Archibald wrote in post #19484957
With my R7, I'm noticing that the focus sometimes goes to high contrast elements near the eye instead of the eye. The eye algorithm finds the eye but then illogically focuses on a nearby item. This happens most with distant subjects that are low contrast with high contrast stuff near the eye. It is inconsistent though.
With better subjects the AF works very well.
In your examples, it looks like the focus is drifting to the high contrast background.
Check out this thread.
https://photography-on-the.net …read.php?t=1528487&page=1 Thanks for the link to that thread, it's good to see other users experiencing the same issue. The main points from that are that the selected focus point is not necessary what the camera is focussing on, and the focus point square is much bigger than it appears in the viewfinder. I still wouldn't expect focus to drift off though, seeing as it's done the hard part and found the eye in the first place, especially when the subject is stationary.
siginu wrote in post #19485359
Look at the uppermost twig on the right side of the bird in the last photo, it's suddenly in focus. It's just focus shift in this case.
It's just hard to get my head around it losing the focus on the eye when it had originally locked onto it. That shouldn't happen with a stationary subject in the space of half a second, should it?
Leigh wrote in post #19485326
This is interesting.
I downloaded the files and processed them through Adobe's DNG Converter to .dng files, and viewed them in Photoshop CS6, and I didn't note any specific blurring?
Thanks for taking the time but, really? Did you see the pigeon in IMG_95145, and the parakeet (IMG_9301)?
Overread wrote in post #19484968
This can be tricky because if you check the sharpening tabs on almost all RAW software, they all apply some sharpening by default. A RAW shot with zero sharpening often looks soft even when well taken. Thus even though sharpening is a destructive element to image data and typically saved till the last stage (or done on separate layers). With RAW shots you have to have some applied to have a shot so that you can functionally make choices regarding which shots are sharp and which are soft and to work with the shot in general.
So yes between different RAW processing packages you will see some variations in sharpness, partly the result of the different code they used to process and partly because of different default base values.
I do use Faithful mode in camera, but even then, I'd be surprised that RAW files would not be sharp straight from the camera? I see pro bird photographers on youtube doing the initial assessment of their images in Faststone (eg the brilliant Jan Wegener) which shows perfect sharpness. I can't do that, I have to import into LR to get a better representation of sharpness. Maybe he sets his in-camera sharpness high? I can have a play with the sharpness settings if that's what I need to do.
---------------
What I suspect is, now that I've discovered eye tracking, I'm using it almost exclusively (because it's so amazing), even with static subjects. And the message I'm starting to get is that this is not advisable, as the AI servo mode combined with tracking can be erratic with stationary subjects. WIth my DSLRs I was shooting one shot focus and recompose pretty much all the time, which is probably a more reliable method.
Is the general rule to not use AI servo and/or tracking on stationary subjects, regardless of whether or not the background is busy?