[FFmpeg-devel] [PATCH] avformat/dashenc: use AVCodecContext timebase when computing missing bitrate

Przemysław Sobala przemyslaw.sobala at gmail.com
Thu Jun 4 10:59:07 EEST 2020


On Tue, Jun 2, 2020 at 10:19 AM Przemysław Sobala <
przemyslaw.sobala at gmail.com> wrote:

> On Mon, Jun 1, 2020 at 3:30 PM Jeyapal, Karthick <kjeyapal at akamai.com>
> wrote:
>
>>
>> On 6/1/20 5:24 PM, Przemysław Sobala wrote:
>> > On Mon, Jun 1, 2020 at 10:06 AM Anton Khirnov <anton at khirnov.net>
>> wrote:
>> >
>> >> Quoting Przemysław Sobala (2020-05-27 17:07:22)
>> >>> ---
>> >>>  libavformat/dashenc.c | 2 +-
>> >>>  1 file changed, 1 insertion(+), 1 deletion(-)
>> >>>
>> >>> diff --git a/libavformat/dashenc.c b/libavformat/dashenc.c
>> >>> index 0cf0df50ef..00a37b175d 100644
>> >>> --- a/libavformat/dashenc.c
>> >>> +++ b/libavformat/dashenc.c
>> >>> @@ -1959,7 +1959,7 @@ static int dash_flush(AVFormatContext *s, int
>> >> final, int stream)
>> >>>
>> >>>          if (!os->bit_rate) {
>> >>>              // calculate average bitrate of first segment
>> >>> -            int64_t bitrate = (int64_t) range_length * 8 *
>> AV_TIME_BASE
>> >> / duration;
>> >>> +            int64_t bitrate = (int64_t) range_length * 8 *
>> >> (c->use_timeline ? os->ctx->streams[0]->time_base.den : AV_TIME_BASE) /
>> >> duration;
>> >>
>> >> That does not look like an AVCodecContext
>> >>
>> >
>> > Of course not. time_base is AVStream's field. I don't know why I wrote
>> > AVCodecContext... Please amend that commit message if possible.
>> Amended and Pushed!
>>
>> Thanks,
>> Karthick
>>
>>
> Thanks.
> What do you think about computing an average bitrate for all segments, not
> only the first one (in case of a static - not dynamic - DASH manifest), if
> one would not want to specify bitrate while encoding using x264 CRF rate
> control? I could prepare such a patch that, if bitrate is not specified,
> it'd be computed at the end, for static manifest, for all segments. It'd be
> more accurate comparing to the first segment's bitrate.
>

Any comments about that?

--
pozdrawiam
Przemysław Sobala


More information about the ffmpeg-devel mailing list