As before, I’m using camera firmware 1.11.

As before, I’m trying to focus on things that aren’t already very well known / discussed about the Z9. And a lot of that is around finding the edge cases where things stop working properly. Please don’t misconstrue that as an overall negative judgement of the Z9 – I like the Z9. (I say this because an earlier revision of this post was reposted on Nikon Rumours, which was very flattering but led to quite some angst in the comments there due largely to this missing context)

All example photos are from High Efficiency* raws as rendered by Lightroom Classic without any edits, other than cropping where appropriate.

Ergonomics

• You cannot use the camera at all – can’t even bring up the menus – if you have a “stowed” lens attached, like the 24-70/4 in compact form. This is not a big deal, of course, but it’s annoying that you can’t do anything with the camera and it doesn’t give any indication why – unlike the Z7, the Z9 doesn’t turn on the LCD in this situation, only the EVF, so it’s very easy to miss the dialog it’s trying to show that’s complaining about the lens.
• It took a little experimentation to figure out how to manipulate the LCD screen easily. If you just grab the side and try to pivot it up or down, for example, it’s almost impossible to do so without fear of breaking something. This seems to be a consequence of its complicated multi-levered design. Fortunately Nikon have included little grabbers on the top & bottom as well, and if you use those appropriately the screen is easy enough to position.
• With the Sigma 150-600 Contemporary attached, I’m finding the Z9 isn’t significantly different in weight from the Z7 (or D500). I also don’t see any meaningful difference in “balance” or any such thing that apparently lots of other people do.
• One thing I hadn’t really considered is that the Z9 is a rounded square, whereas the Z7 was a less rounded rectangle. Consequently the Z9 likes to roll about, such as in a car going around corners, which is annoying – on a recent windy-road drive I had to hold the Z9 the entire time, in my lap, to prevent it attempting to commit suicide against the doors and dashboard.
• Some camera harnesses are problematic with the Z9, unless you have a tripod foot on the lens that you can attach to instead.
  
  e.g. my preferred camera harness is a Cotton Carrier, but the Z9’s big butt forces the camera and lens to stick out way more, which puts a lot more torque on the harness – I worry it’ll eventually snap from the stress (I’ve had one harness break in this way before) – and is just less comfortable all round.

Performance

Autofocus

• It hasn’t completely kicked the Nikon Z habit of obsessive-compulsive behaviour regarding backgrounds. Sometimes with a clear subject front-and-centre, right there in the middle of the frame, it’ll buck all artistic conventions and seek novel compositions in the background.
  
  This is rare, to be clear. The Z7 did this [figuratively] all the time. I think the Z9’s only really done it once, maybe twice so far. So I wouldn’t call this a big deal, even though it’s such an obvious fail when it happens. Still, it’s a bit sad that Nikon still can’t seem to figure this out.

• As others have occasionally commented in reviews, eye autofocus is easily foiled if the eye is obstructed at all, e.g. by hairs, feathers, or similar. In most cases (that I’ve seen so far) it will vacillate randomly between the eye and the obstruction.
  
  As such, it is possible to still get the eye in focus, but you need to take a lot of pictures and hope for the best.
  
  This is not unique to subject-detecting AF modes – the problem occurs in single-point AF as well, for example.
  
  This behaviour in single-point mode is justifiable – in abstract there’s no way for the camera to tell which of the possibilities you actually want – but I would think that when it explicitly detects an eye it would understand how to avoid common obstructions like hairs & feathers, and similarly nearby elements like eyelashes. e.g. focus specifically on circular or oval features, or simply always prefer the furthest detected feature. Alas it does not.

• Eye detection is sometimes wrong about what is an eye – e.g. nostrils on birds. I don’t have much data on this yet, but I get the initial impression that it’s basically just looking for a roundish feature that is darker than its surroundings, and nostrils are often darker than eyes (since eyes often have reflections and/or colour irises), or the eyes might be partially obscured. The behaviour is dependent on the bird – e.g. for California Condors it was a real bugger sometimes in trying to get it to recognise the actual eye, but for many other bird species it had no issues.

• For some animals – e.g. this California Condor which is apparently the Z9 AF system’s nemesis – it doesn’t recognise the actual face at all, but frequently thinks it sees a face elsewhere on the animal. In poses like the one above it mistook the nostril for an eye and prioritised that, but in the photo below it didn’t detect any eyes and instead thought it saw a face in the middle of those neck feathers.

