[LeDave]

I was searching Youtube videos and I stumbled across this one. I was impressed by the SR of Pentax DSLR's, I didn't expect it to be this good!

Pentax K-7 SR at 500mm
[YT]http://www.youtube.com/watch?v=yWNtOoLZjW4[/YT]

[netrex]

Impressive performance I'll have to do this with my white K-x when I get it. Hopefully, regardless of what DPReview said, it should be the same.

[jogiba]

Just don't try to manual focus a handheld 500mm lens while recording.
[yt]http://www.youtube.com/watch?v=m9L-y42Nj58[/yt]
[yt]http://www.youtube.com/watch?v=5zCiuh_UWLc[/yt]

With Bogen 475/501 tripod /head combo:
[yt]http://www.youtube.com/watch?v=kgRROQYmjMM[/yt]
[yt]http://www.youtube.com/watch?v=oKO5f5RFatA[/yt]

[falconeye]

I didn't expect it to be this good!

4 stops are a factor 16.
Even less than 4 stops would make a huge difference in shake when viewed directly.

I keep hearing that in-lens shake reduction (using a stabilized lens with Canikon) cannot deliver this level of stabilisation in movie mode.

I guess this is due to Pentax' over-sized SR-margin in their implementation which is more than would have been required to cover 16x the circle of confusion.

[Steve.Ledger]

I hand held the K-7 for this pan. SIGMA 30mm f1.4 DC

[YT]http://www.youtube.com/watch?v=VjtTMYhwrag[/YT]

I'm happy with the results which are better than I get with the Sanyo Xacti HD1010

[DarkSideofOZ]

I hand held the K-7 for this pan. SIGMA 30mm f1.4 DC

I'm happy with the results which are better than I get with the Sanyo Xacti HD1010

Was that an AE-86 trueno?

[dom777]

I bought an used k-7 a month ago, but just started testing the video mode today. My hope was that the SR would greatly help stabilizing the image when panning handheld, plus reducing the rolling shutter effects.
But what I noticed was this:
When SR is turned on, the image seems more stable indeed. But it seems to "wiggle" around when I stop moving the camera. It happens the most after a quick pan or tilt. The Image seems to stop harsh and then move back a bit. Is this a normal behaviour or do I have to send it to the service? To me it looks like a problem with the SR on my k7! Please let me know if anyone experienced the same.

Could not find anything about this problem in the search, so I am sorry for taking over this thread

[falconeye]

The Image seems to stop harsh and then move back a bit. Is this a normal behaviour or do I have to send it to the service?

This is normal and if you think about it, is unavoidable.

With start of panning, you will force the sensor to one edge, effectively restricting its degrees of freedom from 2 to 1. This must be so because the SR will try to compensate the start of panning action. As it cannot know the difference between start of panning and shake.

Now, at end of panning, the sensor will still sit at this edge. Fortunately, the SR algorithm has a low frequency term built in which causes the sensor to always crawl into the middle to have max. degrees of freedom for the next shake.

What you would need is a pan button but there isn't any ...

[netrex]

This is normal and if you think about it, is unavoidable.

With start of panning, you will force the sensor to one edge, effectively restricting its degrees of freedom from 2 to 1. This must be so because the SR will try to compensate the start of panning action. As it cannot know the difference between start of panning and shake.

Now, at end of panning, the sensor will still sit at this edge. Fortunately, the SR algorithm has a low frequency term built in which causes the sensor to always crawl into the middle to have max. degrees of freedom for the next shake.

What you would need is a pan button but there isn't any ...

If they updated firmware to make the green button do just that, it would be great. Hold it in while panning and it won't try to adjust the panning motion. But then, which way do you pan :P Vertical or horizontal, how would it know which was to not stabilize.

[dom777]

With start of panning, you will force the sensor to one edge, effectively restricting its degrees of freedom from 2 to 1. This must be so because the SR will try to compensate the start of panning action. As it cannot know the difference between start of panning and shake.

Now, at end of panning, the sensor will still sit at this edge. Fortunately, the SR algorithm has a low frequency term built in which causes the sensor to always crawl into the middle to have max. degrees of freedom for the next shake.

Ah, now this behaviour is starting to make sense. Thank you for the clarification
Is this also the reason why the image sometimes appears to "stutter" for a few frames when panning or moving the camera quickly? I could have sworn it is an issue with the SR, but was not able to test this on a second model...