HV20 shutter wobble test
 

I shot some video at various shutter speeds to see if I could get the least amount of wobble. I was jerking the camera very fast to get the most wobble I possibly could.

The reason I did this test was to find the best setting for matchmoving. I've actually had really good solves using 60i video de-interlaced to 540p60 sequences. For some reason this has been working better for me than shooting 24p video but I still have to do more tests. I try to stick to 60i at 1/60 shutter.

http://farm2.static.flickr.com/1051/...7eec39e8_o.jpg

That's interesting. I always thought rolling shutter was more of a problem at higher shutter speeds.

I've only tried one matchmoving test using the HV20 and the PFHoe Pro software. A really simple test, but unfortunately a bit of wobble occurred close to the end of the track. Here's the clip - http://www.avene.org/movies/canberraUFO.wmv

Been meaning to try a shot with a ground plane. Just haven't got around to it yet.

Keep us posted.

Glenn, that was my thinking too until I did these tests. It's not terribly scientific as there's no way for me to duplicate the exact same pan speed for each shutter setting but I think I got them all in the same ball park and the wobble to my eye is pretty much the same on all shutter speeds and also in 24p mode. I only tried 1/48 in 24p though.

Hey Glenn, I did a matchmove here...

http://www.dvinfo.net/conf/showthread.php?t=102878

The 60i works, the 24p goes down in flames.

Wes, that's a great looking composite of the rex. It's too bad the 24p version didn't work. I'm going to have to try a few different versions myself, 25p, 50i etc. Actually, I've been planning to put a few markers on my blue screen. That could prove difficult.

Thanks Glenn!

The cadance of 60p is so much more beautiful than 24p IMHO so I don't mind it too much if 24p won't solve, but of course it's nice to only have to work with 24 frames per second when doing any kind of effects work.

I'll try to post an example later, but I seem to be experiencing wobble issues shooting the Letus35a. Its almost like the whole scene does a hula dance for a split second. Is this the wobble you guys are referring to ?

I have yet to get this wobble. Have no idea what it is. And, yes, I've seen the footage. I tried to duplicate it and couldn't.

Did I get a lemon? :)

Rob, shoot something with straight lines, an edge and jerk the camera back and forth very fast. Then scrub frame by frame. Wobble can be avoided.

I'm not trying to be a jerk...but..when one jerks back and forward in any normal shooting situation would that not be deemed unusable footage anyways? I mean...I have the HV20 and the only time I get that kind of ...."bends" is when I am not being careful in my shot....but whether it was a 3CCD cam or CMOS those types of shots are not used anyways. My only issue with wobble would be in low frequency situations...which I can avoid most of the times.

...just asking...

Ian, the wobble is simply more apparent when you jerk the camera.

This thread is meant to highlight how the wobble affects matchmoving. There are already numerous other threads discussing the wobble in general shooting circumstances.

Wes I read those threads. I misunderstood what you were trying to do. I've seen all the pictures and they all basically look the same (as far as the lean from the rolling shutter). But you mentioned this:

"I've actually had really good solves using 60i video de-interlaced to 540p60 sequences. For some reason this has been working better for me than shooting 24p video but I still have to do more tests."

Do you mean when you deinterlace and downsize the picture you are able to get better results with matchmoving? I've read this before somewhere else but am wondering how this works.

I have also seen some footage (maybe a couple of months ago) with (I forget the name of the software) that was able to do some matchmoving with very wobbly video.

Ian, I was about to ask the same thing. Why on earth would I WANT to jerk my camera back and forth very quickly?

That said, in regard to match mov3, why not simply shoot your match move footage in 60i?

The majority of your shots will likely NOT involve the process anyway, so you can do those in 24p.

I'm sorry, but it seems like a minor complaint when you consider the cost of the HV20.

....whew...I'm glad I'm not the only one who read the thread this way...but... I think I understand now what Wes is tryin to do. My thought on this however is the same as yours..keep it in 60i for those particular shots and 24p for the other majority.

My question to you Wes is, with the type of panning you are using in this test does it matter if rolling shutter artifacts are present or not? Wouldn't it be hard to try and match move even with a CCD based cam (under the same conditions that is)? I've never attempted this type of FX work before but have been interested in it for the last few months so I'm learning from your tests....sorry for my ignorance.

Glenn...I've watched this over and over again..I'm trying to find where the wobble is in this footage. i do see some camera jumps in the beginning (more like shakey cam effect instead of wobbling) but nothing at the end of the footage as you mentioned. What i do see the spaceship wobble a bit. Is that what you were referring to? The video is very still however.