• Similarly in some cases it doesn’t recognise an obvious animal at all, but thinks it sees something interesting in inanimate objects. This is compounded severely by the design flaw where 3D tracking mode will always focus on any detected subject no matter where you tell it to focus, and disabling subject detection (to work around this flaw) requires menu-diving.

• I’ve seen a small number of cases where the camera simply front-focuses for no apparent reason, in single-point AF. Sometimes it’s so severe that nothing in frame is in focus. This is with the 24-70/4 at least. When this happens it consistently misses focus across all photos in a burst (with AF-C active).
  
  To be clear, usually the camera focuses consistently on at least something in the scene – especially with single-point AF it’s generally very reliable. Just not always.
• I haven’t tested it properly yet, but I get the impression the Z9 suffers from the same flaw as the Z7 regarding autofocus performance vs exposure preview. i.e. if your subject is dim in the viewfinder – because you have accurate exposure preview enabled and you’re under-exposing (e.g. it’s an inherently dark subject, or you’re protecting highlights, etc) – the autofocus system performs worse. Simply raise the ISO, for example, and autofocus works much better (but now your actual exposure is wrong, and your photo might be unusable).
  
  This is probably why Nikon have the “View mode (photo Lv)” setting (Custom Settings > Shooting/display > d9), so that you can choose which of these trade-offs you wish to prioritise. But with accurate preview turned off it’s all too easy to screw up the exposure, especially in non-trivial lighting situations where the rudimentary exposure meters just can’t convey what’s going on.
  
  DSLRs didn’t have this issue per se because they had no exposure preview mode – and because the AF system was separate from the image sensor, the AF system could always operate with whatever settings it found optimal.
  
  The crux of my frustration is that it feels like a false dichotomy. I don’t understand why it can’t use whatever ISO is best for AF, but simply adjust the image in the viewfinder to compensate. e.g. if it needs a few extra stops of gain for good AF, then just reduce the EVF / LCD preview by the same amount. Sure, it might clip highlights in the EVF preview, but I think I’d prefer that – the highlights would still be fine in the actual photos taken, and in any case if the subject isn’t in focus blown highlights are irrelevant.

Image stabilisation

• I’m seeing particularly poor image stabilisation performance with the Sigma 150-600 Contemporary. This lens has always had poor image stabilisation, on the Z7 and the D500. But I feel like it’s worse on the Z9.
  
  If I want the majority of hand-held images to be usably sharp at 600mm, it seems like I have to use a shutter speed of at least 1/500. I haven’t really established a baseline by turning image stabilisation off, but I can’t see how 1/500 equates to more than a stop or two of effective stabilisation, at best.
  
  Most of my favourite wildlife photos are taken in the range of 1/10 to 1/100, so this is a big problem for me.
  
  I’m also seeing more (vs the D500, Z7, etc) of the behaviour where the vast majority of photos are unusably blurry and then just occasionally one is actually sharp. There’s not a lot of middle ground. I mention this because with e.g. Nikon’s own 80-400 or the 100-400 you tend to get a much smoother continuum of blurry to not blurry.
  
  It’s the only non-Nikon lens I use with its own image stabilisation, so I can’t draw any conclusions as to whether this is first- vs third-party lens compatibility or anything like that.

Metering

• Ever since it was introduced (D500 era, I think?) I’ve used highlight-weighted metering almost exclusively in stills photography, because noisy midtones can be mitigated but blown highlights can be ruinous. But with the Z9 I found very quickly that it would massively under-expose sometimes in highlight-weighted mode. I switched to matrix metering and have almost exclusively used that so far. It works well most of the time, but of course it does sometimes blow the highlights.
• If “Starlight view (photo Lv)” is enabled (Custom Settings > Shooting/display > d9), the EVF / LCD do not show exposure accurately, irrespective of the setting of “View mode (photo Lv)”. This is extremely surprising and caused me to massively over-expose a bunch of images.
  
  It makes me suspect that “Starlight” mode – which is nominally about autofocus working better in low-light – is nothing more than forcing the sensor into ISO settings that are most amenable to the autofocus system rather than representative of the intended exposure (see earlier point about AF effectiveness vs exposure preview).
  
