Olympus 4th and 5th IBIS motions

[joeletx]

A while ago, there was a discussion about the Olympus 5-axis IBIS is actually only three because the roll and turn 4th and 5th axis move along the X and Y instead. Well, according to this video, that is not true! It does indeed turns and tilts around X, Y, and Z axis. It is amazing, just take a look.

 

[Rocky IDOM]

Cool vid! Thanks for sharing.

 

[bigley Ling]

excellent vid. What is of side interest is how the rear LCD is turned on itself. I like this idea, as it gives the impression the camera not not "so digital"

 

[tt321]

How do you perceive it turning and tilting?

It's shot with a fisheye lens, so pure shifting can appear to involve some tilting on casual observation. I cannot conclusively believe that tilting was involved.

 

[Anders W]

A while ago, there was a discussion about the Olympus 5-axis IBIS is actually only three because the roll and turn 4th and 5th axis move along the X and Y instead. Well, according to this video, that is not true! It does indeed turns and tilts around X, Y, and Z axis. It is amazing, just take a look.

Sigh! No the sensor doesn't tilt and the video doesn't show it to do so. It would also be completely pointless from a stabilization point of view to have the sensor move that way.

The five axes refer to five different types of camera shake, not five different types of sensor movement. The five types of camera movement for which the system corrects are known as pitch, yaw, roll, vertical shift, and horizontal shift. In order to correct for these five types of camera movement, the sensor moves in three ways, not five: up-down (to correct for pitch and vertical shift), left-right (to correct for yaw and horizontal shift), and rotation about the optical axis (to correct for roll).

Pitch and yaw require different motion sensors than vertical and horizontal shift and the correction of the latter is also based on different information than the former in other regards (correction of vertical and horizontal shift requires information on focus distance whereas correction of pitch and yaw doesn't). But pitch and vertical shift is corrected by the same sensor movement (up-down) and the same is true about yaw and horizontal shift (left-right).

For further detail, see this post of mine from three years ago when the five-axis IBIS system was new:

Loading…

www.dpreview.com

 

[Henry Richardson]

Sigh! No the sensor doesn't tilt and the video doesn't show it to do so. It would also be completely pointless from a stabilization point of view to have the sensor move that way.

The five axes refer to five different types of camera shake, not five different types of sensor movement. The five types of camera movement for which the system corrects are known as pitch, yaw, roll, vertical shift, and horizontal shift. In order to correct for these five types of camera movement, the sensor moves in three ways, not five: up-down (to correct for pitch and vertical shift), left-right (to correct for yaw and horizontal shift), and rotation about the optical axis (to correct for roll).

Pitch and yaw require different motion sensors than vertical and horizontal shift and the correction of the latter is also based on different information than the former in other regards (correction of vertical and horizontal shift requires information on focus distance whereas correction of pitch and yaw doesn't). But pitch and vertical shift is corrected by the same sensor movement (up-down) and the same is true about yaw and horizontal shift (left-right).

For further detail, see this post of mine from three years ago when the five-axis IBIS system was new:

http://www.dpreview.com/forums/post/40558437

Thanks, that was the best explanation I have seen. Which of these types of camera shake does the E-M10 not correct for?

 

[joeletx]

If the X and Y movements are used to compensate for rotattions, how does the camera distingush the rotational shifts from plain pan and shift X Y movements?

 

[Anders W]

If the X and Y movements are used to compensate for rotattions, how does the camera distingush the rotational shifts from plain pan and shift X Y movements?

Rotation of the sensor about the optical axis (sensor movement) is used to compensate for roll (camera movement). This type of camera movement is distinguished from other types of movement (pitch, yaw, vertical shift, horizonal shift) by the camera movement sensors (which can distinguish between movements of five different kinds).

 

[Anders W]

Sigh! No the sensor doesn't tilt and the video doesn't show it to do so. It would also be completely pointless from a stabilization point of view to have the sensor move that way.

The five axes refer to five different types of camera shake, not five different types of sensor movement. The five types of camera movement for which the system corrects are known as pitch, yaw, roll, vertical shift, and horizontal shift. In order to correct for these five types of camera movement, the sensor moves in three ways, not five: up-down (to correct for pitch and vertical shift), left-right (to correct for yaw and horizontal shift), and rotation about the optical axis (to correct for roll).

