Nikon Z9 review part 2: Autofocus performance

I was really curious about the Nikon Z9’s autofocus performance and this was the number one question I received on that camera, so I gave it a thorough run-through. We will go through the following:

• Overview of the Z9 autofocus
• Macro and Close-ups performance
• Side-by-side comparison with the D500 (for macro)
• Wide-angle performance
• Wrap-up: autofocus performance

A fiddler ray swims in-between kelp (Nikon Z9, Nikon AF-S 8–15mm f/3.5–4.5 fisheye lens, dual Ikelite DS230 strobes, f/13, 1/40s, ISO 400)

Overview of the Z9 autofocus

On paper, the Z9’s autofocus is well ahead of the D500 and D850, which are Nikon’s most popular DSLRs with underwater photographers.

First, it boasts 493 detection points, covering a whooping 90% of the frame. I remember being thoroughly excited when Nikon announced the D500’s 153 points (51 selectable), which covered about 50% of the frame (92% horizontally, 57% vertically). The D850 and D5 were using the same 153 points AF module, but because of their larger sensor, it only covered 21% of the frame (58% horizontally, 35% vertically). I thought to myself: if the camera is really able to track the subject, then having a 90% frame coverage is a big advantage.

Secondly, the Z9 is meant to focus in very low light: up to -7 EV, or even -9 EV in starlight mode. The starlight mode is meant for static scenes so as far as underwater photography is concerned, -7 EV is a more relevant number. In comparison, the D500/D850 can “only” focus down to -4 EV, and that’s only in the center of the frame, the rest of the focus points being limited to -3 EV. Putting it into perspective, the Z9 is meant to focus when the ambient light is 8 times dimmer than what the D500/D850 could cope with.

Thirdly, the Z9 offers a layer of subject detection modes. I am calling it a layer, because you can choose to activate it “on-top” of the traditional AF-S (autofocus single) and AF-C (continuous auto-focus) modes, that D500/D850 shooters are familiar with. When subject detection is turned-on, the Z9 tries to identify what the subject is, and when it manages to do that, it will display a white rectangle over that subject, giving you the option to focus there. There are five detection modes: people, animals, vehicles, auto and off. 

In the “off” mode, subject detection is simply deactivated and the autofocus functionality is quite similar to that of a Nikon DSLR (but the performance is different, as explained later). The “people” and “vehicle” modes are of little use underwater, so I focused my testing on the “animals” mode. According to the Z9 manual, animal-detection covers dogs, cats and birds, but land photographers have had some success with other mammals, so I was hoping it could recognize some marine life as well. In the “auto” mode, the camera will try recognizing any sort of subject, be it a person, plane or an animal. I didn’t use that mode much, because I didn’t plan on photographing much airplanes nor humans.

Before discussing the real-world performance, let’s remember that Nikon DSLRs had several sub-modes, in the AF-C (continuous tracking) umbrella. In the dynamic area sub-mode, you selected the size of the area where the camera could track the subject, around the initial point. With AF-C 3D sub-mode, you would let the camera use any of the available points, aiding itself with colors, to make sense of what it had to track. Auto-area AF was the most “hands-off” of all AF-C sub-modes: you didn’t have to pick an initial focus point, the camera would decide where what seemed to be the subject, and focused on it. Again, the Z9 offers all these variants of the AF-C mode, and you can choose to activate subject detection on-top.

Macro and Close-ups, with the Nikkor Z 105mm f/2.8 VR S lens

In this section, I will share my observations on the Z9 performance for macro and close-ups.

Introducing the Z 105mm lens

As we know, the ability to track a fast moving subject depends not only by the camera, but also on the lens’ mechanics: some lenses focus faster than others. For example, the F-mount 60mm AF-S macro lens is well-known for being more “reactive” than the F-mount 105mm AF-S lens.

By design, the newer Z-mount 105mm is meant to have an edge over its predecessor, with the introduction of a second focusing motor, and it certainly felt faster and didn’t hunt as much, as the venerable F-mount 105mm.