  Addendum: why use “Starlight” mode in broad daylight? In this case, because I’d had it on the night before and forgot to turn it off. Though in any case I’m still curious as to exactly when it’s supposed to help – e.g. does it only work in extremely low light, or does it boost AF performance in any situation? Nikon’s product material, and 3rd party reviews, say very little in this regard.

Battery life

• The battery lasts about six hours in my use. That’s with the camera on the whole time, GPS enabled (with standby turned off), and sporadic use in 20 FPS mode.
  
  Battery life seems to be mostly affected by how long the camera is switched on and especially how long the EVF or LCD are active. How many photos you take doesn’t seem to be a significant factor. Though video might (I haven’t done a lot of video yet).
  
  I assume the GPS is a significant power suck. Previously, on my Z7, I used a Solmeta GMAX hotshoe GPS because (a) it has a big internal battery that avoids draining the camera’s battery, (b) it was far more reliable than any other GPS unit I ever tried, and (c) having its own battery meant it could keep active irrespective of what the camera was doing or whether the camera was even on. I could use that on the Z9 too, but the combination would be too big for my tastes. If you’re not going to use the built-in GPS you’ll probably see a significant increase in battery life.
  
  I could also turn the camera off between use, which might save a significant amount too – but I’m afraid to do that because I’m not convinced the GPS will work as reliably (the Z9’s manual claims the GPS will continue tracking even if the camera’s off, but frankly I’m sceptical based on past experience with earlier cameras & GPS units).
  
  On the Z7 I’d usually get at best four hours per battery, and that’s even though I’d turn the camera completely off when not actively using it. So it’s clear the Z9 gets significantly better battery life – as you’d expect given it has 157% more capacity.
  
  For a long day of use, or if recording significantly amounts of video, you’d definitely need at least two batteries for the Z9 (or rely heavily on USB tethering).
• Battery charging in-camera officially takes 3h 40m from 0% to 100%, and I haven’t tried to precisely measure that but it seems about right for what I see in practice.
  
  Using a USB battery brick to charge the camera seems to add about 2% every five minutes or so. If you’re out for the day and take a half hour lunch break, for example, you can only add ~12% more juice to the battery. Not all that useful. I find having a USB cable sticking out the side of the camera to be dangerous to the camera’s health (cable yanks, water ingress, etc) so power / charging while it’s in use doesn’t seem practical to me.
  
  I haven’t tested it yet, but in theory you can use any old USB power source to charge a second battery, which will be super helpful (compared to having to use mains power as with previous Nikon battery chargers).
  
  I’d really like to see future Nikon cameras & batteries support faster charging, like most portable electronics do. e.g. USB-PD 60W should be enough to goose the battery from 20% to 80% in just twenty minutes. That’d make it real easy to have a single battery and just top it up every time you take a bathroom break or whatever.

Reliability

Crashes

• It locks up occasionally in playback mode, when reviewing images. It’ll suddenly just stop responding to all controls, and after fifteen seconds it reboots itself. It forgets everything it was doing when it reboots (e.g. the playback image position resets to the most recently recorded image).
  
  [Addendum: some folks suggested it might be the memory card causing this. While it’s hard to definitively rule that out, this happens with at least two different cards from different manufacturers (Pergear & Angelbird).]

GPS

The GPS seems to work quite well so far, though I’ve basically just been moving to a location, preparing, then taking photos, then moving on – a more challenging test will be if I’m moving while taking photos, such as on a boat, or taken photos suddenly after moving. Those are situations in which a lot of GPS units / Nikon cameras will make mistakes, like recording wildly stale location data.

I’ve taken a bit over a thousand photos so far, with GPS enabled.

• Once it’s acquired its position it seems to update it every second, as you’d expect.
• It does actually work indoors (all the time so far for me, but I haven’t been indoors much). This is a pleasant surprise. Some prior GPS units I’ve used have basically not worked at all indoors.
• It seems to be much more capable of actually acquiring an accurate position – when faced with obstructions etc – than most prior GPS hotshoe units I’ve tried, though I’d be [very pleasantly] surprised if it matches the Solmeta GMAX in that regard.
• It’s never failed to record the location so far.
• It’s never gotten the location completely wrong, so far.
  
  It’s not super accurate, though – I see longitude and latitude inaccuracies of up to ~ten metres in some cases. It’s good to just a couple of metres most of the time, though. The Solmeta GMAX is more consistently accurate, but for my purposes ~ten metres of error – occasionally – isn’t a big deal.
  
