Autofocus

The D7200 brings an ostensibly updated AF module to the line: the Multi-CAM 3500 II. While providing the same wide coverage and high density of the 51-point system in the D7100, the module brings enhanced sensitivity to every AF point. All points now work down to -3EV, which makes the D7200 a class-leader in this regard (the D7100 was rated down to -2EV). This may not sound like a big difference but, as we saw with the D750, this is enough to mean that the system will continue to work in very low-lit social situations down to the point that you're likely to need to use flash - meaning it's unusual to hit a point that the camera will stop trying to focus.

The use of a traditional phase-detect module means the D7200 has snappy focus, and can also track along the depth-axis (z-axis) reliably and quickly, even in continuous shooting. With these dedicated modules, we've come to expect very good performance at (depth) tracking an approaching or receding subject with a single, static AF point, whether it be a Nikon, a Canon, or a Sony DSLR. Where we see significant differences between systems, though, is in subject tracking performance - where the camera automatically selects an appropriate AF point to stay on a moving subject. Much of our discussion below, therefore, focuses on this aspect of AF performance.

Subject ('3D') tracking

Focus subject tracking (the ability to keep track of which subject the camera is supposed to be focusing on) is really rather good. The Nikon D7200 uses a relatively modest 2016-pixel color-aware metering sensor to give it an understanding of the subject it's supposed to be following (the one under the active AF point when you initiate focus), yet this is often enough for the camera to be able to latch onto a moving subject.

Whether it can correctly identify the subject distance and refocus in time is something we'll look at further down this page, but purely in terms of identifying and keeping track of its subject, the D7200 does well, as can be seen in the video above. The initial red flash indicates that the shutter has been half-pressed to specify a target. After this point you can watch the camera attempt to switch to the focus point nearest the original target (in this case: the eye of the mannequin). Despite multiple objects in the scene at different depths close to the initial subject (which might confuse a more primitive tracking system), the camera largely remains on target. Not only does it automatically track the eye of of the mannequin despite drastic scene movement and distance change, it even tracks the correct eye.

In continuous 6fps shooting (0:20 in the video above), the system did get confused in one of the three runs we show above, but in the other two, it generally remains near an eye of the mannequin. This is a really tough test, with drastic changes in all three (x, y, and z, or depth, axes), and the constant 'black-outs' caused by the mirror going up at 6fps likely hurts the ability of the camera to accurately follow the subject. The fact that the system does as well as it does is commendable.

All in all, subject tracking performance is impressive, especially considering the low resolution of the metering sensor which, at best, should only give it a minecraft-esque, blocky understanding of the scene. The almost uncanny accuracy of tracking speaks to the benefit of a secondary image sensor that 'sees' the scene to understand the subject and its movement. Mirrorless cameras, with their constant scanning of their image sensors, also perform very well in this particular regard, but amongst DSLRs, Nikon offers the most usable implementation of subject tracking we've seen to date.

Continuous AF performance

Next, we take a look a bit more holistically at continuous AF performance, not only testing subject tracking, but also how well subject tracking combined with the phase-detect AF module actually focuses on the intended target.


1

2

3
v

6

v

5

< 4

7
>

8

This sequence shows how well the autofocus can follow a fast-moving subject, even while the camera is also being moved. The car is a fairly easy-to-distinguish object, so should be relatively easy to identify and track, but the consistency of the focus is still impressive, given the distance of the car relative to the camera is fairly static at first, then decreases quite rapidly (making it hard for the camera to predict).

This sequence was shot in focus priority mode, meaning the camera only shot an image when it was confident of being in focus but, as you can see from the fairly consistent changes in the car's position between shots, there aren't noticeable gaps during which the camera is failing to focus or shoot.

1
2
3
4
v
8
v
7
6
< 5
9 >
10
11
12
v
16
15
14
< 13

This sequence was shot to assess how well the camera copes with both changes of depth (z-axis) as the rider approaches and some (x-axis) movement. Shooting this subject allows us a good degree of control over both variables.

The camera performed extremely well when using the AF-S 70-200mm F2.8 II or 300mm F4 PF lenses: the focus point was able to lock onto the riders face or jersey then track it with a good degree of consistency, regardless of the amount of x-axis movement. The tracking of the subject would become a little less certain as the rider passed through a shadow region but with the focus persistence set at the default level (3), it didn't panic and try to refocus dramatically.

Most impressively, while the hit rate for release priority mode was good (around 88% usable and 55% very sharp), it got even better when the camera was changed to focus priority (Custom Setting a2), without a significant impact on shooting speed. In focus priority we regularly maintained a shooting rate of around 5.5 frames per second, but increased the very-sharp rate to around 75%.

