fftools/ffmpeg_mux: stop rescaling timestamps in of_streamcopy()

This function converts packet timestamps from the input stream timebase
to OutputStream.mux_timebase, which may or may not be equal to the
actual output AVStream timebase (and even when it is, this may not
always be the optimal choice due to bitstream filtering).

Just keep the timestamps in input stream timebase, they will be rescaled
as needed before bitstream filtering and/or sending the packet to the
muxer.

Move the av_rescale_delta() call for audio (needed to preserve accuracy
with coarse demuxer timebases) to write_packet.

Drop now-unused OutputStream.mux_timebase.

fftools/ffmpeg.h

@@ -523,8 +523,6 @@ typedef struct OutputStream {
     /* dts of the last packet sent to the muxing queue, in AV_TIME_BASE_Q */
     int64_t last_mux_dts;
 
-    // the timebase of the packets sent to the muxer
-    AVRational mux_timebase;
     AVRational enc_timebase;
 
     Encoder *enc;

fftools/ffmpeg_enc.c

@@ -463,8 +463,6 @@ int enc_open(OutputStream *ost, AVFrame *frame)
     if (ost->st->time_base.num <= 0 || ost->st->time_base.den <= 0)
         ost->st->time_base = av_add_q(ost->enc_ctx->time_base, (AVRational){0, 1});
 
-    ost->mux_timebase = enc_ctx->time_base;
-
     ret = of_stream_init(of, ost);
     if (ret < 0)
         return ret;

fftools/ffmpeg_mux.c

@@ -76,6 +76,22 @@ static int write_packet(Muxer *mux, OutputStream *ost, AVPacket *pkt)
     if (ost->type == AVMEDIA_TYPE_VIDEO && ost->vsync_method == VSYNC_DROP)
         pkt->pts = pkt->dts = AV_NOPTS_VALUE;
 
+    // rescale timestamps to the stream timebase
+    if (ost->type == AVMEDIA_TYPE_AUDIO && !ost->enc) {
+        // use av_rescale_delta() for streamcopying audio, to preserve
+        // accuracy with coarse input timebases
+        int duration = av_get_audio_frame_duration2(ost->st->codecpar, pkt->size);
+
+        if (!duration)
+            duration = ost->st->codecpar->frame_size;
+
+        pkt->dts = av_rescale_delta(pkt->time_base, pkt->dts,
+                                    (AVRational){1, ost->st->codecpar->sample_rate}, duration,
+                                    &ms->ts_rescale_delta_last, ost->st->time_base);
+        pkt->pts = pkt->dts;
+
+        pkt->duration = av_rescale_q(pkt->duration, pkt->time_base, ost->st->time_base);
+    } else
         av_packet_rescale_ts(pkt, pkt->time_base, ost->st->time_base);
     pkt->time_base = ost->st->time_base;
 
@@ -392,7 +408,7 @@ int of_streamcopy(OutputStream *ost, const AVPacket *pkt, int64_t dts)
     OutputFile *of = output_files[ost->file_index];
     MuxStream  *ms = ms_from_ost(ost);
     int64_t start_time = (of->start_time == AV_NOPTS_VALUE) ? 0 : of->start_time;
-    int64_t ost_tb_start_time = av_rescale_q(start_time, AV_TIME_BASE_Q, ost->mux_timebase);
+    int64_t ts_offset;
     AVPacket *opkt = ms->pkt;
     int ret;
 
@@ -425,27 +441,17 @@ int of_streamcopy(OutputStream *ost, const AVPacket *pkt, int64_t dts)
     if (ret < 0)
         return ret;
 
-    opkt->time_base = ost->mux_timebase;
+    ts_offset = av_rescale_q(start_time, AV_TIME_BASE_Q, opkt->time_base);
 
     if (pkt->pts != AV_NOPTS_VALUE)
-        opkt->pts = av_rescale_q(pkt->pts, pkt->time_base, opkt->time_base) - ost_tb_start_time;
+        opkt->pts -= ts_offset;
 
     if (pkt->dts == AV_NOPTS_VALUE) {
         opkt->dts = av_rescale_q(dts, AV_TIME_BASE_Q, opkt->time_base);
     } else if (ost->st->codecpar->codec_type == AVMEDIA_TYPE_AUDIO) {
-        int duration = av_get_audio_frame_duration2(ost->par_in, pkt->size);
-        if(!duration)
-            duration = ost->par_in->frame_size;
-        opkt->dts = av_rescale_delta(pkt->time_base, pkt->dts,
-                                    (AVRational){1, ost->par_in->sample_rate}, duration,
-                                    &ms->ts_rescale_delta_last, opkt->time_base);
-        /* dts will be set immediately afterwards to what pts is now */
-        opkt->pts = opkt->dts - ost_tb_start_time;
-    } else
-        opkt->dts = av_rescale_q(pkt->dts, pkt->time_base, opkt->time_base);
-    opkt->dts -= ost_tb_start_time;
-
-    opkt->duration = av_rescale_q(pkt->duration, pkt->time_base, opkt->time_base);
+        opkt->pts = opkt->dts - ts_offset;
+    }
+    opkt->dts -= ts_offset;
 
     {
         int ret = trigger_fix_sub_duration_heartbeat(ost, pkt);
@@ -511,10 +517,6 @@ static int thread_start(Muxer *mux)
         MuxStream     *ms = ms_from_ost(ost);
         AVPacket *pkt;
 
-        /* try to improve muxing time_base (only possible if nothing has been written yet) */
-        if (!av_fifo_can_read(ms->muxing_queue))
-            ost->mux_timebase = ost->st->time_base;
-
         while (av_fifo_read(ms->muxing_queue, &pkt, 1) >= 0) {
             ret = thread_submit_packet(mux, ost, pkt);
             if (pkt) {

fftools/ffmpeg_mux.h

@@ -75,7 +75,7 @@ typedef struct MuxStream {
     int64_t    stream_duration;
     AVRational stream_duration_tb;
 
-    // audio streamcopy - state for av_rescale_delta()
+    // state for av_rescale_delta() call for audio in write_packet()
     int64_t ts_rescale_delta_last;
 
     // combined size of all the packets sent to the muxer

fftools/ffmpeg_mux_init.c

@@ -1090,8 +1090,6 @@ static int streamcopy_init(const Muxer *mux, OutputStream *ost)
         }
     }
 
-    ost->mux_timebase = ist->st->time_base;
-
 fail:
     avcodec_free_context(&codec_ctx);
     av_dict_free(&codec_opts);