  Altitude isn’t accurate – its regularly off by tens of metres. e.g. at the beach yesterday it consistently thought I was 12 to 19 metres below sea level.
• Even with all the standby stuff disabled and the camera left on at all times, it does lose its position if you’re not actively using the camera (according to the position read-out in the menus). That concerns me, as this mimics the behaviour of most prior [hotshoe] GPS units with Nikon cameras, and my experience with them is that they fail to correctly tag photos a lot. But, as noted above, so far it’s actually worked well despite this.
  
  I have not yet tried enabling GPS logging – presumably that forces the camera to maintain an accurate GPS location at all times. So that might be a workaround, if there are any issues.
• It doesn’t record heading like the Solmeta GMAX, but then to be honest magnetic compasses tend to be uselessly inaccurate anyway so I’m not really missing that (though if it had it, and it worked reliably, that would be nice).
• There’s no way to tell at a glance if the camera actually has a GPS location fix. There’s a satellite icon that shows up in various places in the GUI, but all it seems to mean is that GPS is enabled, not that it actually has a location lock.
  
  You can dig into the menus to find out, but that’s a bit slow.
  
  Nominally it tells you when it doesn’t yet know its location, by blinking the satellite icon. Indeed I do see that happening when the camera is first turned on and acquiring its location initially. But, even when it stops flashing the camera doesn’t always know its position (according to its read-out in the menus).
  
  I miss the built-in LCD of the Solmeta GMAX where I could see at a glance if it had an accurate position fix (and gauge roughly how accurate via the number of satellites acquired). Maybe Nikon could issue a firmware update that adds a little satellite count badge to the satellite icon, like most GPS units display.

Software support

• Just a minor oddity, but Lightroom’s Import dialog seems to have problems with High Efficiency NEF files – it won’t reliably show thumbnails for them. You can “jostle” it by mousing over the blank spaces to get it to load the thumbnails, most of the time, but this isn’t something that’s required for Nikon NEF files from earlier cameras.

My Z9 has firmware version 1.11.

Note also that I’ve read & watched practically everything out there so far on the Z9, and have no desire to repeat what’s already been covered to death. I’m only noting here things that surprised me or otherwise weren’t well-covered elsewhere. That tends to mean oddities or negative things – but to be clear, my overall impression is very positive; there’s a lot to like about this camera, it’s just that most of that’s been heavily documented already.

Ergonomics

• It’s heavy. Dense. I thought I’d be fine with the weight and was actually only worried about the volume, but in practice it’s the opposite.
  
  The size is fine – not great; smaller would be nice of course, but the Z7 plus a reliable GPS unit in the hotshoe has basically the same bounding box as the Z9, so the Z9 is effectively no larger for my purposes.
  
  The difference is of course most pronounced with baby lenses like the 24-70/4. With the Sigma 105/1.4 on it, it’s not that different to with the Z7. Similarly if you stick a telephoto on it, e.g. a 150-600, the perceived difference vs the Z7 is pretty minor. Still, for casual ‘walkabout’ use, with a small lens like a 24-70 or 24-120, I’m a bit concerned.
  
  I’d assumed it was the bigger battery that contributed a lot of that extra weight, but actually the battery is pretty light. Also much smaller than I’d assumed.

• The grips are a bit big. I’m a guy with fairly large hands, too. It’s similar to the large DSLRs (e.g D500). I know when I went from those to the Z7 I thought the Z7 grip was too small at first, but I very quickly got used to it. I do think the Z7 grip is a tad smaller than ideal – and certainly too short vertically, as my pinky finger inevitably ends up below the Z7 rather than holding it – but overall it’s closer to ideal than the Z9.
  
  Holding the Z7 feels like you’re holding something precise, albeit a bit diminutive, whereas holding the Z9 is like holding a literal brick.
  
  That said, I’m expecting I’ll get used to it.
• The dpad feels kinda stiff. It’s a tad difficult to actually press (and yet feels quite mushy). Not great.
• The control wheels are much nicer than the Z7 ones. The Z7 ones feel nice, at first, but they’re metal and relatively sharp and I know from uncomfortable experience that they can cause blisters on your fingers. The Z9 brings back the rubberised DSLR-style control wheels, which don’t feel as fancy but are much better for heavy use.

