]> git.sesse.net Git - ffmpeg/blobdiff - ffmpeg.c
ffmpeg: set pkt_timebase
[ffmpeg] / ffmpeg.c
index c779cea5c98b191c3c0766484b6f8ad464b0ac00..4afc55b40c384e59f0d762c5dda7d7324ec6b248 100644 (file)
--- a/ffmpeg.c
+++ b/ffmpeg.c
@@ -31,7 +31,9 @@
 #include <errno.h>
 #include <signal.h>
 #include <limits.h>
+#if HAVE_ISATTY
 #include <unistd.h>
+#endif
 #include "libavformat/avformat.h"
 #include "libavdevice/avdevice.h"
 #include "libswscale/swscale.h"
@@ -51,6 +53,7 @@
 #include "libavutil/imgutils.h"
 #include "libavutil/timestamp.h"
 #include "libavutil/bprint.h"
+#include "libavutil/time.h"
 #include "libavformat/os_support.h"
 
 #include "libavformat/ffm.h" // not public API
@@ -63,7 +66,6 @@
 
 #if HAVE_SYS_RESOURCE_H
 #include <sys/types.h>
-#include <sys/time.h>
 #include <sys/resource.h>
 #elif HAVE_GETPROCESSTIMES
 #include <windows.h>
 #include <conio.h>
 #endif
 
-#if HAVE_THREADS
 #if HAVE_PTHREADS
 #include <pthread.h>
-#else
-#include "libavcodec/w32pthreads.h"
-#endif
 #endif
 
 #include <time.h>
 #define VSYNC_VFR         2
 #define VSYNC_DROP        0xff
 
-#define SINKA
-
 const char program_name[] = "ffmpeg";
 const int program_birth_year = 2000;
 
@@ -167,6 +163,7 @@ static int run_as_daemon  = 0;
 static volatile int received_nb_signals = 0;
 static int64_t video_size = 0;
 static int64_t audio_size = 0;
+static int64_t subtitle_size = 0;
 static int64_t extra_size = 0;
 static int nb_frames_dup = 0;
 static int nb_frames_drop = 0;
@@ -179,7 +176,9 @@ static int print_stats = 1;
 static int debug_ts = 0;
 static int current_time;
 
-#if HAVE_THREADS
+static uint8_t *subtitle_out;
+
+#if HAVE_PTHREADS
 /* signal to input threads that they should exit; set by the main thread */
 static int transcoding_finished;
 #endif
@@ -263,13 +262,12 @@ typedef struct InputFile {
     AVFormatContext *ctx;
     int eof_reached;      /* true if eof reached */
     int ist_index;        /* index of first stream in input_streams */
-    int buffer_size;      /* current total buffer size */
     int64_t ts_offset;
     int nb_streams;       /* number of stream that ffmpeg is aware of; may be different
                              from ctx.nb_streams if new streams appear during av_read_frame() */
     int rate_emu;
 
-#if HAVE_THREADS
+#if HAVE_PTHREADS
     pthread_t thread;           /* thread reading from this file */
     int finished;               /* the thread has exited */
     int joined;                 /* the thread has been joined */
@@ -310,6 +308,7 @@ typedef struct OutputStream {
     int64_t *forced_kf_pts;
     int forced_kf_count;
     int forced_kf_index;
+    char *forced_keyframes;
 
     /* audio only */
     int audio_channels_map[SWR_CH_MAX];  /* list of the channels id to pick from the source stream */
@@ -534,8 +533,14 @@ static void reset_options(OptionsContext *o, int is_input)
 
     memset(o, 0, sizeof(*o));
 
-    if(is_input) o->recording_time = bak.recording_time;
-    else         o->recording_time = INT64_MAX;
+    if (is_input) {
+        o->recording_time = bak.recording_time;
+        if (o->recording_time != INT64_MAX)
+            av_log(NULL, AV_LOG_WARNING,
+                   "-t is not an input option, keeping it for the next output;"
+                   " consider fixing your command line.\n");
+    } else
+        o->recording_time = INT64_MAX;
     o->mux_max_delay  = 0.7;
     o->limit_filesize = UINT64_MAX;
     o->chapters_input_file = INT_MAX;
@@ -712,7 +717,7 @@ static FilterGraph *init_simple_filtergraph(InputStream *ist, OutputStream *ost)
 static void init_input_filter(FilterGraph *fg, AVFilterInOut *in)
 {
     InputStream *ist = NULL;
-    enum AVMediaType type = in->filter_ctx->input_pads[in->pad_idx].type;
+    enum AVMediaType type = avfilter_pad_get_type(in->filter_ctx->input_pads, in->pad_idx);
     int i;
 
