[CtrlAltDel_ru]

I have made a test of SR on K-01, K-5IIs, K-3
I can not say K-3 stabilization is much better or much worse.

You Tube

You Tube

[TomGarn]

I have made a test of SR on K-01, K-5IIs, K-3
I can not say K-3 stabilization is much better or much worse.
Video SR test (K-01, K-5IIs, K-3) - YouTube

Thanks !
Good test in tele - wide angle is to slow for this test

K-3 is by far the best ... K-5 just wobbles terribly as usual ! (just like K-7)
(AF is a nuisance, btw.)

[grispie]

K-3 is by far the best

I agree

[grispie]

the proof is in the eating
walked around in my dad's garden as if i was a tourist.
All shots handheld with SR on

ignore the under/over exposures...



Read more at: https://www.pentaxforums.com/forums/75-video-pentax-hdslrs/244233-test-video-...#ixzz2mpHeEKPh

[kadajawi]

I have made a test of SR on K-01, K-5IIs, K-3
I can not say K-3 stabilization is much better or much worse.
Video SR test (K-01, K-5IIs, K-3) - YouTube

Ok... so the K-5 does move back a bit after a pan, while the others don't. You'd only notice it when you look for it though, and it looks to me like something they could have fixed in the firmware with some more sophisticated algorithms... which they probably have in the K-01 and K-3, since those SHOULD have the same problem, unless Pentax is using a more sophisticated algorithm there. Basically I think they could just port that over to a mechanical system.


In this case, perhaps due to faster shutter speeds the K-3 looked rather good... but you just can't always do that. The K-5 corrects less than the others, but it doesn't have the artifacts. To me I'd rather have less artifacts... the warp stabilizer can take care of the last bit of shake IF necessary (though that might put back some artifacts, but I'd expect less than you'd get by relying on a fully electronic system.


Warp stabilizer is a more sophisticated system than the electronic SR used in the K-3... and it can analyze the whole video to smooth it out perfectly, rather than having to guess where you are going to go. The only disadvantage is a slight loss in resolution compared to the built in systems, and that it can get confused by objects moving in front of the camera (while the camera thanks to accelerometer data can avoid it... it would be great if the camera could also record that data, and whatever stabilizing software you want to use can make use of the data). But combining warp and mechanical SR seems to be the way to go for me, as you get a sort of stable video into the camera, with less artifacts, and the electronic SR in the computer will take care of the rest.


Ok, a couple of days ago I had the K-3 in my hands for a short while. We were mostly trying out some bug (?) in combination with the DA 50 1.8, but for a short while I also put in my SD card and took a short video to test the stabilizing system. On the screen there didn't seem to be any stabilization when I shaked the camera left/right a bit... or at least very little. I've just looked at the footage though... and it looks stabilized. Very stabilized. And absolutely awful. lots of jello, blurrying, ... Obviously you wouldn't shoot like that in real life, but... a K-5 wouldn't look like that. And if it falls apart so badly in that test, chances are it will also, to a lesser degree, in real life.

[Dave L]

Ok... so the K-5 does move back a bit after a pan, while the others don't. You'd only notice it when you look for it though, and it looks to me like something they could have fixed in the firmware with some more sophisticated algorithms... which they probably have in the K-01 and K-3, since those SHOULD have the same problem, unless Pentax is using a more sophisticated algorithm there. Basically I think they could just port that over to a mechanical system.

In this case, perhaps due to faster shutter speeds the K-3 looked rather good... but you just can't always do that. The K-5 corrects less than the others, but it doesn't have the artifacts. To me I'd rather have less artifacts... the warp stabilizer can take care of the last bit of shake IF necessary (though that might put back some artifacts, but I'd expect less than you'd get by relying on a fully electronic system.

I find the jello/swaying in the electronic stabilization more disconcerting than any judderyness in the mechanical stabilization, the latter being something I find I can avoid in my K-5 video with careful panning, and/or largely correct in post production. I'd like to see K-5 vs. K-3 comparisons all done at the kind of slow/careful panning that I do, slowest at the start and end of pans; all the test videos I've seen to far seem to focus on unfeasibly fast or sudden panning of the kind that I for one would not do and expect to see good results. We now know the limitations of both systems so it would be good to see what can be achieved handheld within each camera's limits, but with similar 'as smooth as possible' panning techniques used for each, rather than attempting to test the limits.

[TomGarn]

all the test videos I've seen to far seem to focus on unfeasibly fast
or sudden panning of the kind that I for one would not do and expect to see good results.

If you watched that video of the green pool in the forest and the old house,
you could see most pans were very slow and not as fast as you described it.
(You really saw it ?) This kind of roll-back is completely unacceptable of course.
No need to be indifferent.

That quicker tree-to-house-pan could to be avoided, with SR on
but it shows what is happening there mechanically, and with usual
cameras that speed of course is acceptable (and without SR too)

I won't use the SR anymore - while panning with my K-7 - same for K-5
(Or will try to learn new ways of moving, which I doubt now will really work)

The K-3 does the panning-job much better seen in that private-room up above.
(I would have switched to that software-SR while panning if this always works so well)

In my K10 I do have a switch for SR at the camera-body. A shame they put that down.
Nowadays - with two ways of SR possible - and the third choice to simply put it off,
this switch would be the only way to satisfy the user when in search for stabilisation.