• I’d read some complaints that the memory card door is difficult to open, especially one-handed, but it’s completely trivial – push the slider down, then push it left, then just let go and the spring-loaded door opens itself. It takes literally just one finger. Can you finger-paint an L? Then you can open the memory card door.
  
  It’s much easier to open & close than a GoPro case, for example. Possibly too easy, if the goal is to prevent accidental openings.

• I’ve almost always used pure electronic shutter mode on the Z7, for the peace & quiet, but even so it’s still noticeably harder to tell that you’re taking photos with the Z9. I assume because of the complete lack of blackout [most of the time]. Even with the flashing indicators enabled.
  
  [Follow-up: case in point, I went to the zoo a few days ago and unintentionally took 8,000 photos! That’s about 10x what I would have taken with e.g. the Z7, or even the D500. It’s so easy to suddenly have thousands of photos without really noticing. Granted I only kept ~1,500 after culling blurry & duplicate ones, but the culling process alone took a few hours. The Z9 needs to come with little minions to help manage the output. 😜]
• The camera gets warm when recording video (8k/30 and 4k/30 at least) or bursting at 20+ FPS, in a cold room. Not hot, but it does make me a little concerned about over-heating in hot environments. We’ll have to see.
• The EVF is visibly low resolution, just like the Z7 – by which I mean you can see the individual pixels. But honestly I’ve never really felt this was a practical problem – how sharp the EVF is has no bearing on how sharp & correctly focused the actual photo is.
  
  A sharper EVF (and LCD) would certainly be nice, no mistake – I certainly prefer “Retina” displays over their pixellated forebearers – but it’s not a big deal despite what so many others claim.
  
  The EVF still looks better than the cheap, shitty LCD displays most of the world use on their computers.
  
  One purported problem with the low resolution of the EVF is in judging focus accuracy. I think the impact is small if not completely insignificant. There are much better ways to check focus accuracy – e.g. digitally magnifying the view (I assign one of the function buttons to jump to 200%). These are much more effective and indeed necessary if you actually care – with 45 MP no human eye can check critical focus without zooming in a lot.

Performance

This section is particularly preliminary since it’s based on just a couple of hours of use, and this aspect is inherently hard to judge.

Burst shooting

• As others have recently noted, if you use shutter speeds below 1/250 you don’t actually get 20 FPS. It drops first to 15 FPS and then about 10 FPS at 1/30, etc. That’s disappointing – most of the time I’m using less than 1/250, so it appears I’m rarely going to see the claimed 20 FPS.
  
  It’s also weird that in 30 FPS JPEG mode it’s not limited until 1/50 or below.
• The buffer depth is better than I expected based on other people’s tests, possibly because most tests to date were done with 1.00 firmware – Matt Granger’s results show a huge improvement with 1.11 and his numbers match mine for the one card we have in common, the ProGrade 256GB. Or maybe folks were using the HDMI output simultaneously, which Matt Granger also documented as having a significant negative effect on buffer depth.
  
  I get at least 50 HE* photos at 20 FPS before any slowdown, even with a Pergear 512GB CFExpress card which is probably one of the slowest available (it’s optimised for $/GB, not performance).
  
  I was anticipating “having to” buy a performance CFExpress card, but after these initial tests I’m not sure it’s necessary at all. It looks like I can easily take 900+ photos a minute if I want, which is already way more than I should be encouraged to, given I have to sort through them later.

• The buffer depth increases to 80-110 when dropping to 15 FPS. At 10 FPS it is effectively infinite, even with my slowest cards. That’s nice – it essentially matches the D500 in this respect, at least – and the D500 could only achieve an infinite buffer with (for its time) high-end CFExpress cards.

Autofocus

• Autofocus in low light is definitely better than with the Z7, but it still has plenty of room for improvement (which is not to say any other camera is better – I’m pretty sure no camera has what I’d consider truly good low-light AF, yet).
  
  e.g. it still basically can’t focus at all in my TV room [with the 24-70/4], which is a somewhat dim room but not dark by any means – I could sit there and read a printed book just fine, for example. Most of the time it just hunts endlessly. Subject recognition doesn’t work at all.
  
  The Z7 fails even harder – it often doesn’t even try hunting, it just outright refuses to engage the focus motor.
  