     // TODO: support other filter types
@@ -763,6 +768,8 @@ static void init_input_filter(FilterGraph *fg, AVFilterInOut *in)
             exit_program(1);
         }
     }
+    av_assert0(ist);
+
     ist->discard         = 0;
     ist->decoding_needed = 1;
     ist->st->discard = AVDISCARD_NONE;
@@ -787,17 +794,13 @@ static int configure_output_video_filter(FilterGraph *fg, OutputFilter *ofilter,
     AVFilterContext *last_filter = out->filter_ctx;
     int pad_idx = out->pad_idx;
     int ret;
+    char name[255];
     AVBufferSinkParams *buffersink_params = av_buffersink_params_alloc();
 
-#if FF_API_OLD_VSINK_API
-    ret = avfilter_graph_create_filter(&ofilter->filter,
-                                       avfilter_get_by_name("buffersink"),
-                                       "ffmpeg_buffersink", NULL, NULL, fg->graph);
-#else
+    snprintf(name, sizeof(name), "output stream %d:%d", ost->file_index, ost->index);
     ret = avfilter_graph_create_filter(&ofilter->filter,
                                        avfilter_get_by_name("buffersink"),
-                                       "ffmpeg_buffersink", NULL, buffersink_params, fg->graph);
-#endif
+                                       name, NULL, NULL/*buffersink_params*/, fg->graph);
     av_freep(&buffersink_params);
 
     if (ret < 0)
@@ -811,8 +814,10 @@ static int configure_output_video_filter(FilterGraph *fg, OutputFilter *ofilter,
                  codec->width,
                  codec->height,
                  (unsigned)ost->sws_flags);
+        snprintf(name, sizeof(name), "scaler for output stream %d:%d",
+                 ost->file_index, ost->index);
         if ((ret = avfilter_graph_create_filter(&filter, avfilter_get_by_name("scale"),
-                                                NULL, args, NULL, fg->graph)) < 0)
+                                                name, args, NULL, fg->graph)) < 0)
             return ret;
         if ((ret = avfilter_link(last_filter, pad_idx, filter, 0)) < 0)
             return ret;
@@ -823,6 +828,8 @@ static int configure_output_video_filter(FilterGraph *fg, OutputFilter *ofilter,
 
     if ((pix_fmts = choose_pix_fmts(ost))) {
         AVFilterContext *filter;
+        snprintf(name, sizeof(name), "pixel format for output stream %d:%d",
+                 ost->file_index, ost->index);
         if ((ret = avfilter_graph_create_filter(&filter,
                                                 avfilter_get_by_name("format"),
                                                 "format", pix_fmts, NULL,
@@ -842,8 +849,10 @@ static int configure_output_video_filter(FilterGraph *fg, OutputFilter *ofilter,
 
         snprintf(args, sizeof(args), "fps=%d/%d", ost->frame_rate.num,
                  ost->frame_rate.den);
+        snprintf(name, sizeof(name), "fps for output stream %d:%d",
+                 ost->file_index, ost->index);
         ret = avfilter_graph_create_filter(&fps, avfilter_get_by_name("fps"),
-                                           "fps", args, NULL, fg->graph);
+                                           name, args, NULL, fg->graph);
         if (ret < 0)
             return ret;
 
@@ -867,11 +876,14 @@ static int configure_output_audio_filter(FilterGraph *fg, OutputFilter *ofilter,
     AVFilterContext *last_filter = out->filter_ctx;
     int pad_idx = out->pad_idx;
     char *sample_fmts, *sample_rates, *channel_layouts;
+    char name[255];
     int ret;
 
