[D1N0]

Actually movement in all 5 "axels" are compensated by moving and rotating in the focal plane. So the senor itself is always in the same plane, only sliding left/right, up/down and rotating a few degrees.

That's three axis. Horizontal vertical and rotational. Five means wiggling left and right and up and down. If the sensor does not do that then five axis is marketing BS.

[monochrome]

My Pentax's K5IIs, K3, K1 are hunting more than my other brand cameras. That's it!

Have you used the exact same lenses, say a good Tamron, (with different mounts, of course) on all the brands?

[photoptimist]

That's three axis. Horizontal vertical and rotational. Five means wiggling left and right and up and down. If the sensor does not do that then five axis is marketing BS.

It's not BS. In fact the primary correction is the use of 2 linear axes of up-down/left-right motion of the sensor to correct for two rotational axis disturbances (pitch and yaw).

It's only roll axis motion of the camera that provokes an identical-but-opposite roll axis correction by the sensor.

What changes in going from the original 2 or 3 axis SR to full 5 axis is that an additional two axes of linear motion of the camera are corrected with additional amounts of linear motion of sensor which depends on optical magnification ratio (which depends on both focal length and distance).


Tilting the sensor, although it sounds cool, is almost pointless because it would do nothing to correct motion blur except under the most extreme motions.

[D1N0]

It does not make any sense to me

[leekil]

It's not BS. In fact the primary correction is the use of 2 linear axes of up-down/left-right motion of the sensor to correct for two rotational axis disturbances (pitch and yaw).

How does the up/dow/left/right movement compensate for pitch and yaw? It seems like only moving within the sensor plane would not be sufficient for that, in that there would be some tilting of the plane of focus relative to the sensor. Or is the idea that it is minimal enough that in most cases it wouldn't affect the focus and focal plane? Now that I think of it, I have occasionally seen what looks like weird plane-of-focus changes from one frame to another when I have taken two in quick succession.

It's only roll axis motion of the camera that provokes an identical-but-opposite roll axis correction by the sensor.

What changes in going from the original 2 or 3 axis SR to full 5 axis is that an additional two axes of linear motion of the camera are corrected with additional amounts of linear motion of sensor which depends on optical magnification ratio (which depends on both focal length and distance).

Tilting the sensor, although it sounds cool, is almost pointless because it would do nothing to correct motion blur except under the most extreme motions.

In the older Pentax SR diagrams, they show only the linear motion of the sensor, like for the K-70:



With the newer 5-axis SR II, they use a different style of diagram showing the pitch, yaw, and roll, like the K-1:



There's a slight amount of explanation here comparing Sony vs. Olympus, and there is a better version of the K-1 style diagram that makes the pitch & yaw effects more obvious:



A Comparison of How Olympus and Sony's 5 Axis Stabilization Work

Is Pentax's shake reduction completely independent of Sony & Olympus? Or do they share technology?

[Gimbal]

It does not make any sense to me

Look through the viewfinder and do some very small pitch and yaw movements, the scene will appear to move up/down and left/right.

The SR compensates for this by moving the sensor up/down and left/right, and have done so since the K-10d days.

The two new axis are for when the camera is translated left/right and up/down, the scene appears to move accordingly thus the SR compensates for this by moving the sensor left/right and up/down.

[Kunzite]

Yes, the sensor is moving within the focus plane and yes, it does make sense to correct pitch and yaw with such movements. In any case, tilting the sensor cannot actually correct for pitch and yaw - it should be obvious if you do the experiment proposed by Gimbal.

leekil, what we're seeing in the newer diagrams is camera movements, i.e. what the SR is compensating for. Not sensor movements i.e. the actual method of compensating.

[D1N0]

Look through the viewfinder and do some very small pitch and yaw movements, the scene will appear to move up/down and left/right.

The SR compensates for this by moving the sensor up/down and left/right, and have done so since the K-10d days.

The two new axis are for when the camera is translated left/right and up/down, the scene appears to move accordingly thus the SR compensates for this by moving the sensor left/right and up/down.

No sr effects can be seen through the viewfinder. Any pitching and yawing by the user will also result in other movements explaining any motion in live view of course it moves, it has to catch up with very crude shake. Even if it attempts to correct pitch en yaw using just three axis (5 axis is only virtual) then still the focus plane in the image will shift. It is bs in my book. But it also means the sensor is just always in the focus plane on the sensor because it doesn't make any movements to the front and back.

[Kunzite]

What you see through the viewfinder is the movement SR has to compensate for.
If the image appears to be shifting (and it does), the SR must shift the sensor in the opposite direction. There is no workaround to that.
The focus will shift with an insignificant amount, don't forget the motion range compensated for is way, way smaller than that at focus&recompose.

You shouldn't be so hasty to use strong words ("bs") without understanding how it works.

[D1N0]

maybe you shouldn't tell me what to to with a condescending attitude.

[Kunzite]

D1N0, you're calling our explanations "bs". Allow me to be slightly annoyed by that.

[D1N0]

I am calling the use of the term 5 axis stabilisation BS. get somebody to look at your toes. They are obviously unpractically long!

[Kunzite]

While the term "5 axis" is generated by marketing, there are indeed 5 types of motions being corrected for. Translation is different than pitch and yaw, in that it's not an angular movement - and to my understanding, the previous SR systems were using gyro sensors capable of only detecting angular movements. Also, translation becomes an issue at close distances, once again behaving differently.

[zapp]

SR - you know...

Is that the one thing slowing Pentax autofocus.

I just think - you elaborate!

Is in camera shake reduction a dead end?

Awaits fire...

The basic suggestion in the beginning is that Penattax has sensor stabilization (a) and slow AF (b). The hypothesis is that a causes b. The premature conclusion thrown at us is that Pentax AF is a dead end.
Now we are asked to work out the theoretical background.

The answer is No. Hypotheis wrong. Conclusion wrong. That does not mean that Pentax AF is not slow. And yes there ar questionable questions.

[Gimbal]

I am calling the use of the term 5 axis stabilisation BS. get somebody to look at your toes. They are obviously unpractically long!

You obviously don't understand it, so maybe you should cool down until you do.