[UncleVanya]

Apologies for waking a sleeping thread, but I'm recently started seeing a similar effect as the OP on my K-1 which is mostly paired with a DFA28-105. Observed with recent handheld landscape use. Lower left of image looks as though lens is badly decentered but only on a small percentage of images. Others look fine. Tripod shots with self timer (IBIS auto off on timer) seem ok. Horizon correction off. Was thinking I've broken a cam or guide in the lens that caused intermittent misalignment but then noticed a couple of landscape shots with a FA31 that were soft along the extreme left edge, rest of image fine. So now thinking what else could be the cause. Struggling to see how possible shutter shake could only affect one side in landscape orientation only. Currently questioning whether a IBIS fault could be the cause - sometimes allowing sensor on left side (particularly lower left) to move off the focal plane?

If the OP Coven is still monitoring this thread, did you ever find an answer to your problem? To other members, has anyone experienced IBIS faults/failures?

Still evaluating how frequently I can observe this problem - back from a five week South to North drive across Australia (and back) - 9000 kilometres of driving and now over 4500 photos to review/process (ugh!).

See post #10. I would start with two shots with identical setting other than ibis.

I would also suggest that gravity can act differently on a sensor based on orientation that perhaps has a broken or faulty ibis mount on one corner. The test to add fuel to this would be to invert the camera and make some shots upside down and see if they were identically impacted. Also reversed portrait orientation shots might help. This doesn’t eliminate anything but helps add data.

[Wheatfield]

I'd be looking elsewhere for partly soft images. It's not like as if faulty IBIS can only affect part of the sensor. For faulty IBIS to only affect part of the image the sensor would have to be flexible. It isn't.

[UncleVanya]

I'd be looking elsewhere for partly soft images. It's not like as if faulty IBIS can only affect part of the sensor. For faulty IBIS to only affect part of the image the sensor would have to be flexible. It isn't.

Sony had an issue where the ibis attachment points broke under stress and the sensor could move a corner out of the correct plane which clearly wasn’t bending but the magnitude of the alignment drift was higher in one corner.

[southlander]

Sony had an issue where the ibis attachment points broke under stress and the sensor could move a corner out of the correct plane which clearly wasn’t bending but the magnitude of the alignment drift was higher in one corner.

That’s the kind of fault I am thinking about. On the problem shots I seen so far, it would appear the lower left front focuses relative to the rest of the frame. Often I have a framing element in lower left eg a close by tree or bush which can potentially obscure the issue so I will need to pixel peep to look at the more distant (infinity focus) composition elements near said framing elements to see if they are in focus or not. I will be spending more time working through the 4500 images I took over five weeks to see if I can identify when the issue first became apparent and if there is any obvious recurring usage situations. At this stage, tripod shots (IBIS off) look ok which is why I am pondering potential issues beyond the lens.

I want to be reasonably sure of whether it is lens or camera before I send anything off for repair, as based on prior experience, more optically complex issues get shipped off by the local importer to Japan for service. The K-1 and DFA28-105 are my work horses so don’t want to loose both for potentially three months or so. On the other hand, having items sent to Japan means they get worked on by Pentax’s own technicians which is confidence boosting in the quality of service.

Until the last few days I’ve never really though about how the sensor is held steady with high accuracy in the fore/aft z dimension while being able to move with very little friction in the x and y plane and also able to rotate to counteract roll.

I will report back once I’ve reviewed more images.

[kypfer]

Probably totally unrelated … but I just remembered I experienced what could be interpreted as "partial softness" in some images when the software I was using invoked excessive noise reduction.
Although I was using the Digital Camera Utility, a couple of others, using different software, have experienced similar issues.
Whether it was down to the software "over-reacting" or the camera passing "optimistic" data in the EXIF I don't know, but turning down the noise reduction in the software resolved the problem

[southlander]

Probably totally unrelated … but I just remembered I experienced what could be interpreted as "partial softness" in some images when the software I was using invoked excessive noise reduction.
Although I was using the Digital Camera Utility, a couple of others, using different software, have experienced similar issues.
Whether it was down to the software "over-reacting" or the camera passing "optimistic" data in the EXIF I don't know, but turning down the noise reduction in the software resolved the problem

Thanks for that information. Images are RAWs processed in Lightroom. Vast majority of images I took on the trip are daytime using ISO 100 or 200. Zero luminance NR applied, colour NR dialed right down to 5 from the default of 25.

[kypfer]

Thanks for that information. Images are RAWs processed in Lightroom. Vast majority of images I took on the trip are daytime using ISO 100 or 200. Zero luminance NR applied, colour NR dialed right down to 5 from the default of 25.

My first "double-check" would be to see if the issues are replicated when saving the image as an in-camera JPG.
Even though the results might not be exactly what you want, it might help isolate the source of the problem to either the camera or something weird in the program.

[Wheatfield]

Sony had an issue where the ibis attachment points broke under stress and the sensor could move a corner out of the correct plane which clearly wasn’t bending but the magnitude of the alignment drift was higher in one corner.

I dunno. My understanding of how Pentax IBIS works is that there are no attachment points for the sensor. It floats free against a backing plate that is part of the body for all practical purposes and is held in place by magnets.
I just don't see how a sensor could get out of alignment in one corner only. If nothing else the opposite corner would have to be out of alignment an equal amount in the opposite direction unless the sensor was in two pieces, something I am sure would be pretty obvious.

[UncleVanya]

I dunno. My understanding of how Pentax IBIS works is that there are no attachment points for the sensor. It floats free against a backing plate that is part of the body for all practical purposes and is held in place by magnets.
I just don't see how a sensor could get out of alignment in one corner only. If nothing else the opposite corner would have to be out of alignment an equal amount in the opposite direction unless the sensor was in two pieces, something I am sure would be pretty obvious.

The way I see it in my head is that the moving shutter slit and the motion induced vibration might cause misalignment in different degrees during exposure. So while the angle of the whole sensor might be out of whack, only part of it might be exposed at that moment. However I realize this would vary depending on the shutter speed.

[southlander]

Thanks everyone for your responses. As indicated, I need to work though more images from the trip to see if there is a common thread to where I see a problem. One thing that has sprung to mind is whether a particular focal length is involved which might suggest one section of the guides for the 28-105's lens groups is faulty.

Yes, I akso considered the effect of shutter speed on the issue. Below flash sync speed, I would have expected a sensor related problem to be more widely spread over the image, above sync speed only a portion of the image is exposed at any given moment so an IBIS issue could potentially be isolated to part of an image. That is also something I will look at.