably the Z6 is as good as those cameras… small birds in bushes or hummingbirds in flight is another story

 

[SCoombs]

For ducks… probably the Z6 is as good as those cameras… small birds in bushes or hummingbirds in flight is another story

I would say my Z8 noticeably outperforms my old Z7ii for ducks, even though I still find it to be less reliable than I would hope.

 

[fcotterill]

With the Nikon DSLR cameras the D5/D6 had different motors and could focus faster than the D850 with both using the EN-EL18 battery.

The AF motors in DSLRs are for Screwdriver lenses, mostly AFD. The camera motor is irrelevant to AFS lenses.

It was quite noticeable with a fast approaching bird where the D5/D6 could nail focus nearly 100% of the time but the D850 nearly always failed to lock focus.

The D850 and D5 both use EXPEED5 and the Multi-CAM 20K AF module, but the D5 had more clout in its 180K pixel RGB metering sensor with an Advanced Scene Recognition System. IME, their field performance is chalk versus cheese, even with a gripped D850 using the ENEL18.

We're unlikely to ever know, but the deeper layers of AF firmware was probably bespoke to each of the D5 Triumvirate.

The D6 is a distinctly different and much more advanced beast. It's built on EXPEED6 and Multi-Cam 37K autofocus module with 105-point sensors: more densely packaged in a triple-sensor arrangement. These comprise a full array of all-cross-type triplet sensors, unlike the subset in the D5 AF chip.

Nikon also significantly improved the performance of the AF modules, especially pattern detection, including eye and face detection in the D6. These got a further big improvement in FW 1.20 (15 iv 2021). The field performance of the D6 Autofocus is radically different to any other DSLR; it outperforms the Z9 in lock on speed and stickiness. I never bothered with 3D tracking in the D5, yet it's my default BBAF mode on the D6.

With the Z9 if the head direction provides two eyes then eye detection works great. With one eye the detection is not nearly as good.

Bird Detection mode has improved eye detection significantly.

I have also had eye detection fail at 300mm but succeed at 400mm from the same location. The additional image magnification provided more data for the camera's AF system.

Yes I also find filling the frame helps, which is why some bird photographers toggle into DX crop mode.

 

[SCoombs]

The AF motors in DSLRs are for Screwdriver lenses, mostly AFD. The camera motor is irrelevant to AFS lenses.

The D850 and D5 both use EXPEED5 and the Multi-CAM 20K AF module, but the D5 had more clout in its 180K pixel RGB metering sensor with an Advanced Scene Recognition System. IME, their field performance is chalk versus cheese, even with a gripped D850 using the ENEL18.

We're unlikely to ever know, but the deeper layers of AF firmware was probably bespoke to each of the D5 Triumvirate.

The D6 is a distinctly different and much more advanced beast. It's built on EXPEED6 and Multi-Cam 37K autofocus module with 105-point sensors: more densely packaged in a triple-sensor arrangement. These are all-cross-type sensors, unlike the subset in the D5 AF chip.

Nikon also significantly improved the performance of the AF modules, especially pattern detection, including eye and face detection in the D6. These got a further big improvement in FW 1.20 (15 iv 2021). The field performance of the D6 Autofocus is radically different to any other DSLR; it outperforms the Z9 in lock on speed and stickiness. I never bothered with 3D tracking in the D5, yet it's my default BBAF mode on the D6.

If not for the fact that the D6 is 20.8 MP, I might be shooting one of those instead of the Z system. Always wondered why they chose that sensor for the D6. There are several things that one might think could be the reason, but I wonder what the thinking specifically was.

 

[fcotterill]

This article includes a comparative diagram of the triplet AF sensor design.

Nikon D6 : tout savoir sur la nouvelle référence Nikon pour la photo sportive

4 ans après le D5, Nikon dévoile ainsi officiellement le D6 après avoir annoncé son développement en septembre dernier. Que réserve le D6 pour les photographes sportifs ? C’est ce que nous allons voir ensemble.

phototrend.fr

I read an interview in which Nikon explained they prioritized requests from sports photographers for primarily improved Autofocus and networking. The D6 design team prioritized, and they worked with AFP photojournalists.

報道の最前線、その先へパート1 | ストーリーズ | Nikon 企業情報

報道写真の世界を長年支援してきたニコン。世界4大通信者の一つAFP通信とニコンのパートナーシップについてご紹介します。

www.jp.nikon.com

Nikon interview: "We’re at a transitional stage"

In a wide-ranging conversation, Keiji Oishi gave us an insight into Nikon's strategy at all levels of its business, and shared some details of the new D6.

www.dpreview.com

Nikon engineers admitted the Z9 AF cannot win all the races, where the D6 has advantages in some situations

Z 9 開発者Special Interview vol.2～光学性能と小型化を追求したZ マウントシステムの開発意図と誕生秘話～ | ニコンイメージング

新次元の光学性能、システムの小型化、そして使いやすさを追求するニコン Z マウントシステム。具体的な製品の誕生秘話まで、Z マウントの開発意図や背景から、光学設計と商品企画プロダクトマネジャーのインタビュー記事をご紹介いたします。

www.nikon-image.com

 

[TonyF]

I haven't noticed any difference in AF performance between my Z8 and Z9.

If there is a difference between the Z8 and Z9 in AF performance, what would the cause of that difference be attributed to (since the electronic internals are essentially identical)?

It could be for some that the Z9 body handles better with better balance with the longer/physically bigger lenses. From a physical and ergonomics design point of view why would Nikon not allocate priority attention to how the Z9 and its lenses handle? Maybe it could even work the other way around if the Z9 body is simply not right for some. (Hand, finger size etc)

 

[JoelKlein]

The fact that the Z9 had Bird SD before the Z8 tells me that the Z9 and Z8 AF software codes are not aligned.

I remember reading here BCG that the code is written in C. Modern software is written in newer object-oriented languages, which allows the reuse of blocks of code as is. Unlike plain old-fashioned C, which isn't object-oriented and, therefore, requires rewriting, The two cameras are very similar for sure, but the AF code is not. Perhaps it makes a difference.

Another factor is the bigger battery, which powers the Z9 at a higher voltage. Maybe the software clock is faster, too, which gives the Z9 an edge when executing code.