p.1 #1OM-D Auto focus tracking/predicting failures
We'll today I ran some more test photos using the OMD with the new 1.5 firmware. I was hoping AF tracking might have been improved. I saw no evidence that it has changed, at least from a practical standpoint.
Frankly I think it is worse than I realized. I am more familiar with how the camera works (I think) now and was more careful in my settings. Such as: C-AF (TR), single point AF set, L fps. All shots at f1.8 using the 45mm lens so I can easily see DOF and if subject is sharp. A second series of tests using S-AF, single point AF, L fps and I was panning to keep subject in center focus square as best I could.
Subjects were runners and bicyclists coming toward me at a slight angle, almost straight on. In spite of Oly's claim that this camera with certain lenses has the worlds fastest AF, I had almost a 100% focus failure rate. I think it does not have the technology or processing power to predict the situation and adjust for it. In reality it may be tracking better than the previous firmware version but it cannot AF fast enough on a closing subject to take advantage of that better tracking. When shooting S-AF, holding the single focus point on the subject it made no difference. Actually, I think this test is pretty severe and maybe it is testing one of its weakest points. But the OMD is not useful for this type of photography.
I should also add that when in C-AF (tr) mode it did not fire off the shots as quickly as in S-AF. I would assume this means that it only shot when it "thought" it was in focus not just because I had the shutter depressed.
Having the worlds fastest AF camera does you no good if it is not fast enough, or cannot predict movement enough to get a sharp photo much of the time.
Until told otherwise I do not believe any of my settings are wrong or causing this poor performance.
Oct 05, 2012 at 12:55 AM
jcharity
Offline
Activity
[X]
Details
p.1 #2OM-D Auto focus tracking/predicting failures
I've shot automobiles and bicyclists coming toward me with no problem.
p.1 #3OM-D Auto focus tracking/predicting failures
Use C-AF, but not the "TR" mode. It works very well (not perfect, but very well) with the following settings (you should save .jpeg as well as raw files even if you'll just discard the .jpegs, like I do):
4fps (Low burst)
C-AF NOT C-AF+TR
Center point
VIVID color mode (even if you shoot raw) S +0, C +2, RGB +2
p.1 #4OM-D Auto focus tracking/predicting failures
httivals wrote:
Use C-AF, but not the "TR" mode. It works very well (not perfect, but very well) with the following settings (you should save .jpg as well as raw files even if you'll just discard the .jpgs, like I do):
4fps (Low burst)
C-AF NOT C-AF+TR
Center point
VIVID color mode (even if you shoot raw) S +0, C +2, RGB +2
That's really interesting... What does the JPG do in this setting? Does that actually improve AF performance?
p.1 #7OM-D Auto focus tracking/predicting failures
yes, the jpeg settings improve the af performance. i did some internet searches and found these tips from someone who had done a lot of testing, tried them, and they work very well. you need to keep the subject within the selected af point.
p.1 #11OM-D Auto focus tracking/predicting failures
httivals wrote:
Use C-AF, but not the "TR" mode. It works very well (not perfect, but very well) with the following settings (you should save .jpg as well as raw files even if you'll just discard the .jpgs, like I do):
4fps (Low burst)
C-AF NOT C-AF+TR
Center point
VIVID color mode (even if you shoot raw) S +0, C +2, RGB +2
p.1 #12OM-D Auto focus tracking/predicting failures
Bruno - your suggestion of using S +2 makes sense. It was probably a typo on the forum where I read it (I don't remember where), but your suggestion of S +2, makes perfect sense.
p.1 #13OM-D Auto focus tracking/predicting failures
FlyPenFly wrote:
I wonder if it might be from the contrast AF system using JPEG preview to do it's calculations?
Yes, there was an explanation given along those lines I believe from the forum where I gleaned the information.
I took a lot of images at my son's soccer practice, using the 75mm f1.8 all the way open and the success rate was impressive - probably about 2/3. The biggest determinant of success was how large the person was in the frame; the closer to the camera the better. That's probably because when the boys were smaller in the frame, the C-AF system would often lock on the grass rather than the boy.
p.1 #14OM-D Auto focus tracking/predicting failures
Thanks to all who have or will comment on this thread I started. I will test these various setting permutations and report back ASAP. I'm sure others will do he same too. Glad to know I don't have a unique defective OMD body. Love that 45!
p.1 #15OM-D Auto focus tracking/predicting failures
The other thing that helps a lot with action is to have a fast shutter speed. I find that, for example, for my son's soccer games, any shutter speed less than 1/1000, I'm much more like to have motion blur. I get much sharper images at 1/1000 and can see improvements in the % of razor sharp images all the way up to 1/2000.
p.1 #16OM-D Auto focus tracking/predicting failures
Does selecting the smaller focus target "box" help to avoid the "focusing on the grass rather than the player" problem, or is that offset by making it even more difficult to hold the target on the player?
p.1 #17OM-D Auto focus tracking/predicting failures
CalW wrote:
Does selecting the smaller focus target "box" help to avoid the "focusing on the grass rather than the player" problem, or is that offset by making it even more difficult to hold the target on the player?
That's the issue - can you hold the target on the player? I'll try with the smaller box when I get a chance and see which results are better. The internet posting that I read said that one should choose the 14x (smallest) box, but that seems a bit too constrained to me.
p.1 #19OM-D Auto focus tracking/predicting failures
What about the IBIS setting for horizontal focus tracking at average/modest speed? Should it be OFF, IS-1 or IS-3 (horizontal panning)?
httivals wrote:
The other thing that helps a lot with action is to have a fast shutter speed. I find that, for example, for my son's soccer games, any shutter speed less than 1/1000, I'm much more like to have motion blur. I get much sharper images at 1/1000 and can see improvements in the % of razor sharp images all the way up to 1/2000.
p.1 #20OM-D Auto focus tracking/predicting failures
Ok, back with some quick test results using the the recommended settings offered here by httivals. Those included the following:
1) Set to RAW+JPG (small, in my case) or what Oly calls SN+RAW.
2) Picture mode #2 Vivid
3) Sharpness +2
4)Contrast +2
5) Singel AF point.
Shutter set to C-AF.
We'll I'm pleased to say that these settings seem to have resulted in a huge increase in successful in-focus shots of moving subjects. The shutter would not always fire so it seems it has to "lock" focus to allow it to do so. There is no question that the closer the moving object got to the camera the more likely the subject was sharp.
I also tested it using shutter setting of C-AF (tr) and my impression is that this also worked quite well. So, using the cameras auto tracking feature did not disable the better autofocus that was achieved with these settings.
I also took some shots using S-AF. My impression is that these did not have a high a success rate but were better than my previous results.
I have not tried other permutations of these settings to see how they work. I question the need for picture, sharpness and contrast to +2 but as long as you are only interested in the RAW image it really does not matter anyway.
All my shots were at f1.8 using the 45mm lens with IS 1 and at 1/4000th sec. So there was little DOF and extremely limited motion blur. It was very easy to see what was or was not in focus. I hope this is helpful to those who have also been concerned with the issue. I'll be interested in knowing if others using much more telephoto lenses can also get such improved results where it would be essential for accurate focus to have any hope of a useful image.