+
+    snprintf(name, sizeof(name), "output stream %d:%d", ost->file_index, ost->index);
     ret = avfilter_graph_create_filter(&ofilter->filter,
-                                       avfilter_get_by_name("abuffersink_old"),
-                                       "ffmpeg_abuffersink_old", NULL, NULL, fg->graph);
+                                       avfilter_get_by_name("abuffersink"),
+                                       name, NULL, NULL, fg->graph);
     if (ret < 0)
         return ret;
 
@@ -934,9 +946,11 @@ static int configure_output_audio_filter(FilterGraph *fg, OutputFilter *ofilter,
         av_freep(&sample_rates);
         av_freep(&channel_layouts);
 
+        snprintf(name, sizeof(name), "audio format for output stream %d:%d",
+                 ost->file_index, ost->index);
         ret = avfilter_graph_create_filter(&format,
                                            avfilter_get_by_name("aformat"),
-                                           "aformat", args, NULL, fg->graph);
+                                           name, args, NULL, fg->graph);
         if (ret < 0)
             return ret;
 
@@ -948,15 +962,6 @@ static int configure_output_audio_filter(FilterGraph *fg, OutputFilter *ofilter,
         pad_idx = 0;
     }
 
-    if (audio_sync_method > 0 && 0) {
-        char args[256] = {0};
-
-        av_strlcatf(args, sizeof(args), "min_comp=0.001:min_hard_comp=%f", audio_drift_threshold);
-        if (audio_sync_method > 1)
-            av_strlcatf(args, sizeof(args), ":max_soft_comp=%d", -audio_sync_method);
-        AUTO_INSERT_FILTER("-async", "aresample", args);
-    }
-
     if (audio_volume != 256 && 0) {
         char args[256];
 
@@ -982,7 +987,7 @@ static int configure_output_audio_filter(FilterGraph *fg, OutputFilter *ofilter,
                                                                    \
     avio_printf(pb, "%s", ctx->filter->name);                      \
     if (nb_pads > 1)                                               \
-        avio_printf(pb, ":%s", pads[inout->pad_idx].name);         \
+        avio_printf(pb, ":%s", avfilter_pad_get_name(pads, inout->pad_idx));\
     avio_w8(pb, 0);                                                \
     avio_close_dyn_buf(pb, &f->name);                              \
 }
@@ -992,7 +997,7 @@ static int configure_output_filter(FilterGraph *fg, OutputFilter *ofilter, AVFil
     av_freep(&ofilter->name);
     DESCRIBE_FILTER_LINK(ofilter, out, 0);
 
-    switch (out->filter_ctx->output_pads[out->pad_idx].type) {
+    switch (avfilter_pad_get_type(out->filter_ctx->output_pads, out->pad_idx)) {
     case AVMEDIA_TYPE_VIDEO: return configure_output_video_filter(fg, ofilter, out);
     case AVMEDIA_TYPE_AUDIO: return configure_output_audio_filter(fg, ofilter, out);
     default: av_assert0(0);
@@ -1012,6 +1017,7 @@ static int configure_input_video_filter(FilterGraph *fg, InputFilter *ifilter,
                                          ist->st->r_frame_rate;
     AVRational sar;
     AVBPrint args;
+    char name[255];
     int pad_idx = in->pad_idx;
     int ret;
 
@@ -1029,17 +1035,21 @@ static int configure_input_video_filter(FilterGraph *fg, InputFilter *ifilter,
              SWS_BILINEAR + ((ist->st->codec->flags&CODEC_FLAG_BITEXACT) ? SWS_BITEXACT:0));
     if (fr.num && fr.den)
         av_bprintf(&args, ":frame_rate=%d/%d", fr.num, fr.den);
+    snprintf(name, sizeof(name), "graph %d input from stream %d:%d", fg->index,
+             ist->file_index, ist->st->index);
 
-    if ((ret = avfilter_graph_create_filter(&ifilter->filter, filter, in->name,
+    if ((ret = avfilter_graph_create_filter(&ifilter->filter, filter, name,
                                             args.str, NULL, fg->graph)) < 0)
         return ret;
 
     if (ist->framerate.num) {
         AVFilterContext *setpts;
 
