ffmpeg: Maintain higher precision pts between video filter and fps code

This is a bit ugly as it attempts to keep most of the computation
in integers before the double based fps code. The use of integers
is to reduce the chances of rounding differences between platforms

Previously the timestamp was rounded to the encoder timebase
before being converted back to double precision which could cause loss
of precision

Signed-off-by: Michael Niedermayer <michaelni@gmx.at>

1 files changed, 18 insertions, 6 deletions

diff --git a/ffmpeg.c b/ffmpeg.c
index 18d2768221..909dd68600 100644
--- a/ffmpeg.c
+++ b/ffmpeg.c
@@ -883,14 +883,15 @@ static void do_subtitle_out(AVFormatContext *s,
 
 static void do_video_out(AVFormatContext *s,
                          OutputStream *ost,
-                         AVFrame *next_picture)
+                         AVFrame *next_picture,
+                         double sync_ipts)
 {
     int ret, format_video_sync;
     AVPacket pkt;
     AVCodecContext *enc = ost->enc_ctx;
     AVCodecContext *mux_enc = ost->st->codec;
     int nb_frames, nb0_frames, i;
-    double sync_ipts, delta, delta0;
+    double delta, delta0;
     double duration = 0;
     int frame_size = 0;
     InputStream *ist = NULL;
@@ -909,7 +910,6 @@ static void do_video_out(AVFormatContext *s,
         duration = lrintf(av_frame_get_pkt_duration(next_picture) * av_q2d(ist->st->time_base) / av_q2d(enc->time_base));
     }
 
-    sync_ipts = next_picture->pts;
     delta0 = sync_ipts - ost->sync_opts;
     delta  = delta0 + duration;
 
@@ -1233,6 +1233,7 @@ static int reap_filters(void)
         filtered_frame = ost->filtered_frame;
 
         while (1) {
+            double float_pts = AV_NOPTS_VALUE; // this is identical to filtered_frame.pts but with higher precission
             ret = av_buffersink_get_frame_flags(filter, filtered_frame,
                                                AV_BUFFERSINK_FLAG_NO_REQUEST);
             if (ret < 0) {
@@ -1246,9 +1247,19 @@ static int reap_filters(void)
                 av_frame_unref(filtered_frame);
                 continue;
             }
-
             if (filtered_frame->pts != AV_NOPTS_VALUE) {
                 int64_t start_time = (of->start_time == AV_NOPTS_VALUE) ? 0 : of->start_time;
+                AVRational tb = enc->time_base;
+                int extra_bits = av_clip(29 - av_log2(tb.den), 0, 16);
+
+                tb.den <<= extra_bits;
+                float_pts =
+                    av_rescale_q(filtered_frame->pts, filter->inputs[0]->time_base, tb) -
+                    av_rescale_q(start_time, AV_TIME_BASE_Q, tb);
+                float_pts /= 1 << extra_bits;
+                // avoid exact midoints to reduce the chance of rounding differences, this can be removed in case the fps code is changed to work with integers
+                float_pts += FFSIGN(float_pts) * 1.0 / (1<<17);
+
                 filtered_frame->pts =
                     av_rescale_q(filtered_frame->pts, filter->inputs[0]->time_base, enc->time_base) -
                     av_rescale_q(start_time, AV_TIME_BASE_Q, enc->time_base);
@@ -1262,12 +1273,13 @@ static int reap_filters(void)
                     enc->sample_aspect_ratio = filtered_frame->sample_aspect_ratio;
 
                 if (debug_ts) {
-                    av_log(NULL, AV_LOG_INFO, "filter -> pts:%s pts_time:%s time_base:%d/%d\n",
+                    av_log(NULL, AV_LOG_INFO, "filter -> pts:%s pts_time:%s exact:%f time_base:%d/%d\n",
                             av_ts2str(filtered_frame->pts), av_ts2timestr(filtered_frame->pts, &enc->time_base),
+                            float_pts,
                             enc->time_base.num, enc->time_base.den);
                 }
 
-                do_video_out(of->ctx, ost, filtered_frame);
+                do_video_out(of->ctx, ost, filtered_frame, float_pts);
                 break;
             case AVMEDIA_TYPE_AUDIO:
                 if (!(enc->codec->capabilities & CODEC_CAP_PARAM_CHANGE) &&