Nokia Lumia OIS Jello / wobble due to motor vibration

[Mr. Aldi]

I have a numerous of shots from a Nokia Lumia 930 used as a dashboard cam. Unfortunately the OIS cannot be turned of and the motor vibration has caused a severe jello effect in the shots. It's like the video is being projected on a screen that moves in the wind.

I have tried to reduce this with Virtualdub Deshaker, but this only reduced the normal x- and y-axis motion. I also gave the warp stabilizer of Adobe a try but no luck either. It looked like it did reduce it a bit, but very long run times and no stable result.

Am I out of luck to make those shots a bit prettier?

10 MB, 10 sec clip: https://dl.dropboxusercontent.com/u/84771088/WP_20150225_003.avi

Thanks.

[pippas]

Classic CMOS sensor rolling shutter 'jello' effect I'm afraid. Unlike some of the other 'rolling shutter' problems that you can do something with -- 'skewed' verticals or 'flash banding' for example - there's not much - if anything at all - you can do to improve 'jello' distortion.
And it has nothing to do with the OIS either...

[racer-x]

Garbage. Either foam mount your camera phone or get a better camera.

[Mr. Aldi]

Heh heh, ok thanks guys. Now I can let it rest

The foam mount or some dampening is a great idea. Thanks!

[lordsmurf]

Phones are not cameras.
You'll always get low quality issues like this.

[pippas]

Garbage. Either foam mount your camera phone or get a better camera.

No, not garbage .... simply obeying the laws of physics !..

The OP mentioned using this phone camera as a dash cam. That will vibrate. No amount of 'foam mounting' will stop all the rolling shutter 'jello' with the slow scan CMOS sensors used in phone cameras.

You might improve the situation a bit with the faster scans of modern camcorders, and a mount to minimise vibration (foam or otherwise) - but when used as a dash cam any rapid vertical motion will probably cause some 'jello' distortion.

The only way to overcome the problem completely is to use a camera with a global shutter. All CCD cameras use those.... and there are now one or two global shutter CMOS cameras starting to arrive on the market. But they are still expensive.

For more details on the physics of the problem read here: http://dvxuser.com/jason/CMOS-CCD

And for an (exaggerated) example of the difference between CMOS and CCD sensors in high vibration situations, see here:

https://www.youtube.com/watch?v=fSWDP1Tsj1Y

As I mentioned above, OIS doesn't help this problem. If a camera with a rolling shutter moves in a vertical plane, faster than the sensor scan speed, then you will get 'jello' distortion. And it can't be fixed in post.

As I say, you can't beat the laws of physics.