Low-light and low-visibility focusing

To put the Z9 + Z 105mm combo through its paces, I went diving Chowder Bay (Sydney harbor), a great jetty dive for critters, but with low visibility and volatile sediment. I found the perfect conditions to challenge the Z9: 4 meters of visibility and the dim ambient light of the end-afternoon, on a cloudy day. I took fish portraits using AF-C 3D, with and without animal recognition.

Outside the jetty, the Z9 locked focus on the eye of moving fish and tracked it reliably, even though I wasn’t using a focus light, despite the limited ambient light. I was shooting at F/11 and my success rate was above the 80% mark, and I was impressed by how “sticky” the tracking was. 

The eye of goatfish kept in focus despite the animal stirring up silt as it searched for food. Nikon Z9 with Z 105mm, 1/200^th, F/11, ISO 400, 2x Retra Flash Pro

The eye was still tracked as the fish turned away, up to a 120 degrees angle. Nikon Z9 with Z 105mm, 1/200^th, F/11, ISO 400, 2x Retra Flash Pro

Then I moved below the jetty where the ambient light was even lower and it became difficult to spot marine life, unless there was movement. The Z9 was still able to focus on stationary subjects with the focus light off, but I had to turn it on for swimming subjects. As soon as I switched-on my 300 lumens focus light (60 degrees beam), the subject tracking worked beautifully. 

As much as I can, I dive without a focus light because it can spoil the advantage which my rebreather provides, in terms of subject approach. I was quite satisfied with how the Z9 handled the low-light / no-focus light challenge.

Erratic subjects tracking

Since I had switched-on the focus light for the jetty, I went on to shoot the most difficult subject that lives in Chowder Bay: southern pygmy leatherjackets. These cute animals are only a few centimeters long, and they can’t seem to stop buzzing around, swimming erratically all day long (at night they sleep, different story). These are the worst subjects for bokeh shots and I wanted to see if the Z9 / Z 105mm combo could make that kind of photos more achievable. I went shooting at F/4, in AF-C 3D mode, selecting the eye of the fish and tracking continuously. 

18 shots taken in 2 minutes at F/4 (uncropped), 16 of them tack-sharp on the eye!

To my great surprise, out of the 18 shots I took in 2 minutes of the same individual, 16 were sharp on the eye, an impressive 89% success rate.

Southern pygmy leatherjacket. Nikon Z9 with Z 105mm, 1/200^th, F/4, ISO 100, 2x Retra Flash Pro.

Partially obstructed subjects

I also found that “sticky AF” to be very good at keeping the focus, while a distraction passed in between the camera and the subject (I had set the “block shot AF response” to intermediate – 3 over 5).

As the White’s seahorse moved behind pieces of algae, the Z9 kept the focus locked on its eye. I had set the a3 “block shot AF response” to 3 (intermediate). Nikon Z9 with Z 105mm, 1/200^th, F/8, ISO 64, 2x Retra Flash Pro

Animal recognition underwater

I used AF-C 3D modes for all the above tests, even when the subject was stationary, as the Z9 managed to keep the subject in focus while I was approaching and re-composing. Most of the time, I had animal recognition turned-on in case it could help, and I identified three scenarios:

• Scenario 1: the Z9 detected a subject, which it told me by displaying a white rectangle over it. As it keeps thinking, that rectangle sometimes shrank in size, surrounding only the eye of that subject, giving me the option to focus directly there.
• Scenario 2: the Z9 incorrectly thought it has detected the subject, displaying the white rectangle in the wrong place.
• Scenario 3: the Z9 isn’t detecting any subject, and I only saw the red focus square that I could move around (AF-C 3D), no white rectangle whatsoever. 

I found scenario 1 occurred more often with animals that have a regular fish shape, think a seabream, a goatfish, a wrasse. It worked particularly well when they stood out from the background: swimming over sand (goatfish) or in mid-water (seabream, wrasse). For mid-water fish that move fast, I preferred auto-area AF over AF-C 3D, and I found it worked beautifully, especially when animal recognition kicked-in.

It looks like the Z9’s animal recognition also has goatfish covered! Nikon Z9 with Z 105mm, DX mode, 1/200^th, F/5.6, ISO 100, 2x Retra Flash Pro

On another dive with better conditions (8 meters viz, sunny) I found the Z9 did wonders in recognizing potbelly seahorses, saving me from having to position the focus point right on the eye (AF-C 3D). 