+        snprintf(name, sizeof(name), "force CFR for input from stream %d:%d",
+                 ist->file_index, ist->st->index);
         if ((ret = avfilter_graph_create_filter(&setpts,
                                                 avfilter_get_by_name("setpts"),
-                                                "setpts", "N", NULL,
+                                                name, "N", NULL,
                                                 fg->graph)) < 0)
             return ret;
 
@@ -1062,18 +1072,20 @@ static int configure_input_audio_filter(FilterGraph *fg, InputFilter *ifilter,
     AVFilter *filter = avfilter_get_by_name("abuffer");
     InputStream *ist = ifilter->ist;
     int pad_idx = in->pad_idx;
-    char args[255];
+    char args[255], name[255];
     int ret;
 
     snprintf(args, sizeof(args), "time_base=%d/%d:sample_rate=%d:sample_fmt=%s"
              ":channel_layout=0x%"PRIx64,
-             ist->st->time_base.num, ist->st->time_base.den,
+             1, ist->st->codec->sample_rate,
              ist->st->codec->sample_rate,
              av_get_sample_fmt_name(ist->st->codec->sample_fmt),
              ist->st->codec->channel_layout);
+    snprintf(name, sizeof(name), "graph %d input from stream %d:%d", fg->index,
+             ist->file_index, ist->st->index);
 
     if ((ret = avfilter_graph_create_filter(&ifilter->filter, filter,
-                                            in->name, args, NULL,
+                                            name, args, NULL,
                                             fg->graph)) < 0)
         return ret;
 
@@ -1083,9 +1095,11 @@ static int configure_input_audio_filter(FilterGraph *fg, InputFilter *ifilter,
     av_log(NULL, AV_LOG_INFO, opt_name " is forwarded to lavfi "            \
            "similarly to -af " filter_name "=%s.\n", arg);                  \
                                                                             \
+    snprintf(name, sizeof(name), "graph %d %s for input stream %d:%d",      \
+                fg->index, filter_name, ist->file_index, ist->st->index);   \
     ret = avfilter_graph_create_filter(&filt_ctx,                           \
                                        avfilter_get_by_name(filter_name),   \
-                                       filter_name, arg, NULL, fg->graph);  \
+                                       name, arg, NULL, fg->graph);         \
     if (ret < 0)                                                            \
         return ret;                                                         \
                                                                             \
@@ -1136,7 +1150,7 @@ static int configure_input_filter(FilterGraph *fg, InputFilter *ifilter,
     av_freep(&ifilter->name);
     DESCRIBE_FILTER_LINK(ifilter, in, 1);
 
-    switch (in->filter_ctx->input_pads[in->pad_idx].type) {
+    switch (avfilter_pad_get_type(in->filter_ctx->input_pads, in->pad_idx)) {
     case AVMEDIA_TYPE_VIDEO: return configure_input_video_filter(fg, ifilter, in);
     case AVMEDIA_TYPE_AUDIO: return configure_input_audio_filter(fg, ifilter, in);
     default: av_assert0(0);
@@ -1362,6 +1376,8 @@ void av_noreturn exit_program(int ret)
     }
     av_freep(&filtergraphs);
 
+    av_freep(&subtitle_out);
+
     /* close files */
     for (i = 0; i < nb_output_files; i++) {
         AVFormatContext *s = output_files[i]->ctx;
@@ -1380,6 +1396,7 @@ void av_noreturn exit_program(int ret)
         }
         output_streams[i]->bitstream_filters = NULL;
 
+        av_freep(&output_streams[i]->forced_keyframes);
         av_freep(&output_streams[i]->filtered_frame);
         av_freep(&output_streams[i]->avfilter);
         av_freep(&output_streams[i]);
@@ -1553,10 +1570,10 @@ static void do_audio_out(AVFormatContext *s, OutputStream *ost,
     av_init_packet(&pkt);
     pkt.data = NULL;
     pkt.size = 0;
-#if 0
+
     if (!check_recording_time(ost))
         return;
-#endif
+
     if (frame->pts == AV_NOPTS_VALUE || audio_sync_method < 0)
         frame->pts = ost->sync_opts;
     ost->sync_opts = frame->pts + frame->nb_samples;
@@ -1636,7 +1653,6 @@ static void do_subtitle_out(AVFormatContext *s,
                             AVSubtitle *sub,
                             int64_t pts)
 {
-    static uint8_t *subtitle_out = NULL;
     int subtitle_out_max_size = 1024 * 1024;
     int subtitle_out_size, nb, i;
     AVCodecContext *enc;
@@ -1665,6 +1681,8 @@ static void do_subtitle_out(AVFormatContext *s,
 
