followed the directions exactly using the source footage for the tutorial and while it worked, I noticed some shimmering at the bottom of the screen. I have no clue what I did wrong.

First of all, DON'T use the clips that are part of the tutorial. Those are crummy WMV files, suitable for streaming, but not much else. They were meant to simply show what the video looks like before and after. They are definitely not suitable input for Deshaker.

Second, you will always get some "weirdness" at the edges as Deshaker attempts to fill in what would otherwise be black edges, as it moves its equivalent of the pan/crop around. It fills in from video that it finds in previous and later frames, but obviously this cannot be perfect, so you sometimes see strange video at the edge.

If you want to "play around" with Deshaker, I suggest you find some shaky DV AVI footage of your own and use that instead.

John,
I remeber some discussion some time ago with the author of deshaker about the source pixel aspect of NTSC needs to be .909 instead of .911 (default).... or was it the other way around?

Very good info there. Interesting thing is the results were much the same as the ones I'd managed to get the last (and only) time I've tried to fix wobbly footage. It's better for sure but still not 100% stable.
I seem to recall a demo of the one in one of the Avid suite, you could nominate a corner / edge and it'd keep that exactly where it was in the nominated frame. I'm certainly not complaining for the price but it doesn't achieve the magic of transforming a handheld shot into a tripod shot.
/edit/
I just went back and read ALL the guide. In all fairness I should add the footage I was trying to fix had a raging bonfire in about 30% of the frame, it was dusk and the video was pretty noisy in the blacks. Worst possible scenario for generating motion vectors. Funny how when a client thows you a curved ball it's curving in 4 dimensions!
Bob.

In the past, utilities like Virtual Dub was viewed as pirate tools. Spot, are you shifting your view about these tools or just bending with the wind?

Farss, If your serious about tracking camera motion and motion in the scene, you should be using Boujou_Bullit or Luxology_modo or Realviz_matchmoverPro or last, AE6.5pro. Really, the first two are best, outside Apple's world.

and dont forgot combustion 3, it has a great motion tracking/ stabalizer

How were they ever viewed as 'pirate' tools. There's a big difference between something you don' t pay for but you should and something that's someones work that they decided to give away.
Although interestingly enough someone I was talking to recently seemed to think some of the original SoFo audio stuff was 'pirated' code.
Bob.

I've NEVER viewed virtual dub as a pirates tool. Not like I once (and still sorta do) viewed DivX, and how I still view BeSweet and similar tools.
Like the guy who posted this morning.
"Hi, uhh, I've got this DVD that I need footage from, it's of my family, and like...it's like....5.1 surround, and I need to get the audio....how do I do that?" THAT is a pirate looking for pirate assistance warez.
Virtualdub may have been and probably is still used by the enclave of those that steal media, but that's not what it was developed for, and I think it's got some serious credibility for some things.

Apit34356:
Can you elaborate on those softwares?

JJK

Personally, I like Boris RED's tracking capabilities much better.

I remember some discussion some time ago with the author of Deshaker about the source pixel aspect of NTSC needs to be .909 instead of .911

That never came up in either my research or in my discussions with Deshaker's author. (BTW, when I mention "the author" in my guide, I am talking about Deshaker's author, not myself. I realized when re-reading it yesterday after Spot had cleaned it up and posted it sounds as though I am referring to myself in the third person).

I seem to recall a demo of the one in one of the Avid suite, you could nominate a corner / edge and it'd keep that exactly where it was in the nominated frame.

I didn't talk about this capability in the guide, because I didn't like the results, but if you read the author's (that would be Deshaker's author ... ) guide, he talks about using values of -1 for the motion smoothness settings. This "locks" the camera on the position of the first frame. If you want absolute stability, try this, but things will get very strange pretty quickly if your camera doesn't stay very close to its initial position.

In the past, utilities like Virtual Dub was viewed as pirate tools.

I know this comment was intended for Spot, and he has already answered. However, I really don't think VirtualDub has any relationship to the tools used to pirate movies (like Smartripper, DVD Decrypter, DVD Shrink, etc.). In fact, I am not sure how you would use it for any of these purposes. Most people use it as either a capture application (it has some extremely interesting capture properties not found in Vegas or even in Scenalyzer, such as the ability to modify the video properties during capture) or as a signal processing app (which is what I use it for).

There is a TON of amazing software out there -- much of it free, or inexpensive shareware -- that everyone should have in their bag of tricks. Deshaker, VirtualDub, and TMPGEnc (which is useful for cutting MPEG-2 files without requiring recompression) are just a few of the examples. Even DVD Shrink, which is primarily used to "backup" Hollywood DVDs, is extremely useful for the legitimate user. For instance, suppose you just spent ten hours doing a 2-pass, best quality MPEG-2 render only to find that the files are 101% of what your DVD will hold (although don't believe the DVDA estimate -- you'll actually have to prepare the DVD before you find out for sure). Do you want to go back and spend another ten hours re-encoding at 99% of the previous bitrate, or do you want to use DVD Shrink and be finished in ten minutes (with virtually no detectable decrease in quality)? Easy answer, at least for me.

So, even DVD Shrink has legitimate uses.

"In the past, utilities like Virtual Dub was viewed as pirate tools." This stated was for Spot, but not in a personal way, we all know Spot does not support illegal piracy, but more in the effect on the professional business relationships with companies like Sony, CBS, NBC, etc.. being less sensitive to these software apps being promoted or demonstrated at one's website.

"illegal piracy" was a term explained to me by a German businessperson as meaning "high profit
pirating of big ticket USA items".

Okay, so can someone give me the "bottom line?" Before I go and learn a new routine, how does DeShaker compare to SteadyHand?

Before I go and learn a new routine, how does DeShaker compare to SteadyHand?

The whole reason I wrote the guide is that I had been using Steadyhand, and every time the subject of stabilizing video came up in these forums, people that had used both recommended Deshaker. I finally tried it and decided that it is definitely better than Steadyhand.

And you can't beat the price ...

And Dynapel hasn't enhanced their product for years ...

It blows Steadyhand out of the water. Less blur, faster.

Deshaker author speaking... :)

I just thought I should comment on the pixel aspect ratio confusion.
The exact ratio used when you select NTSC in deshaker is 72/79 (=0,91139...).
It was taken from this excellent page:
http://www.uwasa.fi/~f76998/video/conversion/
But the guy who made that page recently changed the ratio to 4320/4739 (=0,91158...).
The difference is very small, but I'll fix it in the next release.

Note that if you believe in that page (and I certainly do), most video programs are using the wrong pixel aspect ratio. Vegas, for example, uses 0.9091.

For info on the math conflict, check here:

Does Vegas have a PAR bug?

Quoting SonyEPM:

Some details in how our numbers are derived:

Funny, I'd written you (Gunnar) about this issue when I first discoverred Deshaker -- and have since just manually changed PAR in all my projects. Otherwise, it's an EXCELLENT tool and I spend a lot of time extolling the virtues of it and VDub to others here and elsewhere -- can't thank you enough for creating a great tool -- even if your PAR values are off :D :D

Incidentally, the article you mentioned was brought up in a previous thread, here.

- jim

I don't know the exact math behind this, but I recognize knowledge when I see it, and the guy who wrote the page I was refering to (Jukka Aho) has tons of it. He definitely seems to know what he's talking about. Did you read the whole page?

Also, I read your forum links and the SonyEPM guy even admits there's a chance he might be wrong. And SonyDennis doesn't say Jukka Aho is wrong, just that the difference is very small and you are free to use whatever PAR you like. Which is very true, but doesn't prove anything.

But this is a rather silly discussion as none of us seems to know anything about this. We're just refering to what other guys say. Please use the PAR you like and I will too. Personally I think the video looks plain ugly when I use 0.9091 :)

piracy ,math geeks, calculations that leave my head spinning

Almost forgot,very informative tutorial John , and great filter Guth
Thanks

winrock, you are correct! Thank you for very informative tutorial John for the masses, great work on the filter Guth, and thanks to Spot for helping promote free ulitities that are needed!

Personally I think the video looks plain ugly when I use 0.9091 :)

Huh... I had the same reaction when I rendered out using .9114 -- everyone looked fatter :D Guess that explains why they say the camera "puts about 20 pounds on"!

At any rate, like I've said -- you've put out an amazing filter, one that runs circles around everything else I've tested and paid much more for -- can't thank you enough!

- jim