[FFmpeg-devel] [PATCH 1/4] lavf: add write_uncoded_frame() API.

Lukasz M lukasz.m.luki at gmail.com
Tue Jan 14 00:16:21 CET 2014


On 13 January 2014 12:07, wm4 <nfxjfg at googlemail.com> wrote:

> On Mon, 13 Jan 2014 10:23:43 +0100
> Nicolas George <george at nsup.org> wrote:
>
> > Signed-off-by: Nicolas George <george at nsup.org>
> > ---
> >  libavformat/avformat.h | 46 +++++++++++++++++++++++++++++++++++
> >  libavformat/internal.h | 14 +++++++++++
> >  libavformat/mux.c      | 65
> +++++++++++++++++++++++++++++++++++++++++++++++++-
> >  3 files changed, 124 insertions(+), 1 deletion(-)
> >
> >
> > Split the API into av_write_uncoded_frame() /
> > av_interleaved_write_uncoded_frame() / av_write_uncoded_frame_query().
> > The QUERY flag stays for internal use in the muxers callback.
> >
> > Also use a more robust magic number as the packet size.
> >
>
> Why does lavd output need to be fast? Trying to write a serious media
> player on top of lavd is a waste of time and won't ever work well with
> the current API.
>

I don't agree. How would you argument slow device is better than fast
device?
Regarding media player on top of it, I don't know what you mean by serious
player,
but I made proof of concept pulseaudio + opengl and it is using less cpu
than ffplay.
I don't see a difference between it and any other player like VLC, mplayer
or ffplay in quality nor AV sync.
I used to use BSPlayer on windows and I would say mine is better in video
quality.
It still needs 2k lines of code based on ffmpeg API, but it is first step
to make it easier and more modular.

I don't want to comment on patch because I don't feel competent in this
area, but this API is nice.


More information about the ffmpeg-devel mailing list