     for (i = 0; i < nb; i++) {
         ost->sync_opts = av_rescale_q(pts, ist->st->time_base, enc->time_base);
+        if (!check_recording_time(ost))
+            return;
 
         sub->pts = av_rescale_q(pts, ist->st->time_base, AV_TIME_BASE_Q);
         // start_display_time is required to be 0
@@ -1692,6 +1710,7 @@ static void do_subtitle_out(AVFormatContext *s,
                 pkt.pts += 90 * sub->end_display_time;
         }
         write_frame(s, &pkt, ost);
+        subtitle_size += pkt.size;
     }
 }
 
@@ -1753,6 +1772,11 @@ static void do_video_out(AVFormatContext *s,
         av_log(NULL, AV_LOG_VERBOSE, "*** drop!\n");
         return;
     } else if (nb_frames > 1) {
+        if (nb_frames > dts_error_threshold * 30) {
+            av_log(NULL, AV_LOG_ERROR, "%d frame duplication too large, skiping\n", nb_frames - 1);
+            nb_frames_drop++;
+            return;
+        }
         nb_frames_dup += nb_frames - 1;
         av_log(NULL, AV_LOG_VERBOSE, "*** %d dup!\n", nb_frames - 1);
     }
@@ -1765,6 +1789,9 @@ duplicate_frame:
 
     in_picture->pts = ost->sync_opts;
 
+    if (!check_recording_time(ost))
+        return;
+
     if (s->oformat->flags & AVFMT_RAWPICTURE &&
         enc->codec->id == CODEC_ID_RAWVIDEO) {
         /* raw pictures are written as AVPicture structure to
@@ -1921,18 +1948,9 @@ static int poll_filters(void)
                 avcodec_get_frame_defaults(ost->filtered_frame);
             filtered_frame = ost->filtered_frame;
 
-            while (!ost->is_past_recording_time) {
-                if (ost->enc->type == AVMEDIA_TYPE_AUDIO &&
-                    !(ost->enc->capabilities & CODEC_CAP_VARIABLE_FRAME_SIZE))
-                    ret = av_buffersink_read_samples(ost->filter->filter, &picref,
-                                                    ost->st->codec->frame_size);
-                else
-#ifdef SINKA
-                    ret = av_buffersink_read(ost->filter->filter, &picref);
-#else
-                    ret = av_buffersink_get_buffer_ref(ost->filter->filter, &picref,
-                                                       AV_BUFFERSINK_FLAG_NO_REQUEST);
-#endif
+            while (1) {
+                ret = av_buffersink_get_buffer_ref(ost->filter->filter, &picref,
+                                                   AV_BUFFERSINK_FLAG_NO_REQUEST);
                 if (ret < 0) {
                     if (ret != AVERROR(EAGAIN) && ret != AVERROR_EOF) {
                         char buf[256];
@@ -1962,7 +1980,7 @@ static int poll_filters(void)
 
                 switch (ost->filter->filter->inputs[0]->type) {
                 case AVMEDIA_TYPE_VIDEO:
-                    avfilter_fill_frame_from_video_buffer_ref(filtered_frame, picref);
+                    avfilter_copy_buf_props(filtered_frame, picref);
                     filtered_frame->pts = frame_pts;
                     if (!ost->frame_aspect_ratio)
                         ost->st->codec->sample_aspect_ratio = picref->video->sample_aspect_ratio;
@@ -2127,15 +2145,16 @@ static void print_report(int is_last_report, int64_t timer_start, int64_t cur_ti
     fflush(stderr);
 