Performance with the Nikkor 18-140mm was less impressive. Oddly the difference wasn't so much a question of focus speed and focusing to the correct distance but of the camera seeming less willing to change AF points to follow the subject. This meant the hit rate with this lens was considerably lower than when using the higher-end sports lenses. We suspect this is because the subject was making up a smaller proportion of the frame (meaning it's harder for the low pixel-count metering sensor to distinguish the subject), suggesting the system works best with larger targets.

Continuous shooting

Of course, none of this subject tracking and fast-focus prowess is terribly useful if the camera can't shoot fast enough. The D7200 will shoot at up to 6 frames per second in standard, 'DX' mode, but this falls to 5fps if you want to shoot 14-bit Raw or increases to 7fps if you're willing to use a 1.3x crop from the sensor.

File type Frame Rate Buffer depth Buffer full rate Time to empty buffer
JPEG Normal ~6.1 fps >100 frames* N/A N/A
JPEG Fine ~6.1 fps ~47 frames ~3.0 fps ~ 8 sec
12bit Raw + JPEG Fine ~6.1 fps ~12 frames

~1.3 fps

~ 9 sec

12bit Raw ~6.1 fps ~28 frames ~3 fps ~ 5 sec
14bit Raw + JPEG Fine 5 fps ~10 frames ~1.1fps ~ 9 sec
14bit Raw 5 fps ~17 frames ~2.3fps ~ 6 sec

*Adjustable, up to 100, with Custom Setting d3

These figures are a great improvement over the D7100, which could shoot just 6 (14-bit) Raw files in continuous mode. The buffer fills quickly if you try to shoot Raw + JPEG, but if your workflow allows it, you can shoot for a descent amount of time in Raw (especially if you switch to 12-bit).

All tests conducted with a SanDisk Extreme Pro UHS-I card, speed class U3. The results were very card dependent, with inferior results using a UHS-II U3 card. Even with the fast UHS-I card we would occasionally get image bursts with significantly fewer images than the specs suggest, before the camera slowed down.

Face detection

The D7200 will face detect when shooting in live view mode but Nikon makes no such claims for viewfinder shooting. Nikon's full-frame cameras have higher pixel-count metering/AF color sensors (91,000 pixels), which allow for a better understanding of the scene and, therefore, actual face detection in viewfinder shooting. The older D7000 and D7100 models would claim they could detect human subjects against a background, whereas the D7200 manual simply states: "in the case of portrait subjects, the camera is able to distinguish the subject from the background". This understandably has led to some confusion surrounding face detection in viewfinder shooting.

So what is the take-home here? In our testing, the D7200 will prioritize faces if you let it choose the focus point ('Auto-area AF'). However, since it has a relatively low-resolution sensor, it's not great at it: the face has to be relatively large in the frame and free from anything so complicated as glasses for the camera to identify it. Skin-toned subjects in front of a face can also confuse the camera (a hand, for example). Although Nikon may feel it's not effective enough to be worth promoting, it is there and it does, to a pretty good degree, work. Just don't expect it to work anywhere nearly as well as it does in Nikon's full-frame bodies.

Focus accuracy

I mentioned in my shooting experience that I had some problems with AF accuracy. I should make absolutely clear that I don't think that the D7200 has a specific problem. With most lenses, the degree of mis-focus tends to be minor and within the range of AF fine adjustment (though this may vary by subject distance, zoom position, or focus point, making adjustments potentially ineffective even counter-productive).

However, because the camera is using a separate sensor as a proxy for correct focus, rather than directly measuring it, we feel confident saying the D7200 simply can't match the focus accuracy of mirrorless cameras. To be fair, though, neither can any DSLR with a dedicated phase-detect AF module.

Live view focus

Like many DSLRs, live view focus isn't the D7200's strength. The main thing holding the Nikon back is that most of the company's lenses simply weren't designed for contrast detection, which requires small, light focus elements and a motor design that can be moved in fine steps and can change direction quickly. Without any on-sensor phase detection system, the D7200's live view focus is slow by most contemporary standards. Not unusable, by any means, but significantly less slick than its primary, through-the-viewfinder focus system.

Ironically, the camera exhibits almost the opposite problem while shooting movies: its AF is too quick. The camera doesn't appear to slow the autofocus down during movie shooting, which means it jumps dramatically, rather than sliding smoothly from point to point. Add to this a very pronounced focus wobble as it performs the inevitable overshoot-and-correct demanded by contrast-detection AF, and you'll find you're usually better-off manually focusing in video. In both respects, a poor performance by contemporary standards, when you consider the likes of the Sony a6000, or Canon 70D - both of which offer quick phase-detect autofocus in movies.