Wobbly snow

[Paul Curtis]

I created a new thread here rather than add it to Colins because this is something specific, but if you look at the snow 60fps shot from

http://blog.lib.umn.edu/mcfa0086/dis...ne/102124.html

(it's the 3rd one down)

Colin says it exhibits the rolling shutter which maybe it does, but im not sure. There's a weird effect like the jello effect but i wouldn't have thought the camera was moving enough to create a rolling shutter artifact. You can see the various parts of the frame wobbling about quite obviously (unusable in fact). Im looking at the 720p download as i cannot open the original files.

I don't understand how the snow would create this as it's not moving fast enough and also the worst you'd get with a vertical rolling shutter is a snow flake disappearing!

This seems to be something else. OIS? A compression artifact? A combination.

I assume it's on the original because Colin observed it and said it was rolling shutter.

Does it do this in rain? Or other areas with changes over the screen? I hope not because that would be a real world problem. Is it only at 60fps?

Can anyone shed some more light on this?

cheers
paul

[Paul Joy]

I'd be surprised if it's caused by the snow as the wobble is happening on the rails too. It looks like it was shot through rising warm air. Was it shot through a window?

Paul.

[Paul Curtis]

Looking at the 24fps version of the snow underneath it does exhibit it but to a lesser degree. I don't think it's a window (although it looks cold enough that if it was me i would be inside for sure...)

I wonder if it's the OIS compensating madly and *that* causing a rolling shutter wobble? Perhaps 60fps is making the OIS work harder.

(Im assuming OIS was on... Colin?)

cheers
paul

[Steven Thomas]

Yes, that was strange, but under those camera conditions, it does not appear to be rolling shutter related.

[Colin McFadden]

I was outside when I shot it. That's on a pretty tight zoom as well. OIS was on, but OIS really shouldn't be able to cause that sort of effect.

If you go frame by frame, you can see the jello effect very clearly. I still think it's a consequence of rolling shutter + the overcrank framerate, which accentuates changes between frames.

The only other thing it could be in my mind is an issue with the codec - motion vectors getting screwed up. It's possible, but I would think I would see evidence of that in other places as well.

I'm just as curious as everyone else though!

[Colin McFadden]

To try and diagnose a bit further, I just did a little screen capture of me scrubbing back and forth over a few frames in the original. Might help?

Original file (Apple Intermediate Codec): https://mediamill.cla.umn.edu/mediam....php?orig=4572

H264 640x480 version:
https://mediamill.cla.umn.edu/mediamill/embedqt/6881

[Paul Curtis]

My thinking about the OIS was

- the OIS is a mechanical solution that moves the lens elements around
- the snow is confusing the electronics driving the elements

so i'm guessing that internally the lens elements might be shifting around imperceptably to stablise something that doesn't need stablising, they're doing this so fast that the rolling shutter is being shown up.

But screwed up motion vectors could also explain it, however you'd have thought that the compression algorthim would handle that. In fact we know mpeg2 can handle that otherwise we'd get no DVDs with snow scenes!

You can see the effect, less pronouced in the 24fps version too.

It's odd. Is it still snowing there? Maybe having a go without OIS would solve this?

I don't see how the rolling shutter would be shown up with simple movement like that. the whole view has to change a lot from line to line to show up during the read out phase and the background is not moving that much. We'd see the jello effect all the time if hand holding on a long lens would be enough.

cheers
paul

[Colin McFadden]

We may have some snow tomorrow morning. I also passed the video along to a technical contact at Sony to see what they make of it.

[Colin McFadden]

The reason I don't think that it's OIS is that OIS should be independent of what's actually being recorded - they do it based on accelerometers/gyroscopic sensors in the body, don't they? I'd certainly believe it if the camera had electronic stabilization (see my post here http://blog.lib.umn.edu/mcfa0086/dis...ne/098974.html) but... hrm..

[Steven Thomas]

I Agree.
Well, just in case, try it again with OIS OFF, then ON.

[Forrest Burger]

Colin, were you in auto focus or manual focus on this shot?

[Colin McFadden]

I believe it was autofocus

[Forrest Burger]

I wonder if that might be coming in to play...the camera trying to focus on the snow flakes at different focal lengths, combined with the OIS?

[Brad Vaughan]

Originally Posted by Colin McFadden

I believe it was autofocus

I'm thinking that's the issue right there.

[Ethan Cooper]

First impression was that it might be the rolling shutter "jello" look, but it just didn't make sense. It seemed to be isolated to certain areas and rolling shutter should be apparent over the entire frame.
My next guess was OIS strangeness but I dunno.. that just didn't seem right either.
I've just about convinced myself that it's a thermal of some sort.
But then again someone mentioned AutoFocus.

Go full manual and see if you can get this to repeat.