IBIS with 14-30 lens - it impacts edge sharpness

[pdk42]

TL;DR - IBIS at slow shutter speeds might give you sharp shots in the centre of the frame, but the edges could be blurred.

Shooting with my 14-30 on the Z7, I'd noticed sometimes that edge sharpness was pretty awful, but at other times it looks fantastic. I thought it might be field curvature, but a quick test showed that at f8 or f11 there is way more DOF than needed to counter any curvature. So, I turned my suspicions to the IBIS (I'd noticed this before on Olympus cameras). So, here are two quick shots of the garden (yard), one handheld, one on a tripod - both 1/10s at f8 at ISO 200 (the light had changed slightly between them so there is a slight brightness difference, but I wanted to have exactly the same exposure). Excuse the ongoing work on the patio!

Subscribe to see EXIF info for this image (if available)

Tripod by Paul Kaye, on Flickr

Subscribe to see EXIF info for this image (if available)

Handheld by Paul Kaye, on Flickrkr

Here are the centres - handheld on the left:

Subscribe to see EXIF info for this image (if available)

And here is the left edge - again, handheld on the left:

Subscribe to see EXIF info for this image (if available)

Q.E.D.

 

[gryphon1911]

I’ve seen this reported in other blogs and such as well. I think the issues stem from the amount of distortion correction the camera has to do at the edge of the frames. Couple that with the micro adjustments for the sensor IBIS and the result is unsurprising.

I think the only way with current technology to combat this would be go with a slightly larger sensor and imaging circle and crop down in camera for the final image.

 

[pdk42]

I’ve seen this reported in other blogs and such as well. I think the issues stem from the amount of distortion correction the camera has to do at the edge of the frames. Couple that with the micro adjustments for the sensor IBIS and the result is unsurprising.

I think the only way with current technology to combat this would be go with a slightly larger sensor and imaging circle and crop down in camera for the final image.

What's interesting is that it seems to be mainly an issue with wide-angle lenses. I saw the same with the PL 8-18 on the E-M1ii. Anyone got any insight as to why that might be the case? I guess it's the IBIS "skew" that is causing it (forward/backward at the edges) rather than translational shifting.

 

[gryphon1911]

I believe it is the degree of distortion that is most difficult to correct. Wider angles have a tendency to need more correction.

 

[pdk42]

I believe it is the degree of distortion that is most difficult to correct. Wider angles have a tendency to need more correction.

Hi Andrew - but isn't distortion correction a static thing? (i.e. that it's about correcting barrel/pincushion distortion inherent in the lens's design). The IBIS problem is dynamic (i.e. it's about movement rather than corrections for optical design limitations).

EDIT - I guess you're thinking that the IBIS shift could move the sensor outside the optimal image circle. However, thinking about it, that would make only one edge blurred, but I'm finding that it's both edges. Here's the right side:

Subscribe to see EXIF info for this image (if available)

I can also see that there is a gradual deterioration as I go away from the centre, so I'm guessing it's some sort of skewing effect.

 

[gryphon1911]

I’m thinking that the lens being static with no IBIS, the in camera correction knows roughly what part of the lens should be there. However, with the sensor moving, it could be any number of millimeters offset and an infinite number of different combinations.

I’m just thinking that because of the inherent large distortion in a wider angle lens, that complicated the cameras job even more. Then couple that with lens optical variations in quality, thickness, no matter how minute. Makes me amazed that they can get the quality they do at all! Those engineers have one heck of a hard job.

 

[canonical]

In a traditional VR system (moving optical element) the element actually re-centers itself when the shutter button is pressed. You often times can see image shift in viewfinder after the mirror goes down again - especially when using long telephoto lenses.

With IBIS the sensor does not return to the center when the shutter button is pressed.