http://crazedmuleproductions.blogspot.com/2009/01/stock-footage-encoding-h264-and-ipod.html
Wow, great stuff, Scott! Exactly what I'm interested in. That's even the
same camera I've been looking into, and the music's great too! =)
I've been waiting for a camera like the 5DmkII for a very long time, to
replace my old 20D. Not that there's anything wrong with it at all, but the
thought of full frame photos (welcome home, my dear 17mm lens!) and shooting
*video* with my 70-200mm EF IS L is very appealing! In the mean time I've
bough a handly little Sanyo Xacti HD700 that records okay 720p under the
right light circumstances.
The cam is just awesome. Yes, some drawbacks for the videographer, but
that output is just gorgeous. Gotta have the right lenses to make the
most out of it. And a tripod, for God's sake! :)
Since my post, I've come up with a solution that in the end also renders
great 1080p video: First, I initiate two renderings from within Cinelerra:
The first renders a series of PNG stills, the second renders a MP3 file (I
should output WAV instead).
Is this your flow for producting the PNGs:
cam image (RAW or JPG) -> convert to PNG -> import into Cinelerra?
You mentioned an issue when the audio was longer than the video. By
rendering both the video (PNG stills) and audio (WAV) separately from
Cinelerra, you get two bonuses: 1) The audio always matches the video, 2)
you can edit audio as well; for instance I noticed, that your video nor the
audio didn't fade out ;)
That is a good solution. I will add that to the my testing grid! :) Oh
yeah, the vid you saw in the stock footage post was just an index of
footage I had taken that night and not edited properly. Here is a nicely
produced example of the 5D's output:
http://crazedmuleproductions.blogspot.com/2009/01/water-new-canon-5d-video.html
On the other hand, I get the feeling that rendering to PNG stills takes 100
times longer: A little over 2 hours for 33 seconds video, consuming 11GB in
820 images, and finally the 41MB final video.
As comparison, rendering the water video from MPEG-TS source took
20minutes on my dual quad core. I spent a ridiculous amount of time
comparing formats that are editable in Cinelerra. I've settled on
MPEG-TS, as it looks great and edits well in Cin.
My script uses a combination of mencoder and MP4Box to produce
mpeg4(h264+acc). I've been shifting between ffmpeg and mencoder, and don't
really know which to prefer. My experience is, that sometimes one works
while the other fails and vice versa.
I have the same issue.
Perhaps a combination of the out two approaches would be perfect: Three
rendering profiles in Cinelerra called something like:
(1 of 3) Render Audio
(2 of 3) Render Video pass 1
(3 of 3) Render Video pass 2
Where, of course, the rendered audio is used as input. I think I'll try it,
just for the fun of it. And maybe compare rendering times with my previous
PNG-stills-approach, and maybe even compare ffmpeg against mencoder.
Love to see the results from that! Let us know what you find.
Thanks for the inspiration ;)
No prob. Welcome to the giant time sink that is Linux video editing!
scott
_______________________________________________
Cinelerra mailing list
[email protected]
https://init.linpro.no/mailman/skolelinux.no/listinfo/cinelerra