Animal recognition picked straight away this potbelly seahorse, most of the time the whole head and sometimes narrowing down to the eye. The seahorse, the sponge and myself were all moving due to search, and animal recognition spared me the effort to manually position my focus point over the eye, when initiating continuous tracking. 

Nikon Z9 with Z 105mm, 1/200^th, F/4, ISO 64, 2x Retra Flash Pro

Yet on subsequent dive in lower visibility, the subject detection didn’t pick on another seahorse species, yet similar looking (white’s seahorse). 

I want to call-out one scenario where animal recognition dramatically improved my focus success rate and this was with Sydney Pygmy Pipehorses (idiotropiscis lumnitzeri). These small syngnathids (2 to 4cm long) are best shot with a wet diopter, but they keep swinging left and right with the surge, making focusing a nightmare.

Kudos to Nikon on this one: the animal recognition reliably identified and tracked the eye, as soon as the Sydney Pygmy Pipehorse (after pre-focusing near the animal). Nikon Z9 with Z 105mm, Nauticam SMC-1, 1/200^th, F/29, ISO 800, AF-C  3D with animal recognition. 2x Retra Flash Pro, 10% crop.

Overall, my testing showed that the Z9’s animal recognition was capable to recognize various fish species, including seahorses and I have had occasional success with octopi. It worked better when the water was clearer and with more ambient light, especially if the animal stood out from its background.

The only circumstances where I would disable subject recognition altogether, is when I can see it is getting confused by the subject: with some pygmy leatherjackets, the Z9 was mistaking their skin patterns of eyes, and focusing on the fish’s flanks as a result. 

Issues with red lights

When night-diving, I usually switch to a red focus light when approaching skittish critters. 

Unfortunately, the Z9’s autofocus performance dropped under a red light: it was hunting for focus at times, and was generally slower. According to Thom Hogan, this is due to mirrorless cameras only using the blue & green photosites of their sensor, when it comes to focusing functions. This caused me to change the way I shoot at night, relying more on white light, which is ok for most subjects, but not all. 

I have read reports of that problem impacting Sony users too, but I am unsure about Canon. I was told Nikon are looking into this, and it would be fantastic to see improvements with a firmware update, time will tell.

Side-by-side D500 comparison

While I was giving the autofocus of the Nikon Z9 a hard time, it struck me that my testing routing was perhaps a bit over the top. Sure, I do use my DSLRs in low-light and low-visibility conditions, I avoid turning on the focus light, but I am no masochist: I also try to make the camera’s job easier, to increase my chances of taking good shots.

Before I could confidently write whether or not the Z9 had outperformed DSLRs for underwater macro, I needed to give the long-timers a fair go. I considered spending a day in one location and swapping cameras in between dives, but the Sydney conditions can change a lot in a couple of hours:  the sky clears up, the visibility changes with the tide, or I photograph that other boxfish which is less skittish than the previous one.

For all these reasons, I resolved to spent two dives carrying both the Z9 and the D500, and alternating between the cameras, to compare them on a mix of macro/close-up subjects, in AF-C 3D mode.

Why comparing with the D500 (instead of a full-frame) you may ask? Well first, I didn’t have a D850/D5 at hand, and I already knew my D810 wasn’t a match for the Z9. Second, the D500 has the same autofocus module as the D850/D5, but importantly, it covers a larger share of the frame (50% vs 21%), which is closer to the Z9’s 90% coverage. Obviously, I needed to work with the same field-of-views and working distances to compare apples-to-apples. The D500 was setup with the F-mount 105mm lens, and the Z9 with the Z 105mm lens. Last but not least, the Z9 was shooting in DX mode, producing about the same image size was the Z9 (19.4 vs 20.9 Mpix), and projecting the same frame, in the viewfinder.

The erratic eastern smooth boxfish swirling in between rocks. Nikon Z9 with Z 105mm, DX mode, 1/200^th, F/5.6, ISO 200, AF-C 3D with animal recognition. 2x Retra Flash Pro.

