Possibly new faster Deshaker

ritsmer wrote on 3/30/2009, 11:00 PM
Gunnar Thalin has released a new version (2008-11-15) of the great deshaker.
I just downloaded and tried it: because of the change to multiprocessor support it ran 3-4 times faster than the old version on my machine - and also it did not make all other apps unresponsive for long times.

Could one of you script gurus (John Meyer?) possible change the famous John Meyer Deshaker script to work with the new version?
it will mean surprising performance for all.

(I also posted this in the Vegas Pro forum)

Comments

Andy E wrote on 4/1/2009, 10:37 AM
Download available from here:

http://homepage.ntlworld.com/a.edmiston/deshaker.htm
johnmeyer wrote on 4/2/2009, 2:30 PM
Andy,

Absolutely brilliant.

You did everything I wanted to do, and much more. Plus, your approach is professional instead of a hack, which is what I did.

I certainly recommend that everyone abandon my old hack and use this instead.

The only problem I encountered (and this obviously needs to be fixed) is that if I select more than one event, as soon as the script is doing the second pass on the second event, I get the error message:

Unable to open file "william-T02.avi" for write: The process cannot access the file because it is being used by another process"

I was using 8.0c. It doesn't appear to run under 7.0d, which I expected and I understand.

I also get this message if I stabilize one event, and after the script is finished, try to stabilize a second event.
Andy E wrote on 4/2/2009, 10:38 PM
John,

Thanks for that.

I'll look into that problem straight away. It looks like I've introduced an unintended "feature" since I've been using it on more than event for a while.

I was using 8.0c. It doesn't appear to run under 7.0d, which I expected and I understand.

Yes. In fact I haven't checked it under 8.0a or b either.

instead of a hack, which is what I did.

You should carry on "hacking" then since there have been a lot of people using the Deshaker script for a long time and singing your praises because of it.

I certainly recommend that everyone abandon my old hack and use this instead.

Probably best to wait until I fix this little "bugette" :-)
johnmeyer wrote on 4/2/2009, 11:45 PM
Andy,

If you need any additional help recreating the problem, just click on my name and send me an email. I don't come to the forum anymore, but heard about your excellent work, and couldn't find any other way to contact you. So, if you email me, I'll get it for sure.
Andy E wrote on 4/2/2009, 11:50 PM
I also get this message if I stabilize one event, and after the script is finished, try to stabilize a second event.

This was the clue.

Assumption is, as they say, the mother of all......

I'm a trimmer user. A lot of the stuff I deshake is old VHS stuff converted to DV-AVI. I create subclips so the Active Take name will be unique. A test project deshaking multiple clips from the same file which I cut up on the timeline gives me the same error.

I'm testing a fix now.
johnmeyer wrote on 4/3/2009, 12:21 AM
Probably just need to increment a pointer ...

BTW, one nifty reason to do the deshaking on the timeline (I posted this, but only once before) is that deshaker always leaves the first and last frame of the clip alone.

So what?

Well, if you've deshaken a lot of clips, you know that sometimes the deshaking algorithm fails and you get all sorts of weird things happening. If you deshake on the timeline, there is a simple solution. First, go back to the original take. Next, cut the event just before, and just after where the bad things happen. Finally stabilize each new event independently. Even though you are now cutting right in the middle of a scene, you won't see anything because deshaker leaves the first and last frames exactly the way they were, so everything lines up perfectly. Even if the video in the center section (where the "bad" things were happening) cannot be stabilized, you at least have good footage on either side, and only a momentary wobble.

So, FWIW, that's why I would recommend doing the deshaking on events on the timeline rather than the trimmer.
Andy E wrote on 4/3/2009, 12:50 AM
So, FWIW, that's why I would recommend doing the deshaking on events on the timeline rather than the trimmer.

Possible misunderstanding here. The deshaking is still done on events on the timeline. It's my workflow to use the trimmer to prepare subclips before placing them on the timeline (where they are then deshaken) and a lack of thought about anyone else's workflow that's caused the issue.