[MPlayer-DOCS] Undocuments -lavcopts and -lavfopts
Michael Niedermayer
michaelni at gmx.at
Sat Feb 25 02:20:08 CET 2006
Hi
On Fri, Feb 24, 2006 at 04:05:06PM +0200, Oded Shimon wrote:
> Was going over old cvslog, saw Michael added some new options quite a while
> ago:
ok, ill try to explain what they do, assuming i remember it correctly :)
>
> -lavcopts vglobal
> from what i understand, puts manditory headers at begginning of file
> instead of at every keyframe (like mpeg-4 quant tables), ask Michael for
> more details...
for encoding:
0 -> default (with ffmpeg.c that will do the sane thing of producing global
headers for mp4/mov/nut/... but local oherwise) i think this doesnt
work for mencoder ATM
1 -> force global headers instead of local ones
2 -> force local headers instead of global ones
for remuxing (ffmpeg.c only currently but lavf demuxer should be changed to
honor this too ...)
0 -> default
1 -> extract global headers from the stream (convert local->global)
2 -> write global headers at keyframes (convert global->local)
>
> -lavfopts muxrate
some containers like mpeg-ps have a (max)bitrate at which muxing is done
this overrides lavfs decision (if you want less problems with WMP and your
VBR MPEG then you can set this manually to the average or so bitrate
the resulting file is not valid MPEG though ... yeah MS never cares
about following the specs)
> -lavfopts packetsize
some containers like mpeg-ps have a packets size independant of the video/audio
frame size, guess what this option does ... (needed for vcd,dvd,...)
> -lavfopts preload
> -lavfopts delay
these control the demux - decode delay (SCR - DTS)
when you have a large number of small frames then they can be stored in the
file so that they are read long before their DTS has come in mpeg-ps/ts
setting these 2 to large values gives the muxer more freedom, setting them
to small values restricts the muxer which can lead to buffer over/underflows
but nicer seeking (large limit means also that the DTS of 2 streams can be
far appart at one spot in the file)
[...]
--
Michael
More information about the MPlayer-DOCS
mailing list