Ziggy99 wrote:
Yes, I'm sending the rig back for testing. This focus issue only arose over the last two sessions.
Note that one of my test shots got the focus right in the VF using DMF, and the achieved focus was still well behind.
I think that you are wasting your time with testing. Something is very wrong with your gear. Could it have been dropped without your knowing about it? In any case, just send it in and get it fixed.
I've had six Sony 7 and 9 series cameras and more than 20 Sony lenses. Everything has always been tack sharp with AF, MF and for a very wide variety of scenarios. AF is one of Sony's strongest features.
Thanks for the assessment.
I now know that the lens back focuses with my A7R III as well.
Never been dropped. Never been in anyone else's hands.
The problem started two sessions ago, at 10 months old.
I haven't heard of the Z's doing it (but don't read much about them).
The Panasonic G9, which is CDAF only, can miss a lock on quite a large subject if there's something behind and near that the AF area overlaps. It doesn't quickly shoot to infinity. I gather that the latest firmware for that camera offers the option of near or further lock priority.
The Canon uses dual pixel AF which AFAIK is a kind of pixel-level PDAF so there goes my theory that this is a contrast detect weakness.
Ziggy99 wrote:
I haven't heard of the Z's doing it (but don't read much about them).
The Panasonic G9, which is CDAF only, can miss a lock on quite a large subject if there's something behind and near that the AF area overlaps. It doesn't quickly shoot to infinity. I gather that the latest firmware for that camera offers the option of near or further lock priority.
The Canon uses dual pixel AF which AFAIK is a kind of pixel-level PDAF so there goes my theory that this is a contrast detect weakness.
No, that sounds very plausible to me, not least because pre-focusing is what I regularly do with my A9 and A7R III to avoid this happening. It loses me shots because birds often don't hang around or there's nothing nearby to get and transfer a focus from.
As to why, I'm still inclined to think that CD is implicated too. Hybrid AF starts with PD to get in the ballpark and then fine-tunes with CD, but I'm not seeing any pulsing. Both are failing in this circumstance. I shoot with a Nikon D500 as well which is about as evolved as you can get in PDAF and it doesn't suffer this weakness. If record shots have to be got, the Sonys get left at home.
I find the small, near, grossly OOF subjects to be a point of AF failure on all my MILCs (Z7, Z50, A9, A9II, A7RIV, R5)....no one is immune...and there is no vaccine
I've had repeated Sony failures with things like a 4" diameter tree trunk, with textured bark in good light, or a 2" branch on the diagonal against a bright sky.
This has to be an algorithm failure and CD is all software so that has to be a suspect.
Ziggy99 wrote:
Maybe if billions was involved there would be.
I've had repeated Sony failures with things like a 4" diameter tree trunk, with textured bark in good light, or a 2" branch on the diagonal against a bright sky.
This has to be an algorithm failure and CD is all software so that has to be a suspect.
One way to improve the situation might be to carry out the very first pdaf operation at a very small aperture. Say f/16 on the A9. Then move the focus to get close before continuing pdaf operations at the desired aperture according to settings.
Using a small aperture wouldn't be very accurate and may have difficulty in low light but, even if this first pdaf operation failed, the camera would resort back to its default aperture and work as now with current accuracy.
Whether f/16 would provide enough additional depth of field to be worthwhile could be tested by setting that aperture and the Aperture Drive in AF setting to silent. At least I think that fixes the aperture as set rather than opening up to focus.
CAF is using hybrid AF, so PD first and then CD. Is the fail happening at the PD attempt or later, at the CD attempt? Or both?
When CD is having trouble you see pulsing. I've never seen that in this form of failure. I can see a blur or I can see a split-second of detail and then the rack out to infinity or something, with no lock recorded in EXIF.
At f16 on a 600mm lens, there's 25cm of DOF around a subject 10m away.
As an aside, this morning I was experimenting on narrow garden stakes, 25mm/1" wide, in a pond that Welcome Swallows like to perch on. Lock failures are normal and the only way to get focus is by transfer ('pre-focus'). When I got a lock I raked the stake from top to bottom in BBF mode. The top few inches were painted blue and the rest was a weathered mid-grey. A lock on the blue would be lost when dropped to the grey. That's a contrast loss and seems to me to be a classic CD failure.
When I rake the stake in tracking mode the box tries to follow the edge.
Nah, 6 weeks and it's still in Japan. Sony Australia did agree to my request for a loaner though and promptly shipped out a 600/4. (I'm not a member of their pro scheme).
You might be right about the A9 not doing CD in CAF. It takes more time. But A7info.exe shows two types of focus: PD and another (which may just be what the camera settled on rather than CD). In the case of the A7R II Sony did say that hybrid applies but not whether it was limited to SAF.
Ziggy99 wrote:
Could work, if the bird hangs around.
Depends on when the fail happens.
CAF is using hybrid AF, so PD first and then CD. Is the fail happening at the PD attempt or later, at the CD attempt? Or both?
When CD is having trouble you see pulsing. I've never seen that in this form of failure. I can see a blur or I can see a split-second of detail and then the rack out to infinity or something, with no lock recorded in EXIF.
At f16 on a 600mm lens, there's 25cm of DOF around a subject 10m away.
As an aside, this morning I was experimenting on narrow garden stakes, 25mm/1" wide, in a pond that Welcome Swallows like to perch on. Lock failures are normal and the only way to get focus is by transfer ('pre-focus'). When I got a lock I raked the stake from top to bottom in BBF mode. The top few inches were painted blue and the rest was a weathered mid-grey. A lock on the blue would be lost when dropped to the grey. That's a contrast loss and seems to me to be a classic CD failure.
When I rake the stake in tracking mode the box tries to follow the edge....Show more →
I didn't mean for us to try to do that manually! I meant that Sony could modify the firmware to work this way.