Qualitative assessment

As I swapped cameras, I observed the D500 was faster at acquiring initial focus on the eye, but then it dropped it quickly, resulting in the focus point jumping on other parts of the fish’s body. If this doesn’t sound like the venerable D500, remember I worked in relatively low light, without a focus light and the visibility was between 5 and 8 meters during these comparative tests.

It took the Z9 a little longer for initial focus, the difference was probably sub-second, but noticeable. However, once the eye was in focus I could see the focus point relentlessly tracking it, whether or not I was moving, re-composing, even when we it got very close to the edges of the frame, well outside of the D500 focus points coverage, even when the subject was turning away (up to a 120 degrees angle). 

As a consequence of these observations, I found myself re-focusing frequently with the D500 (whenever the tracking dropped) and thought this actually worked well. On the Z9, I found that pre-focusing near the subject, before trying to focus on the eye, made the initial eye focus acquisition faster. 

At the end of the first comparative dive, I felt both cameras did a pretty good job, as long as I adapted to their strengths and weaknesses, as described above. So, I needed something else to pick a winner.

Quantitative assessment

For the second dive, I set-off to do a qualitative comparison, to assess which camera was best at quickly acquiring focus, under challenging conditions. Within one minute, I would take 10 to 20 shots of the same subject with one camera, then repeat with the other camera. To give the Z9 and D500 more chances to fail, I used an open aperture (F/5.6) and obviously, no focus light.  

The following five animals were cooperative enough to let me photograph them with both cameras, in consistent conditions:

• Sydney Cardinalfish, hidden in the dark, below a rocky ledge.
• Eastern Smooth Boxfish: erratic, fast moving in between rocks
• Weedy seadragon: cryptic animal, swimming in between kelp
• Common Octopus on the seabed with surge stirring up sand.
• Potbelly seahorse, feeding in between sponges.

A potbelly seahorse feeds in between algae. Nikon Z9 with Z 105mm, DX mode, 1/200^th, F/5.6, ISO 100, AF-C 3D with animal recognition. 2x Retra Flash Pro.

Without further ado, I have summarized in the below table the success rate I got on each subject, after checking for critical eye sharpness in Lightroom:

I had a feeling that the Z9 would do better (due to the more sticky autofocus) but I didn’t expect such a significant difference in success rate. If I had used a smaller aperture or turned-on the focus lights, both cameras would have performed better and it’s likely the gap would have been narrower. Well, at least we know which camera is the ultimate bokeh shots machine.

Z9 vs D500 autofocus: bottom-line

The Z9 has a more sticky autofocus which also benefits from a 90% coverage of the frame, meaning it made continuous focusing on a subject much easier than the well-established Nikon D500. When using both cameras in difficult conditions (open aperture, limited visibility and some low light scenarios, no focus light), the Nikon Z9 achieved critical sharpness in 70% of the shots, against 44% for the Nikon D500. 

The author, with the NA-Z9 housing (left) and the NA-D500 housing (right), ready for a side-by-side comparison in Kurnell (NSW, Australia).

Wide-Angle Autofocus performance

With the time I had for this review, I was only able to shoot wide-angle with the Nikon 8-15mm fisheye, mounted on the FTZ II adaptor. In saying that, I haven’t owned a rectilinear lens for years, so this was the most relevant comparison with my go-to wide-angle setup, which is the Tokina 10-17mm fisheye, behind a 100mm Zen glass dome, with the Nikon D500.

A friendly eastern blue groper follows divers in hope to be fed urchins. Nikon Z9 with 8-15mm fisheye, FTZII adaptor, Nauticam 140mm glass dome. 1/100^th, F/13, ISO 320, 2x Retra Flash Pro.

Over time, I know my D500 / Tokina 10-17mm combo has had close to a 100% success rate in a variety of shooting conditions, and my feel is that the Z9 / 8-15mm combo is in the same league. This is based on 14 hours of diving, covering 9 dive sites, in visibilities ranging from 6 to 20 meters, shooting a mix of reefscapes, schooling fish, gropers and excited fur seals.

Sponges at the entrance of Sydney’s Botany Bay. Nikon Z9 with 8-15mm fisheye, FTZII adaptor, Nauticam 140mm glass dome. 1/100^th, F/13, ISO 320, 2x Ikelite DS230.