  The D500 can focus in similar light in the same room – the D500 is impressive in low-light AF ability, though it tends to over-estimate itself even so; it’ll show a green “got it!” focus indicator but the actual photos will quite often show that it missed. Still, it would at least get some in-focus photos, where the Z9 cannot.
  
  It’s worth noting that my iPhone 11 can autofocus decently in these same conditions – faster & more reliably than any of these “real” cameras. Admittedly it has a wider field of view with much deeper depth of field, so it doesn’t have to be as accurate. But still.
  
  In brighter (but still interior) lighting, the Z9 suddenly starts working and works really quite well, at least for stationary or slow-moving subjects. The transition from basically not working to working well is surprisingly sharp.
• I haven’t really tested it yet, but AF accuracy with a subject moving at non-trivial speed – i.e. my cat trotting towards the camera – isn’t great at first blush. The majority of photos have focus completely missing the head, let-alone the eyes. Oddly focus bounces in & out between every frame, like it’s hunting. The D500 would tend to hit or miss in longer batches. The Z9 is clearly superior here, since having every other photo in focus is way better for capturing a decent moment than missing ten in a row.
  
  The 24-70/4 might be a factor in this, of course. Though it seems to have pretty snappy AF, in my experience.
  
  The Z7 wouldn’t have managed any photos in focus, except maybe by dumb coincident luck, so it’s a very clear improvement over that at least.
• Subject recognition doesn’t work for close subjects, e.g. when the face fills the frame. At least for cats. It stops recognising that it’s even a face, let-alone picking out the eyes. It then tends to focus on merely whatever’s nearest, e.g. the tip of the nose.

• Eye AF consistently front-focuses, because it favours the eyelashes or eyebrows (particularly for animals) over the iris. When depth of field isn’t able to cover this up – e.g. close distances or wide apertures – it’s really obvious, because the Z9 is so consistent with focusing on the wrong thing. A more “manual” focus mode has to be used instead, like single-point AF.

• The AF system continues to not be able to distinguish stuffed toys from real animals. It was a bit stubborn about focusing on a stuffed lion’s face rather than my actual cat, for example. Not a real-world concern for me, of course, but kind of amusing.
• Auto-area AF looks much more promising, at this early stage, than the Z7 (or any DSLRs). So far it’s been pretty reliable about finding the subject (usually my cat, thus far). Not perfect, for sure, but good enough that I’m not frustrated.
  
  For years I almost exclusively used single-point AF on the Z7 because frankly it was the only AF area mode that was usable (with 1.00 firmware). Recently I tried experimenting again, and was quite pleasantly surprised that its other AF modes have been made actually somewhat usable by its firmware updates. I’ve even been dabbling with auto-area AF, in certain easy situations. But it’s immediately obvious that the Z9 is much better.
• Subject detection sometimes finds “subjects” in bizarre places, like random patches of couch fabric or shadows on a bean bag chair.
  
  This would be insignificant if it weren’t for the fact that 3D tracking will always try to track the nearest detected subject – anywhere in the frame, irrespective of where you place the tracking box. So if it detects a non-existent subject – or in any case if you just want to have it track something specific – you have to dive into the menus to disable subject detection entirely. Slow and annoying.
  
  If I place the tracking box over a detected subject and press AF-ON, sure, it should use its subject tracking smarts. Otherwise, it should just do what it’s told and stop trying to be “helpful”.
• So far, 3D tracking doesn’t seem dramatically better than the D500, but I haven’t tested it much in real-world conditions. It’s nice to have it back, and I will be testing it further in various conditions, but I’ve already seen the hallmark flakiness in some cases – e.g. where it just immediately loses the subject and focuses randomly on something else.
  
  I think Nikon’s subject tracking, as on e.g. the D500, has always been rather over-rated. But it does work well sometimes and is super nice when it works.
• I’ve noticed that the Z9 tries very hard to be “sticky”, even with “Blocked Shot AF Response” set to 1 (Quick). But if you release AF-ON and press it again, it instantly refocuses on whatever’s under the AF box. This makes some sense, I guess, but I’m going to have to train myself to feather the AF-ON button. I wish when you set it to ‘Quick’ it would just always focus on whatever is under the AF box.
  
  In auto-area AF and similar modes (as opposed to single-point) it consequently still exhibits some of that annoying behaviour where it loses the subject, focuses on the background, and is then too slow to refocus on the subject.
  