Pitch and yaw require different motion sensors than vertical and horizontal shift and the correction of the latter is also based on different information than the former in other regards (correction of vertical and horizontal shift requires information on focus distance whereas correction of pitch and yaw doesn't). But pitch and vertical shift is corrected by the same sensor movement (up-down) and the same is true about yaw and horizontal shift (left-right).

For further detail, see this post of mine from three years ago when the five-axis IBIS system was new:

http://www.dpreview.com/forums/post/40558437

Thanks, that was the best explanation I have seen.

You are welcome.

Which of these types of camera shake does the E-M10 not correct for?

Vertical shift and horizontal shift, which don't matter much except at high magnifications. For most images, three axes will do about as well as five.

 

[Skeeterbytes]

Can anybody help resolve apparent contradiction between the contention that "5-axis" stabilization only corrects three ways with this Olympus video, which shows rotation on the vertical and horizontal axes, along with the other three?

Much appreciated!

Rick

--
Equivalence and diffraction-free since 2009.
You can be too; ask about our 12-step program.

 

[Anders W]

Can anybody help resolve apparent contradiction between the contention that "5-axis" stabilization only corrects three ways with this Olympus video, which shows rotation on the vertical and horizontal axes, along with the other three?

The video shows only three types of sensor movement: up-down, left-right, and rotation about the optical axis. For reasons already spelled out (see here), these three types of sensor movement are the only ones required to correct five different types of camera movement: pitch, yaw, roll, vertical shift, and horizontal shift. Consequently, there is no contradiction between the contention you cite and what the video shows as long as "5-axis" refers to camera movement and "corrects three ways" refers to sensor movement.

 

[Skeeterbytes]

Can anybody help resolve apparent contradiction between the contention that "5-axis" stabilization only corrects three ways with this Olympus video, which shows rotation on the vertical and horizontal axes, along with the other three?

The video shows only three types of sensor movement: up-down, left-right, and rotation about the optical axis. For reasons already spelled out (see here), these three types of sensor movement are the only ones required to correct five different types of camera movement: pitch, yaw, roll, vertical shift, and horizontal shift. Consequently, there is no contradiction between the contention you cite and what the video shows as long as "5-axis" refers to camera movement and "corrects three ways" refers to sensor movement.

0:15-0:19 shows rotations on the vertical and horizontal axes. This animation from 0:11-0:14 shows the movement more distinctly yet.

So, a contradiction seems to remain.

Cheers,

Rick

 

[Anders W]

Can anybody help resolve apparent contradiction between the contention that "5-axis" stabilization only corrects three ways with this Olympus video, which shows rotation on the vertical and horizontal axes, along with the other three?

The video shows only three types of sensor movement: up-down, left-right, and rotation about the optical axis. For reasons already spelled out (see here), these three types of sensor movement are the only ones required to correct five different types of camera movement: pitch, yaw, roll, vertical shift, and horizontal shift. Consequently, there is no contradiction between the contention you cite and what the video shows as long as "5-axis" refers to camera movement and "corrects three ways" refers to sensor movement.

0:15-0:19 shows rotations on the vertical and horizontal axes. This animation from 0:11-0:14 shows the movement more distinctly yet.

So, a contradiction seems to remain.

The animation in that video is simply wrong with regard to the two tilting sensor movements shown from 0:11 to 0:14. Those two movements do not take place. If they did, it would be of no help with regard to stabilization. It would just shift the focus plane around, just like a tilt lens.

The remaining three sensor movements, shown from 0:15 forward are those that actually take place.

 

[tt321]

Can anybody help resolve apparent contradiction between the contention that "5-axis" stabilization only corrects three ways with this Olympus video, which shows rotation on the vertical and horizontal axes, along with the other three?

The video shows only three types of sensor movement: up-down, left-right, and rotation about the optical axis. For reasons already spelled out (see here), these three types of sensor movement are the only ones required to correct five different types of camera movement: pitch, yaw, roll, vertical shift, and horizontal shift. Consequently, there is no contradiction between the contention you cite and what the video shows as long as "5-axis" refers to camera movement and "corrects three ways" refers to sensor movement.

0:15-0:19 shows rotations on the vertical and horizontal axes.

These are rotations in the shake, not in the anti-shake, so to speak.

This animation from 0:11-0:14 shows the movement more distinctly yet.

So, a contradiction seems to remain.

