Nikon Z Autofocus and Subject Detection Notes

Entry Date: 4/6/2026

Just wanted to put together an article consolidating my observations and own personal experiences on the autofocus systems on the Nikon Z mirrorless system.

As a brief introduction to autofocus systems, all autofocus capable mirrorless cameras do that operation on the camera sensor itself, which brings a huge advantage (theoretically, more on this later) of not having to meddle with AF fine tuning adjustments to account for differences in lens-to-image-sensor and lens-to-AF-module light paths. Most on sensor AF systems started out as contrast detection AF, but when technology improved, on-sensor phase detection AF (OSPDAF) sites were implemented to improve the speed of AF acquisition, similar to how dedicated PDAF modules of the DSLRs work.

While more modern Nikon DSLRs have the ability to use live view for focusing, and all of them (except for the D780) are based on contrast detect algorithms. The Z mirrorless cameras, on the other hard, started out with having OSPDAF modules from the get go, using their experience from the Nikon 1 system (which also used OSPDAF modules for AF operations), and as such offered a much higher speed of acquiring focus as compared to DSLRs doing the same via live view. However, the first generation Z systems (the Z6 and Z7) were slower to focus or tend to hunt in dim lighting as compared to their DSLR brethren doing the same using their dedicated PDAF modules. Firmware updates improved the AF algorithms of the first generation bodies, bringing them to similar performance of their direct successors the Z6II and Z7II, and also including new staple features such as eye detection and subject detection.

One very important thing to note is how subject detection performance is not equal to autofocus performance. While most of us would likely relate one to the other, it is crucial to note that they are two different systems working in tandem so as to understand why some cases of misfocus occur. I will begin listing some of the points I have observed and explain them individually.

• Single Point AF performance in AF-S PDAF mode is fast, sensitive and accurate

The Z6, Z7, Z6II and Z7II do all perform very similarly with high levels of accuracy and repeatability in AF-S Single Point AF mode, even with their base firmware. In dimmer situations, AF-C Single Point AF may get lost every now and then, and this performance is definitely inferior to the dedicated PDAF sensor of the DSLRs. I do recommend using AF-S PDAF mode when using the cameras in dim lighting (such as in studio situations).

• Number of OSPDAF sites affect AF accuracy

The pixel density of the imaging sensor affects the amount of OSPDAF sites it has, and thus affect the accuracy of the AF algorithm. Generally, in good non-hunting lighting, a sensor with more OSPDAF sites will focus more accurately since it can discern between fine details better. For example, I personally find my Z7 focusing more accurately than my Z6 when using Single Point AF and Eye-AF since the actual AF module is smaller. However, it does also have a downside of being weaker in low light, and as expected the Z7 does start hunting earlier than the Z6 when light levels drop, and this is one of the reasons why many people complain about the Z7’s AF being inferior to the Z6.

It is also crucial to know that for two cameras having the same pixel density but different sensor sizes (like the Z7 and Z50), at equivalent focal length image framing, the smaller sensor size will have less OSPDAF modules and hence less accurate AF.

• Sensor architecture affects AF performance

The Z5 behaves quite differently from its siblings. In good light, Single Point AF does perform as expected and on par with the other Z cameras. However, the dimness threshold for it to start hunting is brighter than that of the other cameras, which is highly likely due to its FSI circuitry as compared to the BSI circuitry (and thus, improved sensitivity) of the other cameras.

• Newer CPU allows for more reliable AF-C performance, and better low-light AF-S performance

With the newer EXPEED 7 CPU, AF performance have been improved for cameras using older sensors (like the Zf, Z5II, Z50II). AF-C performance has been improved in non-subject detection modes (i.e Single Point AF or Dynamic AF), and in AF-S the low light sensitivity has been improved so they can focus in dimmer situations without resorting to contrast detect AF. However in good light, I personally find AF-S single point performance is similar to the older CPU in terms in accuracy.

• Newer CPU allows better subject detection

Similarly, a newer CPU can perform more number of calculations per second, and thus subject detection has been drastically improved with the use of a new CPU even when paired with an older image sensor. The Zf/Z5II vs the Z6II is a great example of this - they all use the same imaging sensor, but the subject detection of the Zf/Z5II is way faster and less hesitant than the Z6II.

• Sensor readout speed affects AF-C performance and subject detection

I guess it goes without saying that a sensor with a faster readout speed would be able to pass data to the processor faster for calculations, resulting in better AF-C performance as well as subject detection. In the current EXPEED7 crop of cameras, the Z8/Z9 with their stacked sensors perform the best here, followed by the Z6III with its partially stacked sensor, with the Zf/Z5II/Z50II closely behind.

• Pixel count affects AF-C performance and subject detection

An extension of the previous point, when the sensor architecture and CPU are kept the same, a sensor with higher pixel count would most likely have a slower sensor readout, hence affecting negatively AF-C performance. It does affect subject detection, especially when the subject is moving. However for more stationary or slower moving subjects, after initial subject detection acquisition the performance feels similar.

• Errors due to subject detection and autofocus systems working together

Using both subject detection and autofocus systems together would require the image sensor to pass the data to the CPU to do both tasks simultaneously. It goes without saying that a faster CPU (like the EXPEED7) would perform better than its predecessor (the EXPEED6) here.

Once a subject is detected and the area of focus has been determined, the AF operation is reliant on information from the OSPAF module, which may or may not be directly on top of the detected area. Nikon’s AF system does prioritize closest subject distance, so if the AF point is too large with respect to the detected subject (i.e. lower megapixel sensor on a slightly away detected eye), there is a chance that the OSPDAF module is not actually on the detected subject and may focus differently (like focus ending up on the tip of the nose instead of the eyes). As mentioned earlier, I find AF accuracy is better with more OSPDAF modules, which is further enforced by my own experiences of the Z7’s eye-AF working better than my Z6/Z6II. The Z8, having a similar pixel count to the Z7, with its stacked sensor and EXPEED7 CPU usually delivers the best AF accuracy when using subject detection modes.