You are not logged in.
I need to record a TV program and I was wondering what is the best working format to easily cut commercials and the like.
The final movie will be compressed in theora/vorbis/ogm, but I would need something to work on easily: please not a lossless video codec the output file would be too large.
Suggestions?
Thanks
Offline
Offline
nice... I hoped in something more specific.
Offline
> nice... I hoped in something more specific.
Sorry, but I haven't played w/ video for years.
Formats are generally either easy to cut (uncompressed AVI) or optimized for viewing - mp4.
You seem to know the basic ideas of editing: you get the highest quality possible, then slice & dice it, then export it into a format the client prefers.
I never had problems working on rather large files and it was 8 yrs ago when 80GB drive was big and P4 1.5 GHz was fast.
Offline
I-frame only h264. Configuring x264 with keyint=1 will do it. Then pick a low ratefactor, like crf=10, and you're set.
Offline
The final movie will be compressed in theora/vorbis/ogm, but I would need something to work on easily: please not a lossless video codec the output file would be too large.
If you intend to stay with theora and vorbis, begin there to. Every change of format is going to lessen quality. You may want to consider using matroska for your container until you're done editing though. That way you can use the awesome mkvtoolnix for part of the editing toolchain.
Offline
If you intend to stay with theora and vorbis, begin there to. Every change of format is going to lessen quality.
That makes no sense. If he's going to re-encode, then the intermediate steps should lose as little quality as possible. If you encode into crappy theora and then re-encode that again with theora, you'll end up with mush. And I really mean that, because theora is terrible at keeping sharpness.
In fact, why use theora at all? If it really needs to be royalty-free at least go with vp8. Though I've seen scenes where even vp8 fails quite badly.
You may want to consider using matroska for your container until you're done editing though. That way you can use the awesome mkvtoolnix for part of the editing toolchain.
Now that I agree with. Use matroska for the end container, that is. The intermediate formats don't matter what they're in, and in fact it would be better to use AVI for them, because editors tend to have better support for it. Normally h264 shouldn't be put in avi, but in the i-frame only case it's ok.
Offline
The intermediate formats don't matter what they're in, and in fact it would be better to use AVI for them, because editors tend to have better support for it.
Soooo, almost a decade since I was dabbling in video and not much changed? ;-)
Honestly, is it a problem if you keep a couple GB of data until you're sure you're done with editing?
Offline
it is also some time I did not played with movies, so I might recall wrongly...
but I remember about 60GB of data for a 100 minutes movie. This slowed down the editing program quite a lot...
Anyway thanks, I did not know the vp8 was patent free. I'll try the mencoder lavc's ffvhuff.
Offline
I did not know the vp8 was patent free.
It's not. But it's royalty-free. Doesn't matter for the end-user really, but it is a distinction.
I'll try the mencoder lavc's ffvhuff.
ffvhuff is good. In my tests -lavcopts vcodec=ffvhuff:pred=2:context=1 produces the smallest file.
But why not use x264? This should do:
-ovc x264 -x264encopts keyint=1:crf=10You could even try x264 lossless, see if it's better than ffvhuff:
-ovc x264 -x264encopts keyint=1:qp=0x264 will very likely be slower than ffvhuff though.
Offline
skottish wrote:If you intend to stay with theora and vorbis, begin there to. Every change of format is going to lessen quality.
That makes no sense. If he's going to re-encode, then the intermediate steps should lose as little quality as possible. If you encode into crappy theora and then re-encode that again with theora, you'll end up with mush. And I really mean that, because theora is terrible at keeping sharpness.
If simple editing is what's required, then my re-enocode anything? There should only need to be the first time.
Offline
I need to record a TV program and I was wondering what is the best working format to easily cut commercials and the like.
The final movie will be compressed in theora/vorbis/ogm, but I would need something to work on easily: please not a lossless video codec the output file would be too large.
Suggestions?
Thanks
It's nearly impossible to give any suggestions unless you can give further infos.
e.g.
..how are you capturing you video ?
..what is your source ?
..is it SD or HD ?
and so on.
For SD I have been using Mencoder with good success for capturing analogue cable with on the fly deinterlace and denoise. The results look worse compared to lossless capturing and afterwards encoding but for SD XVid /DivX still does a decent job.
Even for HD (720p) XVid/DivX in higher bitrates (e.g. 4-6 MBit/s) can work nicely.
If your source material is feedable to AVISynth you might consider avsproxy plus Avidemux a valid option.
My archlinux x86_64 host:
AMD E350 (2x1.6GHz) / 8GB DDR3 RAM / GeForce 9500GT (passive) / Arch running from 16GB USB Stick
Offline
If simple editing is what's required, then my re-enocode anything? There should only need to be the first time.
So you'd encode directly into theora or vp8 and then cut that and be done with it? Well, first, you could only use one pass, which is bad for quality (and exception is the constant ratefactor mode of x264, but libtheora and libvpx don't have such a mode). Second, you could only cut at keyframes, which may or may not be where you need to cut.
But if you first encode into an edit-friendly format (ffvhuff or i-frame only h264), you can cut wherever you want and then do a two-pass encode into theora or vp8.
Last edited by Gusar (2010-07-31 23:32:15)
Offline
skottish wrote:If simple editing is what's required, then my re-enocode anything? There should only need to be the first time.
So you'd encode directly into theora or vp8 and then cut that and be done with it? Well, first, you could only use one pass, which is bad for quality (and exception is the constant ratefactor mode of x264, but libtheora and libvpx don't have such a mode). Second, you could only cut at keyframes, which may or may not be where you need to cut.
But if you first encode into an edit-friendly format (ffvhuff or i-frame only h264), you can cut wherever you want and then do a two-pass encode into theora or vp8.
These are all excellent points and I defer to your knowledge here. As well, I agree with what you said earlier about not using Theora in the first place. Generally speaking I would point anyone to x264, but it wasn't defined in the OP.
Offline