| Message ID | 9B7328D6-3555-4D18-B8F1-78EB362B8358@hotmail.com |
|---|---|
| State | Rejected |
| Headers | show |
| Series | [FFmpeg-devel] libavformat/concatdev.c: Add unified_time_base opt | expand |
| Context | Check | Description |
|---|---|---|
| andriy/configure_armv7_RPi4 | warning | Failed to apply patch |
| andriy/configure_aarch64_jetson | warning | Failed to apply patch |
Ye Chuan (12022-03-01): > In some case, the input files have different time base even though > they share the same codec and codec parameters, > While processing the packet, it will adjust the pts/dts by its own > time base instead of the unified one of the output stream, which may > lead to wrong pts/dts or unexpected speed change. > > So this patch add the "unified_time_base" opt, which could be setted by > "-unified_time_base 1" > With the option, it will concatenate the input streams whose time base > are different , and won't course speed change issue. > > Signed-off-by: Chuan Ye <yechuan0102@hotmail.com> > --- > libavformat/concatdec.c | 31 ++++++++++++++++++++++++------- > 1 file changed, 24 insertions(+), 7 deletions(-) Thanks for the patch. I need to examine exactly what it does. In the meantime, I do not know why you keep posting new versions of this patch every few minutes. Please take your time. And before next time: > int segment_time_metadata; > + int unified_time_base; Mangled indentation. Missing documentation. Spelling mistake in the title of the commit message. Regards,
There is some bug on my mail application , when it open the .eml file , the space may be erase by the application, so the patch will not pass the test online From: ffmpeg-devel <ffmpeg-devel-bounces@ffmpeg.org> on behalf of Nicolas George <george@nsup.org> Date: Tuesday, March 1, 2022 at 8:18 PM To: FFmpeg development discussions and patches <ffmpeg-devel@ffmpeg.org> Subject: Re: [FFmpeg-devel] [PATCH] libavformat/concatdev.c: Add unified_time_base opt Ye Chuan (12022-03-01): > In some case, the input files have different time base even though > they share the same codec and codec parameters, > While processing the packet, it will adjust the pts/dts by its own > time base instead of the unified one of the output stream, which may > lead to wrong pts/dts or unexpected speed change. > > So this patch add the "unified_time_base" opt, which could be setted by > "-unified_time_base 1" > With the option, it will concatenate the input streams whose time base > are different , and won't course speed change issue. > > Signed-off-by: Chuan Ye <yechuan0102@hotmail.com> > --- > libavformat/concatdec.c | 31 ++++++++++++++++++++++++------- > 1 file changed, 24 insertions(+), 7 deletions(-) Thanks for the patch. I need to examine exactly what it does. In the meantime, I do not know why you keep posting new versions of this patch every few minutes. Please take your time. And before next time: > int segment_time_metadata; > + int unified_time_base; Mangled indentation. Missing documentation. Spelling mistake in the title of the commit message. Regards, -- Nicolas George
diff --git a/libavformat/concatdec.c b/libavformat/concatdec.c index 0603c6e254..2cb656d87e 100644 --- a/libavformat/concatdec.c +++ b/libavformat/concatdec.c @@ -69,6 +69,7 @@ typedef struct { ConcatMatchMode stream_match_mode; unsigned auto_convert; int segment_time_metadata; + int unified_time_base; } ConcatContext; static int concat_probe(const AVProbeData *probe) @@ -782,13 +783,27 @@ static int concat_read_packet(AVFormatContext *avf, AVPacket *pkt) av_ts2str(pkt->pts), av_ts2timestr(pkt->pts, &st->time_base), av_ts2str(pkt->dts), av_ts2timestr(pkt->dts, &st->time_base)); - delta = av_rescale_q(cat->cur_file->start_time - cat->cur_file->file_inpoint, - AV_TIME_BASE_Q, - cat->avf->streams[pkt->stream_index]->time_base); - if (pkt->pts != AV_NOPTS_VALUE) - pkt->pts += delta; - if (pkt->dts != AV_NOPTS_VALUE) - pkt->dts += delta; + if (cat->unified_time_base) { + /* replace the pkt base on the time base of target output stream */ + AVRational output_tb; + output_tb = avf->streams[cs->out_stream_index]->time_base; + delta = av_rescale_q(cat->cur_file->start_time - cat->cur_file->file_inpoint, + AV_TIME_BASE_Q, + output_tb); + if (pkt->pts != AV_NOPTS_VALUE) + pkt->pts = av_rescale_q(pkt->pts, st->time_base, output_tb) + delta; + if (pkt->dts != AV_NOPTS_VALUE) + pkt->dts = av_rescale_q(pkt->dts, st->time_base, output_tb) + delta; + } else { + delta = av_rescale_q(cat->cur_file->start_time - cat->cur_file->file_inpoint, + AV_TIME_BASE_Q, + cat->avf->streams[pkt->stream_index]->time_base); + if (pkt->pts != AV_NOPTS_VALUE) + pkt->pts += delta; + if (pkt->dts != AV_NOPTS_VALUE) + pkt->dts += delta; + } + av_log(avf, AV_LOG_DEBUG, " -> pts:%s pts_time:%s dts:%s dts_time:%s\n", av_ts2str(pkt->pts), av_ts2timestr(pkt->pts, &st->time_base), av_ts2str(pkt->dts), av_ts2timestr(pkt->dts, &st->time_base)); @@ -931,6 +946,8 @@ static const AVOption options[] = { OFFSET(auto_convert), AV_OPT_TYPE_BOOL, {.i64 = 1}, 0, 1, DEC }, { "segment_time_metadata", "output file segment start time and duration as packet metadata", OFFSET(segment_time_metadata), AV_OPT_TYPE_BOOL, {.i64 = 0}, 0, 1, DEC }, + { "unified_time_base", "adjust to the same time base (from the first stream) while processing packets", + OFFSET(unified_time_base), AV_OPT_TYPE_BOOL, {.i64 = 0}, 0, 1, DEC }, { NULL } };
In some case, the input files have different time base even though they share the same codec and codec parameters, While processing the packet, it will adjust the pts/dts by its own time base instead of the unified one of the output stream, which may lead to wrong pts/dts or unexpected speed change. So this patch add the "unified_time_base" opt, which could be setted by "-unified_time_base 1" With the option, it will concatenate the input streams whose time base are different , and won't course speed change issue. Signed-off-by: Chuan Ye <yechuan0102@hotmail.com> --- libavformat/concatdec.c | 31 ++++++++++++++++++++++++------- 1 file changed, 24 insertions(+), 7 deletions(-)