[MEncoder-users] mplayer / mencoder timing mismatch (WAS: Re: mplayer / mencoder: precision of -ss)

Nitin Jain nitinj1981 at gmail.com
Thu Apr 6 13:07:22 CEST 2006


It depends upon the rate control algo that had been used to encode the
original stream (the one you _pause_ in mplayer)...
For something like h.264, where you might've dozens and scores of P &
B frames strewn between two neighbouring I frames, chances are that a
gap of 50 seconds might not be that surprizing!

Assuming the following stream properties:
duration: 1800 seconds
avg distance between I frames: 30 secs (max: 50s, min: 10s)
encode fps: 15

So,
Total frames: 27,000
Total I frames: 60
%age of I frames: 0.22

Not very sure, but I recall reading somewhere that 2-3% is considered
an okay figure, so 0.22% would be baaad, but not impossible! :D

Oh.. btw, can you tell how good (visually) does the original video
clip seem to be?

-nj

On 4/6/06, Reza Roboubi <reza at parvan.net> wrote:
> Thanks so much for all your replies.
>
> >>If someone specifies "321 seconds" he probablay expects
> >>321  seconds and not some position between 321 and 331 seconds.
> >
> >
> > Yes, well, if someone types-in -ss 300 he probably DOESN'T want to waste
> > time decoding from the last keyframe, up until EXACTLY 300 seconds.
>
> Actually, the issue is a significant lag between how mencoder and mplayer
> talk about time.  I'm trying to seek to a point which mplayer reports as
> (maybe) 1617.9 seconds into a stream.  At this distance, I need to
> put mencoder (maybe) about 50 seconds back, to get close to the desired
> location.
>
> The lag is quite significant, and cannot be explained by iframe/pframe
> arguments.
>
> A more traditional video editing tool may be used, but what I need
> is so simple, and I would be happier if I could even find some time, and
> help develop a patch for it, just so I can do it within mplayer/mencoder.
> (Of course, someone would have to advise me on how to do this.)  In any
> case, real interested to know _why_ this lag exists.
>
> Reza.
>
> _______________________________________________
> MEncoder-users mailing list
> MEncoder-users at mplayerhq.hu
> http://mplayerhq.hu/mailman/listinfo/mencoder-users
>




More information about the MEncoder-users mailing list