Built in shake reduction. Does this really works?

Regarding SR on a tripod, without SR we can assume the sensor is stationary. But in fact the sensor is always being held in position by some magnets under the control of the movement detection. If the movement detection is subject to any input then it will move the sensor, and likewise if it generates any spurious output.

Is there any possibility of the movement detection producing a signal when it is on the tripod? Probably low, but on a longer exposure it might drift slightly with changing temperature or even react to loud noise - we don't know. For example live view gets the sensor quite warm, and then it cools, so that could account (in theory) for changing temps in the body which could interfere with the proceedings. Any noise signal from the movement detection will be worse than no signal.

Switching SR off probably only switches off the movement detection, since the sensor is still magnetically suspended.

Another scenario - you have SR on and take a tripod-mounted shot at a speed like 1/4 sec. Pressing the shutter will move the body very very slightly, the SR will wake from its slumber and start to compensate, but the shutter opens before it has settled. It was settled before in the static case but that very slight wobble on pressing the button might cause it to overshoot for a fraction of a second as it reacts to the new tiny movement.

Like Smeggy I believe some things more when the reasons to believe them are more evident.

When SR is off on K20D and later bodies the sensor is clamped into position, not free moving.

On a tripod, I would suggest this means the highest possible accuracy and therefore a good idea where work of the highest precision is required.

To respond to Smeggy - why switch SR off on a tripod?

The SR is optimised to the frequency of normal hand shaking - a quite low frequency. When rigidly mounted on a tripod, the natural frequency of movement is much higher (the hand normally damps out the high frequency). Control theory indicates that there must be one dominant time factor for stability. The movement due to the hand is slow, the sensor can react quicker than the hand shakes. When on the tripod, the vibration is much faster than hand shaking but smaller. This frequency, if close to the response of the sensor movement, could result in undamped movement which might be worse than having the SR switched off.

I hope this helps - not easy explaining it in simple terms...

Regards
Tim

Thanks Tim, that sheds more light on it.

Useful info and diagram in this link.

Regarding clamping in position I think it is always free-floating but with SR off it is held in one position by the electromagnets rather than being physically clamped.

It was free floating on the K10D but it was changed for the K20D.

That's one big advantage with "In-Body" SR. If they come up with an improved version, it's available in the latest camera and works with all your lenses.

With "In-Lens" SR (aka "Legacy SR" from the film days ) when they come up with an improvement as Nikon did (mentioned in a link recently) you have to change ALL your lenses and SR won't even be available for some.

Ahhh that, in bold is starting to make some sense. Thanks

Thanks the Pentax pdf was very helpful although I had to laugh when it said 24 million lenses were compatible with it - Think the Marketing Person entered 'feedback loop' of stupidty there.

I'm pretty sure the sensor is not physically clamped when off on the 'K-5' - If you test in live view where you can see the sensor movement, you can still see some overshoot, when SR is 'off' by quickly pannign the camera and stopping it abruptly. Although not as much as when SR is on.

More about it than you ever wanted to know in this Pentax patent: http://www.patents.com/us-20080226276.html

If you wade your way through to item 196 it does say what happens when SR is off - the calculations are still performed in order to hold the sensor electromagnetically in its mid-position.

The sensor movement is also discussed for the K-7 in regard to its possible interaction with blur induced by shutter acceleration in this article - complete with the pictures that are omitted in the patent application above. See section 5.5 in particular.

Japanese patent with diagrams here.

Phew! I think I need a cup of tea.

Had a look at this and my brain hurts.

Thank god there's a layman's summary in there. Although worryingly it does suggest K7 owners need a 'massive' tripod to prevent shake. Remind you of anything???

You only need a tripod like that at around 1/80th

The SR will stay one when capturing the photo, if that wasn't the case why would you need SR to begin with?

This from Canon, no details though.

Excellent stuff. I'll have a more detailed look at that later

yes of course. Dunno What I was thinking when I typed that


Senile

I also have a practical problem with this and also agree with the poster who simplistically demonstrated that the software was flawed. My problem is that camera shake will be a diminishing variable from very shaky hands through fairly steady, to a monopod, to a cheap tripod to an expensive tripod! It's clever how it can distinguish between steady hands and a person with shaky hands using a monopod! There must be a crossover period that computes shutter speed/focal length/measured shake in the decision whether to turn it off or not.(Or can we just say that monopods and tripods always produce a different oscillating frequency so always turn off?) I always forget on my cheap tripod and it seems ok!
Dave