     if (is_last_report) {
-        int64_t raw= audio_size + video_size + extra_size;
+        int64_t raw= audio_size + video_size + subtitle_size + extra_size;
         av_log(NULL, AV_LOG_INFO, "\n");
-        av_log(NULL, AV_LOG_INFO, "video:%1.0fkB audio:%1.0fkB global headers:%1.0fkB muxing overhead %f%%\n",
+        av_log(NULL, AV_LOG_INFO, "video:%1.0fkB audio:%1.0fkB subtitle:%1.0f global headers:%1.0fkB muxing overhead %f%%\n",
                video_size / 1024.0,
                audio_size / 1024.0,
+               subtitle_size / 1024.0,
                extra_size / 1024.0,
                100.0 * (total_size - raw) / raw
         );
-        if(video_size + audio_size + extra_size == 0){
+        if(video_size + audio_size + subtitle_size + extra_size == 0){
             av_log(NULL, AV_LOG_WARNING, "Output file is empty, nothing was encoded (check -ss / -t / -frames parameters if used)\n");
         }
     }
@@ -2228,13 +2247,6 @@ static int check_output_constraints(InputStream *ist, OutputStream *ost)
     if (of->start_time && ist->pts < of->start_time)
         return 0;
 
-    if (of->recording_time != INT64_MAX &&
-        av_compare_ts(ist->pts, AV_TIME_BASE_Q, of->recording_time + of->start_time,
-                      (AVRational){ 1, 1000000 }) >= 0) {
-        ost->is_past_recording_time = 1;
-        return 0;
-    }
-
     return 1;
 }
 
@@ -2251,12 +2263,20 @@ static void do_streamcopy(InputStream *ist, OutputStream *ost, const AVPacket *p
         !ost->copy_initial_nonkeyframes)
         return;
 
+    if (of->recording_time != INT64_MAX &&
+        ist->pts >= of->recording_time + of->start_time) {
+        ost->is_past_recording_time = 1;
+        return;
+    }
+
     /* force the input stream PTS */
     if (ost->st->codec->codec_type == AVMEDIA_TYPE_AUDIO)
         audio_size += pkt->size;
     else if (ost->st->codec->codec_type == AVMEDIA_TYPE_VIDEO) {
         video_size += pkt->size;
         ost->sync_opts++;
+    } else if (ost->st->codec->codec_type == AVMEDIA_TYPE_SUBTITLE) {
+        subtitle_size += pkt->size;
     }
 
     if (pkt->pts != AV_NOPTS_VALUE)
@@ -2304,7 +2324,7 @@ static void rate_emu_sleep(InputStream *ist)
         int64_t pts = av_rescale(ist->dts, 1000000, AV_TIME_BASE);
         int64_t now = av_gettime() - ist->start;
         if (pts > now)
-            usleep(pts - now);
+            av_usleep(pts - now);
     }
 }
 
@@ -2337,6 +2357,7 @@ static int decode_audio(InputStream *ist, AVPacket *pkt, int *got_output)
     else
         avcodec_get_frame_defaults(ist->decoded_frame);
     decoded_frame = ist->decoded_frame;
+    av_codec_set_pkt_timebase(avctx, ist->st->time_base);
 
     update_benchmark(NULL);
     ret = avcodec_decode_audio4(avctx, decoded_frame, got_output, pkt);
@@ -2414,13 +2435,27 @@ static int decode_audio(InputStream *ist, AVPacket *pkt, int *got_output)
         ist->resample_channels       = avctx->channels;
 
         for (i = 0; i < nb_filtergraphs; i++)
-            if (ist_in_filtergraph(filtergraphs[i], ist) &&
-                configure_filtergraph(filtergraphs[i]) < 0) {
-                av_log(NULL, AV_LOG_FATAL, "Error reinitializing filters!\n");
-                exit_program(1);
+            if (ist_in_filtergraph(filtergraphs[i], ist)) {
+                FilterGraph *fg = filtergraphs[i];
+                int j;
+                if (configure_filtergraph(fg) < 0) {
+                    av_log(NULL, AV_LOG_FATAL, "Error reinitializing filters!\n");
+                    exit_program(1);
+                }
+                for (j = 0; j < fg->nb_outputs; j++) {
+                    OutputStream *ost = fg->outputs[j]->ost;
+                    if (ost->enc->type == AVMEDIA_TYPE_AUDIO &&
+                        !(ost->enc->capabilities & CODEC_CAP_VARIABLE_FRAME_SIZE))
+                        av_buffersink_set_frame_size(ost->filter->filter,
+                                                     ost->st->codec->frame_size);
+                }
             }
     }
 
