I decided to do some simple AF-tests at Stockholm Marathon. The equipment used was the Nikon D300 with the 8 fps MB10 battery grip and the Nikkor 70-200VR lens.
I have shot many marathon races before, in Stockholm, New York and London, with good performance from the autofocus system and did not expect any particular problems.
The settings used was AF-C dynamic, 51 points active. As you may know the D300 let you customize the AF in a huge number of ways. I choose the 51 point setting for this test because it is default on AF-C and it has served me well on a varity of situations, from running dogs to soccer, alpine skiing and marathon races.
But for this application it is possible that 21-point, 9-point or even single point would have been better, since tracking is usually rather straight forward.
The MB10 was loaded with regular AA-batteries - I dont use the 8 fps option often enough to justify the high speed NC pack, at least not now. The batteries lasted the whole day for 800+ shots, and have not shown any sign of giving up yet, more than a week later. I shot jpegs, and switched Auto-ISO and Active D-Lightning off, since those settings lower the maximum fps from 8 to 7.5.
The D300/MB10/70-200 combo:
Usually I shoot short bursts of maybe 3-5 frames when I got the composition I want: But now I wanted to shoot long series to get a number on hit rate and also examine some Internet rumors I heard that the maximum frame rate on Nikon DSLRs slows down when tracking, even in release priority mode.
Side note - tracking
I thought this sequence was a little interesting. (If you click on the image, an 800×3600 pixels window will open.) As I wrote in my post about the Stockholm Marathon race, the third elite runner, (Phillip Bandawe, Tanzania) was obscured by slower runners and I could not get any good shots. However, look how the focus is stuck on him all the time in the six frame sequence despite the shifting light and disturbing foreground.
I took several long series in the range of 30-50 images. The sheer amount of images caused me some problems of how to present the results in a sensible way. First my idea was to make huge contact sheets in Photoshop, but there is a size limit in the software which I use the create these pages.
Here are the last 16 frames of a 34 images series of Said Regragui, Sweden, who finished third in the race. ISO 200, 1/1250s, F 4.5, 200 mm.
Here are 50 percent crops of the last 14 images (click to expand):
Looks good except 173 which is soft. The softness of the fingers on 174 seems to be motion blur, the whole image looks like this (click to expand):
You can see the images in a larger size here.
The last 10 images from a 20 frames series:
You can see the last 16 images in a larger size here.
All looks good to me.
The image below is from another 20 images series:
The series in reduced size:
To see the series larger, click here.
I have more series, but looking trough them is a tedius exercise. The difficulty in doing a rating are the odd borderline images where it can be hard to decide what is acceptable or not, but I think it is on the safe side to say that the hit rate of the AF is 90 percent or better in conditions like this. Even if you shoot at 8 fps.
Frame rate - stated and real
Now, how was it with the frame rate? Does it slow down while tracking? Lets look at the first series again and the exif-info. The image 0175 has an exif creation date that says 2009-05-30, 14.46:55.
If we count forwards to eight we find the image 0182, which has the same creation date including the same second - 2009-05-30, 14.46:55. In other words, 8 frames per second. No slow down.
From the last series. Image 0285. Exif creation date: 2009-05-30, 14.59:18.
The eight image during the same second, 0292 with the same exif: 2009-05-30, 14.59:18.
So in these sequences the stated frame rate of 8 fps is reached.