  Still, that you can re-press the AF-ON button and have it do what you want is a big improvement from the Z7, which would often stubbornly refuse to refocus on the subject until you manually adjusted focus to be on the right subject (and even then it would sometimes obstinately refocus immediately on the background again).

Image stabilisation

• Image stabilisation doesn’t seem any different from the Z7 – which is to say it’s good, but could be better. e.g. ¼s exposures hand-held at 70mm still have a low success rate. I might be seeing more interesting (positive) results with the 105/2.8, that supports the enhanced “Synchro VR”, but it’s a bit early to tell.

Responsiveness

• There’s a noticeable delay – half a second or so – for the Z9 to turn on and be ready to use. It’s faster than the Z7, but far from the instantaneous that many reviewers have been claiming.
  
  That’s with all the ‘slowdowns’ disabled, like the sensor shield and restoring AF position. Enabling those doesn’t actually make much difference, as far as I could tell.
• It’s occasionally a little slow when doing certain things in video mode, e.g. immediately after you stop recording it can take a moment or two to start responding to the controls again.
  
  The Z7 is similar. The DSLRs might have been better – I don’t remember, but in any case it’s largely irrelevant given how bad they were at video (e.g. unusable AF, no viewfinder support, etc).
  
  One thing the Z7 was infuriating about was taking up to ten seconds after burst shooting before it’d let you switch into video mode, seemingly because it refuses to switch modes until its buffer is empty. I don’t yet know if the Z9 suffers from this too, but because the Z9 buffer clears quite quickly, it’ll hopefully prove far less annoying even if it does.
• The GUI feels snappier than the Z7, but I haven’t compared them side-by-side yet. I mainly feel like image review is significantly swifter – zooming in on a photo is instantaneous, whereas the Z7 usually takes a moment. Skipping rapidly back and forth between photos is also instantaneous, whereas the Z7 had just a bit of delay. Overall it’s subtle but the effect is to make the Z9 feel significantly more powerful.
• It’s subtle if present, but it feels like the scene view (on LCD and EVF) has less latency than the Z7. The AF indicators do lag the subject / camera movement, but it’s only significant if you have very fast movement, and I don’t yet know if it actually impacts focus accuracy.
  
  It’s subtle enough that if I didn’t already know Nikon are claiming improvements in EVF latency, I wonder if I’d have noticed anything at all.
• The EVF & LCD lag significantly in some situations, involving slow shutter speeds (e.g. 1/10) and/or low light. Similar to the Z7 (and my iPhone, for that matter). To be clear, this is when not taking any photos, just looking at the screens. I think most cameras do this – I guess they figure it’s better to lower the refresh rate than have [more] noise in the image preview.
• The 60 Hz “limit” of the EVF & LCD doesn’t seem problematic to me. I have various devices with 120 Hz displays (e.g. my iPad) but frankly I’ve never seen any meaningful difference. But I haven’t used an EVF with a refresh rate higher than 120 Hz, so I guess I can’t be sure.

Video quality

• The oversampled 4k (30 FPS or lower in FX mode) is very nice, at least in H.265 10-bit HLG (I haven’t played as much with other video formats).
  
  There’s no mistaking that it has been processed – it’s definitely a little crunchy if you look closely, and especially if you compare vs the 8k down-sampled in playback. The 8k does contain more genuine detail – though not 4x as much, by any stretch – and looks much more natural with no evidence of artificial sharpening.
  
  It reminds me a bit of the results from recent iPhones in their “Deep Fusion” mode, though not as strong (thankfully).
  
  The processing settings – including several aspects of sharpening – are configurable in strength, but I haven’t played with that yet. I’m pretty comfortable with the defaults (I intend to use the over-sampled 4k mode, as opposed to 8k mode, when I want quick results that don’t require post-processing).

Reliability

• As hoped, the Z9 has yet to fail to take a photo when told to (other than in contrived, pathological cases like after filling the buffer with an 800-photo burst). It’s clearly better than the Z7. Which is a low bar, of course. The Z7 is a very unreliable camera for actually taking photos, thanks to its anaemic buffer and slow image processing. I’m hoping that continues to hold true in more demanding testing – I’ve missed the D500, which was the first and (’til now) only camera I ever used where you could trust that it would work when you pressed the shutter button.

