[MPlayer-dev-eng] [PATCH] mencoder, B-frames, and video/audio delay

Michael Niedermayer michaelni at gmx.at
Sat Jan 21 21:34:46 CET 2006


Hi

On Fri, Jan 20, 2006 at 06:53:30PM -0800, Loren Merritt wrote:
> On Fri, 20 Jan 2006, Corey Hickey wrote:
> >Loren Merritt wrote:
> >>>>BTW I think there's a corresponding problem at the _decoder_ end too,
> >>>>which you should check out. (Use h264 with deep pyramid to test it..)
> >>>
> >>>You mean b_pyramid with lots of b_frames? I'm not seeing any problem
> >>>here... I tried adding b_pyramid to each of my test clips and they
> >>>looked fine (with my wrongly-patched mencoder). The mplayer man page
> >>>says b_pyramid "increases decoding delay to 2 frames". If that's the
> >>>problem, I don't know if I would notice a 2/24 second delay. Can you
> >>>elaborate?
> >>
> >>Right, x264 doesn't support pyramid with depth > 2.
> >>But you can still modify the "delay" header to say 16, even if it only
> >>uses 2. (x264/encoder/set.c:187)
> >
> >I'm confused as to what you two mean by "depth". Do you refer to the
> >number of B-frames?
> 
> I mean decoding delay.
> 
> depth 0: no B-frames.

I->P->P->P


> depth 1: B-frame 1 depends on P-frame 0. (no B-refs)

I---------->P
I->B<-------P
I---->B<----P
I------->B<-P


> depth 2: B-frame 1 depends on B-frame 2 depends on P-frame 0.

I---------->P
I---->B<----P
I->B<-B->B<-P

yeah ok its simplified, there can be more dependancies ...

[...]

-- 
Michael




More information about the MPlayer-dev-eng mailing list