[clackers]

I wondered about whether the rule applies to electronic SR. I did a couple of clips with my K-3 and there was no SR artifact that I could see with camera on my heavier tripod. I changed to a longer lens (very heavy 70-210/3.5 Tamron 19AH), unbalanced on head, and physically thumped on the tripod legs and was unable to get the frame to jump. I don't know how things might be with a lighter tripod at full extension or with the legs set in a moving stream, however.


Steve

I think you have a point, Steve. Since the sensor's not moving, there wouldn't be any feedback through the tripod.

[stevebrot]

Really, AF changing all the time is a bug-a-boo in the few times I have attempted video on my K-3II.

Yep, really. The K-70 is the first Pentax dSLR that supports continuous AF while shooting video. I just tried tracking AF mode on my K-3 and while it did a good job of tracking prior to full shutter press, it stopped doing active focus at that point. I am able to do AF on demand during video capture when in multi-point and single point modes, with both shutter half-press and with the back button, but even then, it will only do one attempt per press, at least on the K-3 firmware v1.30. I did not try this using wired or IR remotes.


Steve

[stevebrot]

I think you have a point, Steve. Since the sensor's not moving, there wouldn't be any feedback through the tripod.

There might still be a fault with the sensor's static suspension on the OP's camera that is not related to SR.

I don't do much video, but in general have found the feature to be stable, if not particularly current or flexible. Something like the example on this thread is outside anything I have seen from my camera.


Steve

[stevebrot]

"The K-3 will autofocus during recording whenever you press and hold the AF button. The focusing is very slow (likely to keep audible noise levels down) and tends to hunt with longer lenses"

Yes, though it will not do continuous focus even with the button depressed and even in tracking mode. I would have used this feature in the past had it been available. When you press the AF button, there is a sequence of faint clicking sounds which may or may not be followed by a focus attempt. If attempted, the action will be slow and will stop if focus is attained or if it does a full rack without success. One must release and press again to do another try. I also checked to see if the previous focus point was being tracked for additional attempts. If so, the tracking is not very effective.

I don't know what is happening with the example video and spontaneous operation of the AF system may be part of the mix. If so, a fault is present since the K-3 normally does not support continuous AF adjustment.


Steve

[clackers]

It's not just video ... AF-C has never been available in Live View stills either.

Perhaps the K-70 features it now?

[Azergoth]

Hi guys, thanks for your replies

You should almost always disable SR in video mode with the K-3, because the camera doesn't offer mechanical stabilization and only offers the poor electronic solution, which can result in the nasty side effects you're seeing (especially when the camera is already stabilized by a tripod or monopod).

Now I know it and will certainly do in the future. I don't shoot video very often and I don't care much about video features, but it's a "nice to have" for occasional use. Now I know

Electronic stabilization is driven by inertia and detected by an array of microgyros on the main board. Movement within the frame will not trigger video SR on the K-3 (I just checked). I just reviewed your clip and there is a general upward drift* of the frame from start to finish with a much smaller leftward drift. also. The branch is moving quite a bit throughout the clip (is floating), but the jerking motion and frame shift affects the rocks visible on the stream bed too. If it were my clip, I would suspect the head was not fully secure and may have been assisted by a breeze. Were you using suspended weight from the tripod?



Let's just assume we don't know if it was on. Video SR is set separately from still photo SR with the setting retained between sessions and power cycles.


Steve

* ...meaning that the subject is drifting downward

Good point. I was supposing the SR system analyzed the scene instead of using the gyroscopes (as for the mechanical SR). In this case, I was on a pretty sturdy tripod (Brelebach + Sirui billhead). Not the highest end carbon tripod, but good wooden tripod. The ballhead is also a "mid range", decent ballhead. With the lens well balanced on its collar, I don't think there was much motion at all. And I was sheltered from wind in a tent, also

Have to switch SR (electronic or mechanical) off when a camera's on a tripod, Azergoth!

For video, I trust you. But for stills, I've never turned it off and it has never been an issue. And I shoot quite a lot nightscapes and have had any trouble. I'd be happy to be proven wrong, though, but I think this is more of a myth of the early days than reality.

I suggest that you look at your focus mode in LV. (pg 46 in the manual)
Focus mode set to "Tracking"
"Tracks the subject in focus.
The shutter can be released even if the subject is not in focus"

I propose that the camera is seeing the water and focusing on its movements, therefore giving the "jerky" motion which is remarkably synchronized with the "flow".

