PetKal Offline Upload & Sell: Off
|
p.1 #1 · p.1 #1 · 5DIII/1DX Servo AF: "recipes" and a simple model | |
This simplified approach to Servo AF parameter setting is intended for folks who are struggling with the manual like I am, and who are keen on getting the drift of their new cameras' Servo AF performance fast and easy, while experts are not likely to find anything new here.
First of all, let me say that I believe that one should explore and evaluate a piece of gear on the type of target which represents their general photography interest as closely as possible. For example, if you like shooting antique porcelain dolls, then by all means test your gear on dolls.
Since I like to photograph birds here and there, it only makes sense that I'd test my new lenses and cameras on birds. However, therein resides a problem of test target availability. Even when available, no two bird flights are alike, the weather and light changes all the time, and a number of flights photographed may not allow sufficiently high confidence level in the results.
Canon have provisioned their 5DMkIII and 1DX user's manuals with six AF tracking "recipes" which are motion type based, and they are meant to help users select the most appropriate recipe for their tracking application, by selecting a recipe which comes close to the application. The recipes consist of different combinations of three tracking parameters: sensitivity, acceleration/deceleration and AF Pt. switching speed.
For brevity's sake, let us refer to them as A,B,C. In addition, users can custom assign different values to each of A, B and C.
Well, those recipes I found a bit confusing, thus I "dismembered" them by starting to look at A, B and C on their own. The ambiguity and overlapping of recipes is thereby reduced to a simplified question: given my particular tracking application, what should A, B and C be set to ? Now we come to the verification stage, or test shooting if you will. The issue is not only how well our 1st iteration of A,B and C settings work, but one should also do a bit of sensitivity analysis by permutating, say, A only, and observe the impact of it on tracking effectiveness.
Clearly, that approach can get a bit difficult with test targets such as non-captive birds.
There is one easy way of helping the test process: instead of chasing moving targets outdoors, we can use a moving lens on a fixed target at home.
A plastic action figure on my desk becomes my test target. A painting placed behind the target becomes my "busy" background. I shoot the target with a lens 135-200mm FL, and the target head should cover about 10-15% of the VF area.
Now let us try to do one little experiment with the setting A (tracking sensitivity):
Select centre AF point with 9 AF pt. expansion and, obviously, the camera should be in Servo AF mode.
(1) Dial Setting A to (-2)
Focus on the target (e.g. a plastic figurine). Move/play the lens over the target in any direction, back and forth, in such a way that all of the AF pts (including the expansion ones) clear the target periodically. Observe the AF behaviour, such as how long it takes for AF to jump from the target to the background.
(2) Dial Setting A to (+2)
Repeat the same process as above.
The net effect of the above test will be a fair difference in the observed AF tenacity or "stickiness".
In addition, you can have the lens fixed on the target, and then you pass an object quickly in front of the target and observe the resistance of AF to getting derailed from the target, again, as a function of different Setting A values.
There are similar ways in which a target movement can be simulated in order to get a better feel for the effect of permutations in B and C settings as well.
There you have it, lottsa writing here although the approach outlined is really very straightforward. Not perfect by any means, as one should always do real life testing as well, but the exercise described has helped me get a better feel for the new AF system.
|