ibis (A7RII) and shifted lenses?

[daf]

Can Ibis work properly with shifted lenses?

 

[k-hawinkler]

Great question! Thanks.

 

[hiepphotog]

I don't see any reason that it wouldn't work.

 

[Vivek]

Great question! Thanks.

I will be more interested in any practical applications of this. It is something totally new.

 

[sbay]

I will be more interested in any practical applications of this. It is something totally new.

I've often used the 24 TSE handheld with shift for architecture / street scenes. Usually my shutter speeds were fine without IS but I suppose it wouldn't hurt (especially with 40-50MP cameras).

 

[mjm6]

I suspect that it won't work properly. That doesn't mean that it won't actually improve the image somewhat under normal circumstances, but it isn't going to be able to make the same level of improvements that are possible with an axially centered optical system.

Since the camera has 5-axis IBIS, it is going to be attempting to deal with sensor shifts (x,y) and also rotation (yaw, pitch, roll). Here is a link to an image to show this:

https://www.google.com/search?q=sony+5+axis+image+stabilization&rlz=1C1ZMDB_enUS561US561&espv=2&biw=1852&bih=874&source=lnms&tbm=isch&sa=X&ved=0CAgQ_AUoA2oVChMIka7dkOSUxwIVDBU-Ch35xwCt#imgrc=GcqSpnX9fqaBNM%3A

Anyway, the x and y IS should be OK, but the yaw, pitch, and roll will not, because they are designed around a lens that is on the axial center of the chip. Shifts would require a different formula for y,p,r corrections to be reasonable, and I suspect that even if it were possible to make the calculations, the IBIS doesn't have the range of movements necessary to fully compensate (the further off center you go, the greater the range of movements necessary in the sensor to compensate).

So, as I said, it probably won't work correctly when the lens is shifted, but it may work sufficiently well for normal shooting. Should be OK when not shifted.


---Michael

 

[k-hawinkler]

Well, naively I would assume one would use a tripod with such a lens and turn IBIS off.
Then IBIS would just hold the sensor in the default position, no?
Should work just fine. Where am I wrong?

 

[hiepphotog]

I suspect that it won't work properly. That doesn't mean that it won't actually improve the image somewhat under normal circumstances, but it isn't going to be able to make the same level of improvements that are possible with an axially centered optical system.

Since the camera has 5-axis IBIS, it is going to be attempting to deal with sensor shifts (x,y) and also rotation (yaw, pitch, roll). Here is a link to an image to show this:

https://www.google.com/search?q=sony+5+axis+image+stabilization&rlz=1C1ZMDB_enUS561US561&espv=2&biw=1852&bih=874&source=lnms&tbm=isch&sa=X&ved=0CAgQ_AUoA2oVChMIka7dkOSUxwIVDBU-Ch35xwCt#imgrc=GcqSpnX9fqaBNM%3A

Anyway, the x and y IS should be OK, but the yaw, pitch, and roll will not, because they are designed around a lens that is on the axial center of the chip. Shifts would require a different formula for y,p,r corrections to be reasonable, and I suspect that even if it were possible to make the calculations, the IBIS doesn't have the range of movements necessary to fully compensate (the further off center you go, the greater the range of movements necessary in the sensor to compensate).

So, as I said, it probably won't work correctly when the lens is shifted, but it may work sufficiently well for normal shooting. Should be OK when not shifted.


---Michael

Michael, the way you describe it would mean the sensor somehow is able to use the liveview feed to compensate for the vibration. I believe these IBIS (or OIS) would just have a pre-determined compensation depending on the input focal length and the current vibration input. Shifting would not change the focal length, I assume. The shift would try to match the induced vibration so that the light ray would hit exactly the same spot during the exposure time. So the T/S 24 would have the same compensation as a normal 24 given the same amount of vibration.

And as it turns out, 3rd party lens would not have yaw and pitch compensation, especially the T/S lens.

 

[mjm6]

Michael, the way you describe it would mean the sensor somehow is able to use the liveview feed to compensate for the vibration. I believe these IBIS (or OIS) would just have a pre-determined compensation depending on the input focal length and the current vibration input. Shifting would not change the focal length, I assume. The shift would try to match the induced vibration so that the light ray would hit exactly the same spot during the exposure time. So the T/S 24 would have the same compensation as a normal 24 given the same amount of vibration.

And as it turns out, 3rd party lens would not have yaw and pitch compensation, especially the T/S lens.

I believe the camera has accelerometers in it to detect motion. Yaw, pitch, and roll are "twist" movements, where one side of the chip is moving in one direction and the other side in the opposite. My point was that this can't be properly handled if the image is shifted.

If the camera drops down to simple motion adjustments when the camera doesn't detect an FE lens, then it'll probably do just fine, just not as well as the full 5-axis IBIS would be able to do.

---Michael

 

[dmward]

Certainly not a definitive test, but I just tried OIS on and off with a 24mm TSE on my A7II.
I had the IBIS set to manual and the focal length dialed in. Although I've read that if the lens data is transmitted to the camera auto IBIS will work as well.

At 1/8 second handheld the images look reasonably sharp with in both instances. It did seem that the EVF was a bit steadier at high magnification focus assist. That suggests that IBIS does work, at least to some extent.

As mentioned, most of the time when using the shift feature I have the camera on a tripod. When doing handheld with shift its high shutter speeds where IBIS is less important.

 

[hiepphotog]

I believe the camera has accelerometers in it to detect motion....

---Michael

Exactly, that's why I'm confused as to how shifting, an optical mechanism, would interfere with the IBIS, purely based on mechanical vibration.

 

[ggibson]

By the way, anyone using adapted lenses on the A7rII should be aware that there is a bug with IBIS that the A7II also has which is confirmed present on the A7rII also. When turning OFF the camera to change lenses:

- If you attach a 3rd party lens after using an OSS (eg the 24-70 FE) lens then IBIS does NOT work.
- If you attach a 3rd party lens after using a non OSS lens (eg the 55 f1.8) then IBIS does work.

The workaround is leaving the camera on while you switch lenses, which is a bit weird to me, but sometimes I do it by mistake anyways

BTW, I assume this is only with lenses/adapters that are NOT electronically coupled. Presumably, the Loxias or a Canon lens on an AF adapter would engage IBIS properly after an OSS lens. I don't have an A7rII to test though.

 

[mjm6]

Exactly, that's why I'm confused as to how shifting, an optical mechanism, would interfere with the IBIS, purely based on mechanical vibration.

Because the camera is detecting motion, running the motion through a formula that has the focal length in it to determine the offsetting displacement of the sensor, and then firing a micro actuator to offset the movement to keep the sensor in the same place relative to the projected image based on that information.

When the movement is rotational, the displacement necessary to offset the image movement is a function of both the focal length and the distance from the optical center of the image, and since that is essentially unknown in the case of a TS lens, the camera can't properly calculate the displacement necessary.

However, if the camera drops out the rotational aspects and just uses linear displacement, it will do OK for the most part, and I think that is what it will do with the TS lenses, as you have said.

---Michael