[FFmpeg-devel] [PATCH] fftools/ffmpeg.c: allow forcing input framerate on streamcopy

[Carl Eugen Hoyos]

2018-10-19 20:39 GMT+02:00, Leo Izen <leo.izen at gmail.com>:
> On 10/19/18 2:26 PM, Carl Eugen Hoyos wrote:
>> 2018-10-19 4:58 GMT+02:00, Leo Izen <leo.izen at gmail.com>:
>>> ---
>>>   fftools/ffmpeg.c | 8 +++++---
>>>   1 file changed, 5 insertions(+), 3 deletions(-)
>>>
>>> diff --git a/fftools/ffmpeg.c b/fftools/ffmpeg.c
>>> index da4259a9a8..5d68194676 100644
>>> --- a/fftools/ffmpeg.c
>>> +++ b/fftools/ffmpeg.c
>>> @@ -2045,12 +2045,14 @@ static void do_streamcopy(InputStream *ist,
>>> OutputStream *ost, const AVPacket *p
>>>       if (ost->enc_ctx->codec_type == AVMEDIA_TYPE_VIDEO)
>>>           ost->sync_opts++;
>>>
>>> -    if (pkt->pts != AV_NOPTS_VALUE)
>>> +    if (ist->framerate.num)
>>> +        opkt.pts = av_rescale_q(ist->pts, AV_TIME_BASE_Q,
>>> ost->mux_timebase) - ost_tb_start_time;
>>> +    else if (pkt->pts != AV_NOPTS_VALUE)
>>>           opkt.pts = av_rescale_q(pkt->pts, ist->st->time_base,
>>> ost->mux_timebase) - ost_tb_start_time;
>>>       else
>>>           opkt.pts = AV_NOPTS_VALUE;
>>>
>>> -    if (pkt->dts == AV_NOPTS_VALUE)
>>> +    if (pkt->dts == AV_NOPTS_VALUE || ist->framerate.num)
>>>           opkt.dts = av_rescale_q(ist->dts, AV_TIME_BASE_Q,
>>> ost->mux_timebase);
>>>       else
>>>           opkt.dts = av_rescale_q(pkt->dts, ist->st->time_base,
>>> ost->mux_timebase);
>>> @@ -2602,7 +2604,7 @@ static int process_input_packet(InputStream *ist,
>>> const AVPacket *pkt, int no_eo
>>>           avpkt = *pkt;
>>>       }
>>>
>>> -    if (pkt && pkt->dts != AV_NOPTS_VALUE) {
>>> +    if (pkt && pkt->dts != AV_NOPTS_VALUE && !ist->framerate.num) {
>>>           ist->next_dts = ist->dts = av_rescale_q(pkt->dts,
>>> ist->st->time_base, AV_TIME_BASE_Q);
>>>           if (ist->dec_ctx->codec_type != AVMEDIA_TYPE_VIDEO ||
>>> !ist->decoding_needed)
>>>               ist->next_pts = ist->pts = ist->dts;
>> How can this be tested?
>>
>> Carl Eugen
>
> I'm not entirely sure. I ran "make fate" and it passed, and I
> successfully rescaled a 30fps clip to 15fps, 20fps, 45fps, and 60fps
> using -r:v as an input option. I'm not entirely sure what the standard
> procedure is for performing more rigorous tests.

What I meant was:
Which kind of input and output did you use to test your patch?

Carl Eugen