[Ffmpeg-devel] I'm giving up

Michael Niedermayer michaelni
Wed Dec 6 16:44:56 CET 2006


Hi

On Wed, Dec 06, 2006 at 03:03:54PM +0100, Baptiste Coudurier wrote:
> Hi
> 
> Michael Niedermayer wrote:
> > Hi
> > 
> > On Wed, Dec 06, 2006 at 01:27:58PM +0100, Baptiste Coudurier wrote:
> >> Nico Sabbi wrote:
> >>> Luca Abeni wrote:
> >>>
> >>>> BTW, talking about the code... I see that support for the 
> >>>> CODEC_FLAG_GLOBAL_HEADER is not implemented yet. How can ./ffmpeg
> >>>> -i sourcevideo -vcodec ffh264 -vtag avc1 -an targetvideo.mp4 work?
> >>>> (Or am I missing something as usual?) Anyway, if you are interested
> >>>> I can try to work on it and provide a patch for supporting
> >>>> CODEC_FLAG_GLOBAL_HEADER in one week or two.
> >>>>
> >>>>
> >>>> Thanks, Luca
> >>>>
> >>>>
> >>> why spreading that shit??  Bytestream format is the only reasonable
> >>> one. Fuck mp4!
> >> Nonsense. You don't need startcodes at all with mov/mp4, and having nal
> >> size is really easier and faster than searching for startcodes.
> > 
> > total nonsense, you MUST read the whole data once to remove startcode
> > escaping, RTFS h264.c
> > 
> > if they would have droped the excaping too then yes there would a speed
> > advantage but they didnt, so its just yet another format and you still need
> > exactly the same number of passes over the data ...
> > 
> 
> Ok, my bad for talking before actually reading code. That formatting in
> global header capable containers could be a good idea though.
> 
> Would nut support that, what could make H264 in nut faster than any
> other container ?

something like that does not belong to the container

the most sane thing IMHO would be to move the (un)escaping into a bitstream
filter or AVParser, that way h264 with and without the escaping could be
easily and quickly be handled independant of the container similar to the
mp3 header compression (and it has the same compatibility issues of course)

[...]

-- 
Michael     GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB

If you really think that XML is the answer, then you definitly missunderstood
the question -- Attila Kinali




More information about the ffmpeg-devel mailing list