Connectivity

• Pairing with Snapbridge on my iPhone went fine. The process is a bit inelegant, but no different than with any other Nikon cameras, and worked first time.
• Snapbridge is very snappy for camera remote control. I’m not sure how it compares with the Z7 as honestly I’ve virtually never used the Snapbridge app, but I was impressed with how low-latency it is (and laughed that you can use it to control the camera’s LCD in playback mode – I’m not sure what the point of that is but it’s amusing).
• Image download over wifi to an iPhone is pretty slow. About ten to fifteen seconds per ~50MB image. I don’t recall if that’s different from the Z7 – but I remember when image download was only supported over Bluetooth, and that was excruciating… something like up to a minute per 2MB image. So progress, I guess, though still inexplicably slow.
• It refuses to charge at all over USB from some chargers – including chargers that work just fine with the Z7. I’m guessing it requires some minimum wattage that exceeds what basic USB power bricks can provide. Not a big deal, but a tad disappointing (I’d be fine with it charging slowly, as long as it still charged).

Factory settings

• GPS is off by default and the UI is surprisingly unintuitive about it – it took me an hour and several attempts before I finally confirmed that it was off and figured out how to enable it.
  
  I guess I can understand why it might be off by default, given its potential for significant battery drain, but then standby power saving mode is disabled by default, and that has a much bigger negative impact.
• Standby power saving mode is disabled by default. This makes the camera consume a noticeable amount of power when switched on but not being used.
  
  This makes some sense as a default, I suppose – standby mode presumably incurs some delay when you snap the camera out of it, which the nominal audience for this camera might hate. But I was surprised to see 10% of the battery disappear just because I left the camera sitting on a desk for an hour without formally switching it off. This was never a problem on the Z7 or any of the DSLRs – though admittedly the Z7 has a noticeable, annoying delay to wake up from standby.
• The video format settings are just stupid by default – 4k/30 H.265 8-bit (SDR), if I remember correctly. They should be 8k/30 H.265 10-bit with HLG or N-log enabled (or ProRes – not my preference but I could at least respect that choice).
• With HLG in use the video display is flat by default. And the preference for fixing that is really obtusely named – “View Assist” – and buried in Custom Settings > Video > g8. For someone that wants to use HDR simply to get better dynamic range, I see no point in making the live preview ugly as a side-effect.
• AF is set to single point by default. Not a big deal, but kinda weird for this camera in particular – a major point of the Z9 was its new “handle anything” auto-area AF, so I’d expect that to be selected by default.
• Video AF is set to AF-F by default… I do actually use that occasionally, but I’d be surprised if so-called video professionals use it much at all – unless you’re a Youtuber you know that shifting focus during a scene is to be done carefully and rarely. AF-C makes much more sense to me.

One thing I have very quickly discovered, however, is that Lightroom “Classic” deliberately chooses not to perform some functions if it is le tired.  Or it thinks your computer is le tired.  By which I mean, if there is pretty much anything else running and consuming CPU time (and/or RAM?), it refuses to even attempt some operations.  HDR merges is the first one I hit.  I was a bit flummoxed by it just happily queuing up a number of HDR merge operations, and them just sitting there in its queue, with no indication of error – just never executing.

Only after I quit or disabled a bunch of other processes – any and all that were using any measurable CPU time – did it finally, about ten seconds later, decide that it was now willing to consider my ‘requests’.

#%@!ing fussy little turd.

It’s worth noting that it’s not the only popular app, on macOS, that does this same bullshit.  Time Machine is another big one.  At least in Time Machine’s case I can see a more plausible line of reasoning behind it, even if it is misguided – the user’s probably not explicitly waiting for a Time Machine backup to complete.  As in, not all the time.  Sometimes they are. And they certainly expect backups to happen at all, which on a consistently busy machine simply doesn’t happen.  So Time Machine’s reluctance to function on a working machine is still stupid overall.  But Lightroom refusing to complete a user initiated, user-interactive, and user-blocking operation, is just patently stupid by its very notion.

Update:  Worse, now it doesn’t work at all.  And a quick web search shows many other people having the same problem, and Adobe as usual doing nothing about it.

Incidentally, I tried to log in to Adobe’s forums in order to ‘Me too’ those issues, only it won’t let me log in anymore, falsely claiming my password is invalid.  Good job, Adobe, good job.