[UncleVanya]

A few learned folks on this forum have asserted in the past that the five axis stabilization used by various manufacturers only compensates by moving the sensor in a plane perpendicular to the light path - and that no tilting of the plane of the sensor occurs.

That made sense to me. The rationale being that any tilt would take part of the image out of focus - but... then I watched this video and it seems pretty clear to me that tilting is happening.

A close-up of Olympus’ 5-axis image stabilization system

Can someone help me understand this? I'm trying to figure out how the sensor can be tilted away from the normal plane and not negatively impact focus.

[stevebrot]

The short answer is that it depends on the system and whether or not stabilized lenses are part of the equation. How the various implementations work for all possible cases is deep magic of dark forces and not for open discussion by mere mortals. If it works, it works.


Steve

[jatrax]

"Any sufficiently advanced technology is indistinguishable from magic." - Arthur C. Clarke

Particularly if they don't want you to understand it...................

[Gimbal]

I watched the video and it clearly shows that the sensor isn't tilting.
If you watch the reflection in the sensor you can see the lens of the recording camera, this reflection doesn't move, it stays in place even though the sensor moves right and left, up/down and a does little rotation. Had the sensor been tilting the reflection would have changed A LOT.

[Steve.Ledger]

I watched the video and it clearly shows that the sensor isn't tilting.

There is tilt. (Pitch and yaw)

Here's a an infogram showing the axis..



And an Olympus video.

Youtu.be

Youtu.be

[UncleVanya]

There is tilt. (Pitch and yaw)

Here's a an infogram showing the axis..



And an Olympus video.

Assuming this is accurate - how does this avoid causing more problems than it solves? Tilting the camera doesn't change the lens / sensor orientation to one another and motion in that direction should change the focus across the plane of the sensor.

This is different than the Sony system or even the Panasonic system which can combine optical and in body IS together. The Oly systems are pure in body only.

---------- Post added 04-11-16 at 04:12 PM ----------

I watched the video and it clearly shows that the sensor isn't tilting.
If you watch the reflection in the sensor you can see the lens of the recording camera, this reflection doesn't move, it stays in place even though the sensor moves right and left, up/down and a does little rotation. Had the sensor been tilting the reflection would have changed A LOT.

That could be true. I'm not sure since I don't know quite how the microlenses on the front of the sensor behave - are they flat mirror reflectors or are they more curved? I'm not clear on this. I also see other lights that seem to come and go which implies a shift in the plane of the sensor picking up other reflections.

[Steve.Ledger]

Sony also has pitch and yaw. How else do you suppose 5-axis is achieved?
Also, Panasonic have only just introduced 5-axis with the yet to be released G80/85 which works with non-OS lenses (as well as with OS lenses)



Using is beliveing, I have an Olympus OMD and the IBIS is incredible (and quiet during video recording )

[UncleVanya]

Sony also has pitch and yaw. How else do you suppose 5-axis is achieved?
Also, Panasonic have only just introduced 5-axis with the yet to be released G80/85 which works with non-OS lenses (as well as with OS lenses)



Using is beliveing, I have an Olympus OMD and the IBIS is incredible (and quiet during video recording )

Honestly these are marketing slides and they describe the direction the camera can move and the sensor will compensate for - they don't clearly articulate HOW this is accomplished. And you are correct that 5 way is supposed to work with or without OIS - it's unclear if this is tilting the sensor or doing something else to compensate.

[Fogel70]

One important thing to know is that all image stabilization correct for pitch ad yaw. Which is done by either move the image sensor or a lens element in X- and Y-direction
That is the basics of all systems (2-axis or more).

3-axis systems add compensation and motion sensor for roll.
5-axis system also compensation and motion sensor for X- and Y-motion.

The only difference between pitch/yaw and X/Y motion is the type of motion sensor used, but the image sensor or lens element move in X/Y direction for both. The projected image is moving in X/Y direction for both of them. Tilting the sensor do not help for correction yaw/pitch unless the lens also tilt the same amount. But this will not work as a lens may be too heavy and would risk serious damage on the one holding the camera.

[Gimbal]

The sensor doesn't pitch or yaw, period.
When the camera pitch, the sensor moves up/down, when the camera yaw, the sensor moves left/right.

You all have Pentax cameras and can easily check this yourselves, unmount lens, turn on camera, enter focal length (make sure SR is on otherwise you will not be asked for focal length), set a manual exposure of 10 sec, press shutter button and watch the sensor while you pitch and yaw with the camera body. The sensor will shift left/right up/down, but never tilt.

[monochrome]

IIUC:

2-axis is Up/Down (1) & Left/Right (2);
3-axis adds roll, which is top corner Up/Down & opposing corner Down/Up (compensates for non-level horizon);
5-axis adds pointing camera slightly to one angle of subject plane (Yaw) (4) and tilting camera up or down from subject plane (Pitch) (5);

once SR has been invoked with a half-press.

K-1 has 5-axis IBIS.

[UncleVanya]

The sensor doesn't pitch or yaw, period.
When the camera pitch, the sensor moves up/down, when the camera yaw, the sensor moves left/right.

You all have Pentax cameras and can easily check this yourselves, unmount lens, turn on camera, enter focal length (make sure SR is on otherwise you will not be asked for focal length), set a manual exposure of 10 sec, press shutter button and watch the sensor while you pitch and yaw with the camera body. The sensor will shift left/right up/down, but never tilt.

Does Pentax claim 5 axis IS for any body other than the K-1?

---------- Post added 04-11-16 at 05:00 PM ----------

One important thing to know is that all image stabilization correct for pitch ad yaw. Which is done by either move the image sensor or a lens element in X- and Y-direction
That is the basics of all systems (2-axis or more).

3-axis systems add compensation and motion sensor for roll.
5-axis system also compensation and motion sensor for X- and Y-motion.

The only difference between pitch/yaw and X/Y motion is the type of motion sensor used, but the image sensor or lens element move in X/Y direction for both. The projected image is moving in X/Y direction for both of them, so tilting the sensor would not help for correction yaw/pitch. The only difference is that X/Y motion of the camera is usually not noticeable unless you focus really close.

I think this is accurate - but the dang marketing slides always make it seem otherwise. It makes no sense to work as described on the marketing slides and that video tricked me a little. I'm pretty sure that sensor isn't tilting now that I look again but...

[RobA_Oz]

The diagrams used to describe the 5-axis systems actually show the possible camera movements that contribute to image blur. It's easy to assume those movements are replicated on the image sensor, but for the makers to claim the IBIS works to compensate for those movements only requires it to sense them. Image sensor movements in the vertical and horizontal axes deal with X-Y plus pitch and yaw motions, and differential movements at either end of the image sensor account for roll.

[stevebrot]

Does Pentax claim 5 axis IS for any body other than the K-1?

If they have, it did not make it into the product descriptions.


Steve

[stevebrot]

...for the makers to claim the IBIS works to compensate for those movements only requires it to sense them. Image sensor movements in the vertical and horizontal axes deal with X-Y plus pitch and yaw motions, and differential movements at either end of the image sensor account for roll.

Yep, though the devil is in the detail of doing the fine control for the correction. Pitch and yaw are harder to correct for.


Steve