A suggestion, not a end all, but change the focus mode in LV.

Indeed, I was thinking something similar but regarding SR. I use back button AF, so no AF here.

I wondered about whether the rule applies to electronic SR. I did a couple of clips with my K-3 and there was no SR artifact that I could see with camera on my heavier tripod. I changed to a longer lens (very heavy 70-210/3.5 Tamron 19AH), unbalanced on head, and physically thumped on the tripod legs and was unable to get the frame to jump. I don't know how things might be with a lighter tripod at full extension or with the legs set in a moving stream, however.


Steve

Is this the same DA* 300mm that you are using a piece of foil on as a temporary power contact until the replacement part comes?


Steve

Yes it is. If I remember well, it's the day before it broke. Or was it already broken and has it impacted the SR? Anyway, I use back button AF, so it shouldn't batter, right?

To me it appears that the sensor is locked onto the highlights of the water - moving left to right.

That's why I was thinking about a SR issue

---------- Post added 08-20-19 at 11:59 AM ----------

Ok, I have in fact other videos from those days. Some look "good" and some are also "jerky". But they were shot in the same conditions and with the same settings. So, granted, their cinematographic quality is open for debate, but some are "good" and some bad.

This one is stable, not much water in the scene:

Youtu.be

Youtu.be

This one, on the other hand, is also jerky. Note the abundance of water in the scene:

Youtu.be

Youtu.be

Less water and stable again:

Youtu.be

Youtu.be

[Azergoth]

Ok, I've been thinking about a silly test: tryin to replicate this behavior ... at home. So here is what I did: found some video on youtube that might trigger the same behavior, get my setup as sturdily as possible and film the screen. The result is astonishing:

Here's the video i've been filming:

Dipper - Cinclus cinclus (gularis) - YouTube

Without SR, stable as it could be. Note I've intentionally left the progress bar to asses motion.

Youtu.be

Youtu.be

With SR, jerky as I've encountered:

Youtu.be

Youtu.be

My conclusion is: the camera IS using the image content for the SR an NOT the accelerometers.

[clackers]

My conclusion is: the camera IS using the image content for the SR an NOT the accelerometers.

Thanks for the experiment, Azergoth, I'll remember that!

[PDL]

Again, check your settings for Live View and see if tracking is turned on.

Run your SR on, SR off test again, but with Live View set to --- grid and turn off video SR. Just try it and see what happens.

You say you are using Back Button Focus - OK.
You agree that the sensor is locking on (tracking) the highlights of the water.

Just change the Live View mode to anything other than tracking. It appears to me that the tracking mode of Live View is being carried over to video, since you have to be using Live mode to shoot video.

[clackers]

Again, check your settings for Live View and see if tracking is turned on.

Run your SR on, SR off test again, but with Live View set to --- grid and turn off video SR. Just try it and see what happens.

You say you are using Back Button Focus - OK.
You agree that the sensor is locking on (tracking) the highlights of the water.

Just change the Live View mode to anything other than tracking. It appears to me that the tracking mode of Live View is being carried over to video, since you have to be using Live mode to shoot video.

Actually, that's very possible.

A good test is simply to go into Manual Focus.

That's what film making and cinematography's based on, anyway.

[Azergoth]

Again, check your settings for Live View and see if tracking is turned on.

Run your SR on, SR off test again, but with Live View set to --- grid and turn off video SR. Just try it and see what happens.

You say you are using Back Button Focus - OK.
You agree that the sensor is locking on (tracking) the highlights of the water.

Just change the Live View mode to anything other than tracking. It appears to me that the tracking mode of Live View is being carried over to video, since you have to be using Live mode to shoot video.

Hi PDL, no I'm not in Tracking mode. I'm in "select" mode, if you must know. But it's NOT and AF issue, AF not doing ANYTHING because I'm not touching the AF button. With back button AF, it's full manual AF unless you touch the AF button.

Yes, the sensor is tracking the water, but NOT the AF. That's all the point

Actually, that's very possible.

A good test is simply to go into Manual Focus.

That's what film making and cinematography's based on, anyway.

I was in manual focus --> back button AF 😉
CDAF in video is poor, by the way. I'd avoid it as much as possible in video, as you mentioned.