On the seals colony, I spent about 2 hours producing 1300 shots (testing high speed bursts, more on that later), and there were only 5 out of focus, which happened when the Z9 focused on bubbles which were stuck on the dome. That is a 99.6% success rate with the seals, while working in the shallows, with surge, waves, and 12-15 meters visibility, except when I had foam all around me. If memory serves from previous years, my D500 is probably a bit less performant on seals, but having deleted all bad shots long ago, I have no way to tell if that was 99% or 95%. 

A fur seal quickly passes by, nearly touching my dome. Nikon Z9 with 8-15mm fisheye, FTZII adaptor, Nauticam 140mm glass dome. 1/200^th, F/11, ISO 400, AF-C Auto-area mode. 2x Ikelite DS230.

For the seals, the Z9 was set to AF-C, Auto-area mode and Animal recognition was enabled, picking up seals and occasionally their eyes. For general wide-angle work, my go-to mode was AF-C 3D, with animal detection left on by default, turned-off on the few occasions where it propose to pick something that wasn’t the subject.

There was one dive with horrendous conditions where the Z9 struggled to focus on the subjects, which were a few juvenile grey nurse sharks (aka sand-tiger sharks). I had misread the weather, and entered a narrow bay where a long surge wacked my knees on rocks, and kept stirring up the bottom. The visibility dropped to about 2 meters, with lots of floating particles and it was a cloudy day. In those conditions, the Z9 struggled to grab focus on the shark silhouettes, favoring some of the floating debris. I wonder if my D500 would have done better, but in hindsight, had I entered the water with my D500 and found those conditions, I would have aborted the dive straight away. I only stayed and persevered, because I had some testing to do.

A fiddler ray swims in between kelp, no issue focusing in the low afternoon light and 6 meters visibility. Nikon Z9 with 8-15mm fisheye, FTZII adaptor, Nauticam 140mm glass dome. 1/40^th, F/13, ISO 400, AF-C 3D mode. 2x Ikelite DS230.

Overall, my gut feeling is that the Z9/8-15mm does slightly better than the D500 with fast-paced action, and does as good of a job on less challenging wide-angle subject matter.

Let’s not forget a fisheye lens is the ultimate selfie machine! Nikon Z9 with 8-15mm fisheye, FTZII adaptor, Nauticam 140mm glass dome. 1/80^th, F/11, ISO 320, 2x Retra Flash Pro.

Wrap-up: autofocus performance

Autofocus was a big question mark and it turns out I like the Z9 autofocus performance better than that of my long-term, trusty companion, the D500.

A wider 90% frame coverage and the “stickier” autofocus tracking enable a new way of working: I position the focus point on my subject (AF-C 3D), and then re-compose and/or get closer, while having a high confidence that the subject will be tracked, even if it swims around and partially turns-away. On some fish species, animal recognition will kick-in, saving me the effort to position the focus point. These features allowed me work faster, which meant getting shots that I could have missed, with a DSLR.

In low ambient light, the accuracy was better than my D500 (with respective 105mm macro lenses),  a commendable performance.

I haven’t tried my F-mount Nikon 60mm AF-S macro lens with the Z9, but I anticipate it would perform even faster than the Z 105mm lens. If I had a chance to do some blackwater diving, that would have been my first choice, but I’d be curious to know how the Z 105mm performs for that sort of dive.

One negative is night diving with a red focus light, where I found my DSLRs still do a better job, but as soon as I revert to white light, the Z9 leads again. When using the optional Nauticam EMWL wet lens, the Z9 (and some other mirrorless cameras) currently suffers from a back-focus issue. See my EMWL article for details and ways to work-around that problem.

All-in-one, If I was given the choice between the Z9 and any DSLR for a dive trip, I would pick the Z9 every time, except if I was planning on using only the EMWL, for that whole trip. 

UPDATE October 2023: visit this post for my review of the newly-released Nauticam solution.

Z9 review index:

• Part 1: Prelude and Introduction
• Part 2: Autofocus performance (this post)
• Part 3: Viewfinder experience
• Part 4: Battery life, Burst shooting, Max Synchro Speed and Image quality
• Part 5: The Nauticam NA-Z9 housing
• Part 6: Conclusion