One factor that is important when picking a camera is its auto-focus (AF) performance. Cameras differ on both accuracy and responsiveness (or acquisition) of focus and also tracking or keeping a subject in focus while the subject is moving. Accuracy means that the camera thinks it is in focus but it isn’t really (you end up with out of focus images even if the camera told you that it was in focus). Responsiveness is the lag between being out-of focus to being in-focus.
It is a challenge to assess a cameras performance as it depends on so many factors such as power from a driving processor or battery, available light and particular lens. Does the camera have a processor, or two? Does the camera perform in dim light and how does it perform in bright light? How can you separate the cameras native autofocus capability from that of the capability of the lens? But perhaps lens choice is irrelevant to separate, because you may choose a camera because of its lens anyway – and perhaps one should assess autofocus with the fastest lens available.
I looked at two sites to get insight on autofocus performance of Nikon, Canon and Sony cameras – both resources use scientific rigor when assessing performance - Popular Photography and Imaging resource and found conflicting results.
Popular Photography below (higher better performance) favors the autofocusing systems of the D3X and Sony A900 while imaging resource favors the Canon.
Data Source: [D300] [D700] [D3][D3x] [A900] [5DII]
Popular photography.. more to come.
Based on Imaging Resource Data:
Note that the Imaging Resource data above shows over a 2X difference between the D3 and D700 (D300) for the “Shutterlag, Full AF” numbers while the “pre-focus and shutter” numbers are pretty close between the cameras. We noticed that they use two difference lenses while testing between the D3 and D300/D700, namely the fast focusing 24-70 f/2.8 on the D3 and D3x but the slower focusing lens, the Sigma 70 f/2.8 EX on the D300 and D700. We asked the Imaging resource researchers about this issue but they believe that lens choice should not make a significant difference as they test iteratively or multiple times- so the lens is already in focus in the final measurement. In this manner they claim that they isolate the cameras performance the lens performance, i.e., they measuring the camera’s ability to ‘determine’ focus instead of the lens’ autofocus performance.
Personally, I challenge that assumption, for two reasons – first the data from popular photography seem to differ and second removing possible variables (such as lens choice and power) in experiments is more fair and it is only one way to eliminate lens choice as a question and that is to not use different lenses.
The focus engines between cameras (the D3 and D700) are the same, Nikon’s Multi-CAM3500FX system. The real difference is that the D3 has two processors giving it more ‘power’ while the D300 and D700 have only one. So the question is whether the D3 performs true parallel processing or not. Leveling the field between the cameras by adding more power to the D300/D700 or add battery grip with AA.
I offered imaging resource my equipment for retesting (my camera, lens and battery grip) for further testing – they responding by saying they would get back to me but so far I have not heard an additional response. Possible because of they are busy with the upcoming PMAi.
Review Site Data:
