motion stability [??]

[mattias]

this is from istabilize. took 20 seconds plus export. fcp's single point algorithm failed miserably on such a low res clip. it really needs one sharp and contrasty point to track. plenty of those in the high res version i assume.

http://www.mattias.nu/stuff/stability_istab.mov

/matt

[johnnhud]

...here's an example[/url] from somebody with more than 3 hours of experience in shake(sorry, but couldnt help it ;)
++ christoph ++

Looks freaking awesome! All I did was drag in the file, apply the image stabilize node, do a 2 point track and didn't change any of the variable information in Shake, then file out to a quicktime h.264 file. I was pleased with my results and knew that I could do better with the high rez file, but you made that lower rez file rock solid steady.

Another cool thing about Shake is that the trial version does not apply any watermarks or anything to your final export.

Again christoph, I'm impressed. What did you do that was different than I did?

[steve hyde]

..These examples are amazing!

I'm working on preparing a higher res file. What compression settings should I use? Should I set key frames to make the file smaller? The 20 second clip is 613mbs. I'm still learning about video compression so, as always, advice is welcome and appreciated.

Steve

[johnnhud]

if your using quicktime pro, choose to compress codec h.264 at 640x480.

if your not using quicktime, download DIVX and use it's converter to compress a 1500Kps 640x480 or you can even choose a profile like (High/HD) and not even have to do custom settings.

http://www.divx.com (it's free, quicktime's not)

[steve hyde]

....I am using the compressor in Adobe Premiere Pro 2.0. This time I used H.264 at 720 X 480 for a higher resolution file that packed down to 34mbs:

http://www.steve-hyde.com/03.mov













Steve

[mattias]

quicktime is free. quicktime pro is not, but that's just a special version of the *player application*, it's the exact same version of quicktime. there are plenty of programs that export using the quicktime engine that are free. at least on the mac.

/matt

[steve hyde]

....Premiere Pro does it. I used a PC to make that .mov

Steve

[Mafishus]

There is a 3rd party plug-in that comes with Premiere 2.0. It is on the install disc in a folder called 3rd Party Plugins (I think). It does a pretty decent job and costs you nothing additional. Just apply it to the clip on the timeline and render. Good luck!

[johnnhud]

....I am using the compressor in Adobe Premiere Pro 2.0. This time I used H.264 at 720 X 480 for a higher resolution file that packed down to 34mbs:

http://www.steve-hyde.com/03.mov

Steve

and here is mine.......http://homepage.mac.com/WebObjects/File ... lic/04.mov

[christoph]

What did you do that was different than I did?

well, basically i averaged 4 different tracking points in one stabilize node, which is common technique with any graniy/noisy footage. in other words make a Tracker1, add 3 extra tracks, examine the results and tweak if needed, then put an expression like:
(Tracker1.track1X+Tracker1.track2X+Tracker1.track3X+Tracker1.track4X) /4
into the track1X field of the stabilize node, and
(Tracker1.track1Y+Tracker1.track2Y+Tracker1.track3Y+Tracker1.track4Y) / 4
into the track1Y

also turning on the preprocess option or putting a slight blur in front of the tracker (not the stabilizer!) often helps as well.

This time I used H.264 at 720 X 480 for a higher resolution file that packed down to 34mbs:

for any tracking/stabilizing work, any recompression is nearly as bad as scaling, specially with keyframes. what you want to use is a copy as close to the original, so in your case, a 3 sec clip of the original file would be best... share using a torrent or a free file hoster if you dont have enough webspace.
your recompression (or original clip?) does also show som interlacing artifacts, which makes the job even harder, but all that said, here's probably as good as it gets with the clip you posted.

++ christoph ++

[steve hyde]

Thanks for taking the time to work on these clips. The results you guys have come up with makes me optimistic.

Cristoph,

Do you think I would see better results with this technique if I captured the footage from the Digibeta vs. the DV? I assume the Digibeta would look better since there is more information..

Once I get a rough cut, I'm thinking I might ask the posthouse to dump the DigiBeta material, with handles, onto an external drive so that I can do these kinds of post effects with better results.. Is it worth it?

Steve

[christoph]

Cristoph,

Do you think I would see better results with this technique if I captured the footage from the Digibeta vs. the DV? I assume the Digibeta would look better since there is more information.. Is it worth it?

well, depends on how dear these shots are to you and what is going to happen with them later, and your taste for estetics..

for example personally i never stabilize super8 clips 100%, because they loose a part of the characteristic that we like (so usually i reduce it by 50-80% depending on shot and editing)..

the sample you posted seem to have some interlacing, i'd check the original if it's there as well, and if it's standard 3:2 pulldown that can be removed. if it is, the DV footage is good enough if all you need is a DVD... if it's random interlacing, i'd definitely retransfer for stabilizing, and if you can handle/afford a digibeta workflow, the better.

++ christoph ++


addon: reading your post again, i now realize that you probably have a digibeta transfer already and are working with a DV dub for editing. if you dont have a lot of special effects, a clean EDL, and timecode accurate dubs and captures making an online from the digibeta is kind of easy and shouldnt be too expensive (prices for these kind of work vary a lot, depending how well you know the guys, their workload, if they like the project or not, and if you are a cute girl or not ;)
still double check if you can get rid of the pulldown (which is where i assume the interlacing comes from), then stabilize, then reapply puldown for the TV copy.

[steve hyde]

....Thanks again to all for your helpful suggestions and experiments with my clip. At this point I really need to educate myself on digital video methods and theory. My understand of pull downs, key frames, compression algorithms, chroma keying etc etc.. is pretty fragmented. I have ordered a couple of books on the subject. Here is what I found. Does anyone have any suggestions on books that cover this stuff clearly and coherently? Is there anything worth reading that is more current than these?



Poynton, Charles A., 1950-
A technical introduction to digital video / Charles A. Poynton New York : J. Wiley, c1996

and

Symes, Peter (Peter D.)
Video compression demystified / Peter Symes
New York : McGraw-Hill, 2001


Steve

[christoph]

At this point I really need to educate myself on digital video methods and theory.

hmm.. from personal experience i can tell you that it really takes quite a long time until you really understand what's going on, and even if you invest a lot of time it hardly ever helps with the inspiration for a good film.

however, it does help to save a lot of money and getting the best technical results once you do sit down and actually leave the books to film something, but basically it comes down to if you like doing this kind of things, or if you absolutely need it for your kind of work and cant afford to pay anybody else.

++ christoph ++

[steve hyde]

At this point I really need to educate myself on digital video methods and theory.

hmm.. from personal experience i can tell you that it really takes quite a long time until you really understand what's going on, and even if you invest a lot of time it hardly ever helps with the inspiration for a good film.

however, it does help to save a lot of money and getting the best technical results once you do sit down and actually leave the books to film something, but basically it comes down to if you like doing this kind of things, or if you absolutely need it for your kind of work and cant afford to pay anybody else.

++ christoph ++

...Yeah, I'm not looking for inspiration in technical manuals....just looking to gain some technical expertise on making technical choices. I have several projects on my pallete at the moment - some of them have already been shot... some will be shown on the web, some will be finished on Digibeta, some will be pillar boxed for HDTV, some will be letterboxed for SDTV and some will be full 4:3 for SDTV and some will be designed for presentation on LCD projectors and some will be designed for HDTV projection.... I'm not letting myself get bogged down in technical crap, but I do know that I have to make technical choices when preparing material for each venue...

It is all quite overwhelming really and that is why I'm looking for some peer reviewed publications on the subject.. Please let me know if you find any..

Thanks,

Steve