Jeff
Offline
• • • • • •
Upload & Sell: On
|
 p.31 #4 · ['NEW Fix' UPDATE!] - MkIII AF still broken | |
OK, so here's the scoop:
I shot 7 buffer-filling sequences of both my kids running, biking, and scootering at me. Center-point, no expansion, AI Tracking Sensitivity at default (as well as everything else). My two sons were wearing very different clothes, one contrasty, the other not. There are three numbers; the first is the # clearly in focus, the next # reflects those that are marginal (i.e. could go either way), the last is the total # of shots in the sequence.
Seq. #1: 15-18/21 in focus, 71-83%
Seq. #2: 19/20 in focus, 95%
Seq. #3: 12-17/20 in focus, 57-85%
Seq. #4: 15-19/20 in focus, 75-95%
Seq. #5: 15-18/20 in focus, 75-90%
Seq. #6: 11/17 in focus, 65%
Seq. #7: 14-16/20 in focus, 70-80%
--------------------------------------------------
TOTAL: 101-118/138 in focus, 73-86%
Now that I would consider to be acceptable, compared to the consistent 50-70% of my previous body. No consistent front- or back-focusing, and I didn't critically evaluate for AF point vs. point of focus (i.e. I eyeballed it, since I was using the center point). Since the results were basically good, I did take the time to throw out clear cases of 'user error', which would up the 'hit rate' a bit. For example, note that in Sequence #6 the hit rate is way down (for that particular 'contrasty' son), because I lost the subject, and the camera went grossly OOF to the front, but it was not able to recover (all the last 6 images), even after the point got back on the subject. This also happened in sequence #7 (dang squirrelly scooters!), but the camera did 'slowly' recover focus back to the subject over 4 frames.
Unfortunately, I am reserving the right to pass judgment on any potential 'focal plane' issues until I've had adequate time to test appropriately for that. Completely different issue, requires a different test; I shot a few out the back, and let's just say I have some questions.
Anyway, in my mind this is how this camera should perform, and I can say without hesitation that the AI Servo performance of this fresh-off-the-line Blue Dot (S#55886x) is at least an order of magnitude better than my early-production (S#50285x) MkIII. If my first camera had performed like this out of the box, I likely never would have noticed any of the secondary issues that I will now investigate, mostly out of curiosity.
Regards, to all.
Jeff
PS: Here are two examples, the first a typical 'in-focus' shot viewed at 50%, the second a typical in-focus shot at 100%:
|