Cheers,

Rick

 

[Skeeterbytes]

0:15-0:19 shows rotations on the vertical and horizontal axes. This animation from 0:11-0:14 shows the movement more distinctly yet.

So, a contradiction seems to remain.

The animation in that video is simply wrong with regard to the two tilting sensor movements shown from 0:11 to 0:14. Those two movements do not take place. If they did, it would be of no help with regard to stabilization. It would just shift the focus plane around, just like a tilt lens.

The remaining three sensor movements, shown from 0:15 forward are those that actually take place.

Hmm, well, it seems improbable that Oly would present their own technology incorrectly, but if they have I'm sure there's some documentation to settle the question.

One implication of such motion is eventually having in-camera swing and tilt, which would be yet another productivity extension of the IBIS module.

Cheers,

Rick

 

[tt321]

0:15-0:19 shows rotations on the vertical and horizontal axes. This animation from 0:11-0:14 shows the movement more distinctly yet.

So, a contradiction seems to remain.

The animation in that video is simply wrong with regard to the two tilting sensor movements shown from 0:11 to 0:14. Those two movements do not take place. If they did, it would be of no help with regard to stabilization. It would just shift the focus plane around, just like a tilt lens.

The remaining three sensor movements, shown from 0:15 forward are those that actually take place.

Hmm, well, it seems improbable that Oly would present their own technology incorrectly, but if they have I'm sure there's some documentation to settle the question.

In that animation, even the x and y translational shift compensations are illustrated incorrectly. As illustrated it only works under one very restrictive condition. In most cases the correct compensation movements are not like that.

The only 100% correctly depicted compensation is the rotation around the optical axis.

One implication of such motion is eventually having in-camera swing and tilt, which would be yet another productivity extension of the IBIS module.

In camera swing and tilt will do what swing and tilt do, which are not related to motion compensation but focus plane adjustments.

The existing shift facility is not now used for mimicking shift lenses. Has anyone done that?

 

[clengman]

0:15-0:19 shows rotations on the vertical and horizontal axes. This animation from 0:11-0:14 shows the movement more distinctly yet.

So, a contradiction seems to remain.

The animation in that video is simply wrong with regard to the two tilting sensor movements shown from 0:11 to 0:14. Those two movements do not take place. If they did, it would be of no help with regard to stabilization. It would just shift the focus plane around, just like a tilt lens.

The remaining three sensor movements, shown from 0:15 forward are those that actually take place.

Hmm, well, it seems improbable that Oly would present their own technology incorrectly, but if they have I'm sure there's some documentation to settle the question.

Nah, This is not a technical illustration. It's just a cartoon for marketing purposes. I kind of doubt that the illustrators 1) Had anything more than very cursory input from engineers that actually know how the system works and 2) Would have an easier time grasping the concept than your average dpreview reader.

Given that, even if the illustrator had a very good understanding of the technology and knew that the sensor never tilted either about x or y axes, I doubt that it would be a benefit to try to portray this completely accurately. It's much easier to show the camera moving by 5 different types of movement and the sensor moving in 5 different ways to compensate. Even if it's not accurate, it gets the point across with no further explanation to the average consumer.

If you try to make the cartoon completely accurate, the average consumer isn't going to immediately understand why camera pitch can't be corrected by sensor pitch. Making it clear how this works then requires more than just an animated cartoon. You have to include text and technical writing. That's not good marketing.

It's clearly an example of technical information being dumbed down for the masses. It's unfortunate that the result for some curious users is a misunderstanding of the technology.

One implication of such motion is eventually having in-camera swing and tilt, which would be yet another productivity extension of the IBIS module.

Cheers,

Rick

--
Equivalence and diffraction-free since 2009.
You can be too; ask about our 12-step program.

 

[OrdinarilyInordinate]

Cool! I'm going to visualize that every time I use my E-M1 now...

 

[Skeeterbytes]

In camera swing and tilt will do what swing and tilt do, which are not related to motion compensation but focus plane adjustments.

The existing shift facility is not now used for mimicking shift lenses. Has anyone done that?

Not to my knowledge, not yet anyway. Sensor shift fore and aft for focusing would be implicitly possible, and for that there is a precedent.

Cheers,

Rick

 

[Skeeterbytes]

Presumably then there's a technical paper somewhere, a patent filing perhaps, detailing how it actually works? Can you point me to it?

Cheers,

Rick