[gazonk]

In my K10 I do have a switch for SR at the camera-body. A shame they put that down.
Nowadays - with two ways of SR possible - and the third choice to simply put it off,
this switch would be the only way to satisfy the user when in search for stabilisation.

On the K-3, you can program the RAW/Fx-button to turn SR on and off. I wish they would add that to my K-5 with a firmware update!

[Dave L]

If you watched that video of the green pool in the forest and the old house,
you could see most pans were very slow and not as fast as you described it.
(You really saw it ?) This kind of roll-back is completely unacceptable of course.

Thank you and yes, I remember it now! But it looks jerky at times when viewing the pans on Youtube in a browser so I downloaded the 720p version and hey presto! No jerkiness in VLC Player! The green pool pan and most of the others I find to be acceptable with the mechanical K-7 shake reduction but there is of course that "roll-back" effect at the end of all but the very slowest pan and tilts (though oddly, it didn't bother me).

I was interested to see what happened when running your video through the software shake reduction in Powerdirector 12, so I tried it. I set it at the maximum that allowed all your captions to remain on screen, 75%, and haven't tried any of the other settings. Obviously, as a third generation re-encode via a low-bitrate Youtube 720p video there is a loss of definition but I hope you don't mind if I share it with you? I uploaded it anyway and will remove it if you wish.

I found that some parts were improved, some were the same and in a few sections, it was slightly worse but I think it's good to have the option to apply software SR to all *or parts* of a mechanical SR video. I think that's why I prefer mechanical SR but I don't know if computer software SR on top of camera software SR would work as well. If you can download my file, the differences show up most clearly when playing both versions side by side and synchronized, in two instances e.g. of VLC Player.

You Tube

You Tube

[kenyee]

Doesn't software SR need extra resolution?
It's too bad you can't save whatever readout the chip had so you can do SR on your desktop.
H.264's next resolution up from 1080p is apparently 2,560×1,920 which I doubt the K-3 is doing :-P
H.264/MPEG-4 AVC - Wikipedia, the free encyclopedia

[kadajawi]

Doesn't software SR need extra resolution?
It's too bad you can't save whatever readout the chip had so you can do SR on your desktop.
H.264's next resolution up from 1080p is apparently 2,560×1,920 which I doubt the K-3 is doing :-P
H.264/MPEG-4 AVC - Wikipedia, the free encyclopedia

You lose a bit, depending on how much has to be corrected and how you set it, but it can work without being very noticeable.

[Painter]

Here is some K-3 video I shot it was unplanned in retrospect I wish I had tried some 720 60p and tried it slow mo but anyway I had it set for 1080 60i to go to the second card slot. Vimeo free doesn't let me post full HD so they re-encode it but I was happy with the results. Also using the AF button to refocus with a DA* lens works really well.

Vimeo

[kadajawi]

The wobble near the end is quite bad...

[TomGarn]

Thank you and yes, I remember it now! But it looks jerky at times when viewing the pans on Youtube in a browser so I downloaded the 720p version and hey presto! No jerkiness in VLC Player! The green pool pan and most of the others I find to be acceptable with the mechanical K-7 shake reduction but there is of course that "roll-back" effect at the end of all but the very slowest pan and tilts (though oddly, it didn't bother me).

I was interested to see what happened when running your video through the software shake reduction in Powerdirector 12, so I tried it. I set it at the maximum that allowed all your captions to remain on screen, 75%, and haven't tried any of the other settings. Obviously, as a third generation re-encode via a low-bitrate Youtube 720p video there is a loss of definition but I hope you don't mind if I share it with you? I uploaded it anyway and will remove it if you wish.

I found that some parts were improved, some were the same and in a few sections, it was slightly worse but I think it's good to have the option to apply software SR to all *or parts* of a mechanical SR video. I think that's why I prefer mechanical SR but I don't know if computer software SR on top of camera software SR would work as well. If you can download my file, the differences show up most clearly when playing both versions side by side and synchronized, in two instances e.g. of VLC Player.

Muztenbecher Villa - Hamburg 2013 (Pentax K-7 SR Test) [shake reduction @ 75pc applied in PD 12] - YouTube

Wow - great test you did here. It really works and you can watch it while compensating
Nice to see it really caught all those nasty roll-backs well - except that very quick one ...
It works ! ... and I know those painfull outtakes of mine by heart. I am amazed it can see
what it got to do: That's fun to watch.

Just what we love - the sharp resolution and the original wider angle - that loose a lot - but
that may be acceptable for some shots you really need - but correct them - if possible.

Powerdirector 12 - Can that be added to Premiere CS6 ?
I didn't look for software like that before ... taht may even be allready included on Premiere ?

P.S. You may leave that video online of course - but please just give it a new title. Thank you !

[kadajawi]

@Tom: Powerdirector is probably some editing software like Premiere, just more consumer focused. Just play with the warp stabilizer that's included with Premiere. Electronic SR, regardless of in camera or on the computer, only works with short exposure times. As soon as you get motion blur due to camera shake it's game over. That's where mechanical SR shines.