[FFmpeg-devel] [PATCH] doc/ffmpeg: add notes about timestamp handling

[Michael Niedermayer]

On Thu, Jan 23, 2014 at 05:52:29PM +0100, Stefano Sabatini wrote:
> ---
>  doc/ffmpeg.texi | 22 ++++++++++++++++++++++
>  1 file changed, 22 insertions(+)
> 
> diff --git a/doc/ffmpeg.texi b/doc/ffmpeg.texi
> index fa5f8b5..be9f206 100644
> --- a/doc/ffmpeg.texi
> +++ b/doc/ffmpeg.texi
> @@ -211,6 +211,28 @@ described.
>  
>  @c man end STREAM SELECTION
>  
> + at chapter Notes on timestamps handling
> +
> +By default, @command{ffmpeg} removes the initial offset in the input
> +timestamps, so that they will start with 0. To disable this behaviour,
> +you can apply the option @option{copyts}, which will disable the
> +application of the offset removal and other magical heuristics used to
> +fix input timestamps,

> depending on the @option{vsync} synchronization
> +mode used, the input codec, the input timestamp values, the mood of
> +the original code programmer and astrological influences.

can you post your research data upon which you base this claim of
relation to astrological influences?

also please open tickets for all cases that show unexpected behavior
(bugs tend to be fixed more readily with good bug reports)


> +
> +Note in particular that the @option{itsoffset} value has no effect if
> + at option{copyts} is not enabled, unless it is set to 0, in this case a
> +magical synchronization mode will be enabled.

the 0 check is really checking for "has the user set itsoffset"
a differnt default value could be used to detect this if users set
its offset to 0 explicitly for some good reason ?



> +
> +If you want to understand this better, read the fine source of
> + at filenane{ffmpeg.c} code, good luck.
> +

> +Finally, note that many muxers will apply further heuristics, in
> +particular they may reset the initial offset (some muxers don't
> +support initial timestamps starting from 0, or they don't accept
> +negative timestamps).

the avoid_negative_ts option should be mentioned here



> +
>  @chapter Options
>  @c man begin OPTIONS
>  
> -- 
> 1.8.1.2
> 
> _______________________________________________
> ffmpeg-devel mailing list
> ffmpeg-devel at ffmpeg.org
> http://ffmpeg.org/mailman/listinfo/ffmpeg-devel
> 

-- 
Michael     GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB

Asymptotically faster algorithms should always be preferred if you have
asymptotical amounts of data
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 198 bytes
Desc: Digital signature
URL: <http://ffmpeg.org/pipermail/ffmpeg-devel/attachments/20140123/93f46969/attachment.asc>