[Gimbal]

I came up with (what I thought) an interesting test regarding SR and panning.


This is what I did.
I attached the camera on a folded tripod and turned the head so that the folded legs came up in the middle of the viewfinder (I used a da21mm lens). Then I lifted the tripod and turned on live view and wiggled the camera around. With SR off, the legs was locked in the middle of the view no matter how I twisted and turned, as expected. With SR on one can see the SR working trying to smooth out the shakes and the legs are no longer locked in the middle but slide around in the center of the view. There is no problem seeing how much the legs/sensor can move before hitting the limits.

Now if I start panning the legs are still held in the middle, not hitting the side-way limits or even moving near them, but they are free do move up and down. With other words, a classical panning mode.

If I start panning with a sudden acceleration, the legs/sensor is immediately locked sideways, if I start softly the sensor first hit a side limit and then moves in and locks in the middle. Seems to work great.

Looks to me as there is a well working panning mode.


Now I try taking photos while panning, using about 1/20s. I don't have any actual object to track so I just start swinging and firing.

First with SR off.
The tripod legs are sharp and the surrounding is a blur, as expected.

Then with SR on.
I wanted to see fairly sharp legs, if any motion blur it should only be vertical. But instead they where kind of blurry, especially sideways. This suggests that the panning mode doesn't work well at all, in fact, quite the opposite.

Disappointing, panning doesn't seems to be working in photo mode.


Now, a last test. Panning, SR on, photo from live view.
And hey, the legs seems to only be blurred vertically. That means that the SR is only working vertically, panning mode is effective.


So I would call this a bug.
We clearly have an effective panning mode, but it is only used if shooting from live view. But who does that? It's just to slow in reality.

Someone call PENTAX.

[Adam]

This is why DSLRs shouldn't have live view

[bobmaxja]

IT,s not a bug. You should not use the Sr when panning

Read p139 OPERATING MANUAL

The shake reduction fwill not fully work when shooting with a very slow speed shuter, for example when panning or shooting night scenes. In this case, it is recommended to turn off Shake Reduction function and use the camera with a tripod

[creampuff]

How is it a bug? Camera manual already explains it.
Hmm, I don't think most of us shoot with a tripod leg in the middle of the viewfinder.

[Gimbal]

So if one take the quoted manual text literally, one should only use SR with short shutter speeds in bright scenes. But who needs SR in those situations?

It's all about what they mean with "very slow shutter speed". IMO below 1/20 (or there about depending on lens) is the lower limit for SR, below that is tripod territory.

[cbope]

So if one take the quoted manual text literally, one should only use SR with short shutter speeds in bright scenes. But who needs SR in those situations?

It's all about what they mean with "very slow shutter speed". IMO below 1/20 (or there about depending on lens) is the lower limit for SR, below that is tripod territory.

I recently shot in both a museum and an aquarium with my K-5 and found that with SR on, I could get acceptable shots down to 1/20s about 50% of the time. This was using the kit 18-55 WR lens, mostly at the short end of the zoom range (e.g. wide angle). It's not perfect 100% but I believe the SR system works pretty well. ISO was mostly at 3200 (auto) and I got several really nice shots. Probably not tack sharp blown up to 20x30, but acceptable given the poor lighting and lack of flash in these shots.

[Lowell Goudge]

no offence to the OP but I think the entire test is flawed and that the only real test is to go out and shoot, your time would be better spent.

Shake reduction is not flawed, but to my knowledge there is no valid test for using points of reference physically attached to the camera as an indication that shake reduction is working or not.

As for how effective SR is, even going back to the K100D when it came out, I tried it in a stoor with a 50-200mm zoom or something like that, and could easily hand hold at 1/50th when zoomed to 200mm. that was good enough for me to say it works.

Since I got my K10D, (my first body with SR) I have shot A LOT at speeds well below 1/ (FL * 1.5) the rule of thumb for "acceptable sharpness" and not had issues, with the exception of panning, because SR can and does introduce some motion blurr unless you are panning exactly at right angles to the direction of motion, SR will see any acceleration, even that from panning a subject moving at an angle, as shake, and try to compensate.

In short, SR is great at taking camera induced blurr out of shots, but is reliant on a stationary subject.

[Gimbal]

Shake reduction is not flawed, but to my knowledge there is no valid test for using points of reference physically attached to the camera as an indication that shake reduction is working or not.

Right, it doesn't show if SR is working correctly, but it does show if it is enabled, and if it is enabled in both horizontal and vertical direction. Normal for a panning mode is that the SR is only active in the vertical direction.

So to sum it up, the K5 supports panning with SR enabled, but only from live view. And I'm the only one thinking that is weird. Have to admit I'm a bit surprised.

[Aegon]

So if one take the quoted manual text literally, one should only use SR with short shutter speeds in bright scenes. But who needs SR in those situations?

It's all about what they mean with "very slow shutter speed". IMO below 1/20 (or there about depending on lens) is the lower limit for SR, below that is tripod territory.

I think the logical negation of "very slow" is "not very slow", which is different than "short and bright".

Put another way, use SR when your shutter speeds are not very slow.

That is a lot more reasonable than "one should only use SR with short shutter speeds in bright scenes".

[TaoMaas]

I came to the conclusion some time back that the SR added blur in certain situations (and I quite liked the effect), but I'd never been able to duplicate it consistantly. Based on what Gimbal found in his tests, I kinda think that what I was seeing were times when I caused the SR to overcompensate or max out. This gives me something to go on now...so, thanks!

[Gimbal]

I also tried my K10d, in normal mode only as it doesn't have live view.

It seems as the k10 also detects a pan and then only allow the SR to operate in vertical mode.



Notice the blur in the background, I'm panning sideways, but the blur on the tripodlegs are mostly vertical. Thus the SR is now only working in a vertical direction, as expected if there is a panning mode.


Here is the same thing shot with the K5, from live view.



Same story, panning sideways, blur on legs only vertical. Thus SR is locked sideways.


And finally, the K5 in normal mode.



The legs are also blurred sideways, really bad one might add. SR is not locked sideways even though I was panning.

I call it a bug. Unless someone can come up with a better explanation.

Anyhow, with the K5, turn off SR during a pan.


(for some reason exif is partially stripped by PS, but all pictures where marked as "stabilized")