ilkka_nissila Offline Upload & Sell: Off
|
Since the D7500 uses Advanced Multi-CAM 3500 II and has a higher resolution matrix metering sensor than the D7200, the AF is likely to be similar to that in the D750.
A situation where a subject is moving past obstacles such as branches is difficult for any autofocus system. I think the main problem here is that the D500/D5 AF is so fast and sensitive it will try to focus on branches and the subject in turn, unless the programming (Blocked subject response and Subject motion) can be used to tune the system for the subject in this case. I've never felt too comfortable relying on the camera handling a situation where the subject is temporarily blocked, so I avoid it and shoot the subject with a clean and direct line of sight where possible.
The flip side of the coin is that the D5 will respond very quickly to changes in subject motion where previous cameras would miss focus and take a long time to reacquire focus at times. I think the advantages are obvious even if it means the operator has to be precise as well to get the most out of the system. In some situations I've had D810 get 30% of the shots in focus at wide apertures whereas the D5 >90. The difference isn't always this big obviously. In my experience the D5's Multi-CAM 20k produces the largest differences in focus performance (compared to the old system) with fast primes (f/1.4, f/2) and in particular, the longer distances now are handled well as well whereas with D8x0 family cameras this was an issue, especially with the D800. I shoot a lot of verticals where the subject's face is under a linear point using the Multi-CAM 3500 family sensors and the focus results can be erratic, whereas the D5 with its cross type sensors does a solid job of focusing on facial detail near the edge of the frame even in an approaching subject situation. Unfortunately, these cross sensors at the last columns of the array, are only used as linear type with many f/4 and all f/5.6 lenses which reduces the advantage of the new system when using such lenses for this type of applications somewhat. It's still good but not quite as bombproof as with f/2.8 and faster lenses.
I shot some half-marathon runners on Saturday using 9-point dynamic area (it may have been six points actually since the actual number of points in the dynamic area (or group area depends on which columns one is using), it seemed to work as expected, not losing focus when the selected focus point was temporarily outside of the head. Since the runners were close (half body with 300mm) there is no way it could have done that without continuing to focus using the other points. I didn't get perfect results but a very high percentage of shots in focus still, using the 300mm f/4 PF. The 9-point area was about the right size considering the size of the runner's head in the frame and the precision with which I could hold the focus point on the target. In this kind of situations the D810 would have a tendency to focus on the hair because it has more texture than the face, but the D5 is able to see facial detail and focus on it with confidence.
I've used group area AF a lot in similar situations in the past as it is fast and reliable but with 9-point dynamic I could choose a subject in the middle of a group and keep that subject in focus for the duration of their approach instead of having to choose a subject which is already relatively close. It's more precise in that sense. Also if the subject wears a cap, group area AF would easily focus on the tip of the cap rather than the face, whereas with 9-point dynamic I didn't get that kind of errors. Still, I've found group area AF works great for distant subjects where the head is small in the frame and it is still my favorite choice for photographing action in general. The cap issue is definitely solved better using the 9-point dynamic though.
Following the tips on Nikon's NPS sports AF pages the D5 AF works as described at least for me. It operates on a very fast time scale and this is great for me, it allows greater precision of focus in complicated action situations. It's quite amazing to see only a few percent of focus rejects on a routine basis, using fast lenses (a single digit error rate is typical for walking or slowly moving subjects; for runners I would say I got 10-20% slightly out of focus using 9-pt dynamic on the face using the 300/4 PF). I can definitely imagine it failing to focus on a subject flying behind obstructions though. I don't think this is an easy problem to solve. If at all possible, I would shoot from a position with a clear view of the subject to avoid this problem.
|