[vlc-devel] [PATCH] avcodec: added skip-frames option for encoder

Paweł Wegner pawel.wegner95 at gmail.com
Thu Jul 26 09:49:39 CEST 2018


> I don't see how this patch can't break encoding with broken streams not
> sending proper PTS or PTS at all.

This is just an option. Without enabling it explicitly the behavior is the
same as previous.

On Thu, Jul 26, 2018 at 9:41 AM Francois Cartegnie <fcvlcdev at free.fr> wrote:

> Le 26/07/2018 à 09:34, Paweł Wegner a écrit :
> > No, it is not related to the decoder output. Decoder has `skip-frames`
> > option as well, but it is not useful since it is
> > the encoder which drops frames.
> > My problem is that those timestamps are different every time I run the
> > code. Looks like they are absolute,
> > not relative to the beginning of the video. Is this intended?
> >
> > On Thu, Jul 26, 2018 at 8:58 AM Steve Lhomme <robux4 at ycbcr.xyz> wrote:
> >
> >> On 24/07/2018 21:59, Paweł Wegner wrote:
> >>> Encoder will encode only one frame per timestamp. Since timestamps may
> be
> >>> inaccurate, skipping frames may render encoder's output
> nondeterministic.
> >>> Added an option which instructs to encode all frames which come out of
> >> avcodec.
> >>
> >> Is this related to the output of the avcodec decoder ? Because that's
> >> how I understand the commit log and doc of the option. But the option is
> >> on the encoder, which may receive frames from other decoders.
>
> They are scheduled on your system, so that's system clock.
>
> I don't see how this patch can't break encoding with broken streams not
> sending proper PTS or PTS at all.
>
>
> --
> Francois Cartegnie
> VideoLAN - VLC Developer
> _______________________________________________
> vlc-devel mailing list
> To unsubscribe or modify your subscription options:
> https://mailman.videolan.org/listinfo/vlc-devel
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.videolan.org/pipermail/vlc-devel/attachments/20180726/3ca737f4/attachment.html>


More information about the vlc-devel mailing list