+    if (decoded_frame->pts != AV_NOPTS_VALUE)
+        decoded_frame->pts = av_rescale_q(decoded_frame->pts,
+                                          ist->st->time_base,
+                                          (AVRational){1, ist->st->codec->sample_rate});
     for (i = 0; i < ist->nb_filters; i++)
         av_buffersrc_add_frame(ist->filters[i]->filter, decoded_frame, 0);
 
@@ -2442,6 +2477,7 @@ static int decode_video(InputStream *ist, AVPacket *pkt, int *got_output)
         avcodec_get_frame_defaults(ist->decoded_frame);
     decoded_frame = ist->decoded_frame;
     pkt->dts  = av_rescale_q(ist->dts, AV_TIME_BASE_Q, ist->st->time_base);
+    av_codec_set_pkt_timebase(ist->st->codec, ist->st->time_base);
 
     update_benchmark(NULL);
     ret = avcodec_decode_video2(ist->st->codec,
@@ -2466,6 +2502,16 @@ static int decode_video(InputStream *ist, AVPacket *pkt, int *got_output)
     if(best_effort_timestamp != AV_NOPTS_VALUE)
         ist->next_pts = ist->pts = av_rescale_q(decoded_frame->pts = best_effort_timestamp, ist->st->time_base, AV_TIME_BASE_Q);
 
+    if (debug_ts) {
+        av_log(NULL, AV_LOG_INFO, "decoder -> ist_index:%d type:video "
+                "frame_pts:%s frame_pts_time:%s best_effort_ts:%"PRId64" best_effort_ts_time:%s keyframe:%d frame_type:%d \n",
+                ist->st->index, av_ts2str(decoded_frame->pts),
+                av_ts2timestr(decoded_frame->pts, &ist->st->time_base),
+                best_effort_timestamp,
+                av_ts2timestr(best_effort_timestamp, &ist->st->time_base),
+                decoded_frame->key_frame, decoded_frame->pict_type);
+    }
+
     pkt->size = 0;
     pre_process_video_frame(ist, (AVPicture *)decoded_frame, &buffer_to_free);
 
@@ -2758,6 +2804,37 @@ static InputStream *get_input_stream(OutputStream *ost)
     return NULL;
 }
 
+static void parse_forced_key_frames(char *kf, OutputStream *ost,
+                                    AVCodecContext *avctx)
+{
+    char *p;
+    int n = 1, i;
+    int64_t t;
+
+    for (p = kf; *p; p++)
+        if (*p == ',')
+            n++;
+    ost->forced_kf_count = n;
+    ost->forced_kf_pts   = av_malloc(sizeof(*ost->forced_kf_pts) * n);
+    if (!ost->forced_kf_pts) {
+        av_log(NULL, AV_LOG_FATAL, "Could not allocate forced key frames array.\n");
+        exit_program(1);
+    }
+
+    p = kf;
+    for (i = 0; i < n; i++) {
+        char *next = strchr(p, ',');
+
+        if (next)
+            *next++ = 0;
+
+        t = parse_time_or_die("force_key_frames", p, 1);
+        ost->forced_kf_pts[i] = av_rescale_q(t, AV_TIME_BASE_Q, avctx->time_base);
+
+        p = next;
+    }
+}
+
 static int transcode_init(void)
 {
     int ret = 0, i, j, k;
@@ -2992,6 +3069,9 @@ static int transcode_init(void)
                     codec->bits_per_raw_sample = frame_bits_per_raw_sample;
                 }
 
+                if (ost->forced_keyframes)
+                    parse_forced_key_frames(ost->forced_keyframes, ost,
+                                            ost->st->codec);
                 break;
             case AVMEDIA_TYPE_SUBTITLE:
                 codec->time_base = (AVRational){1, 1000};
@@ -3061,6 +3141,10 @@ static int transcode_init(void)
                 ret = AVERROR(EINVAL);
                 goto dump_format;
             }
+            if (ost->enc->type == AVMEDIA_TYPE_AUDIO &&
+                !(ost->enc->capabilities & CODEC_CAP_VARIABLE_FRAME_SIZE))
+                av_buffersink_set_frame_size(ost->filter->filter,
+                                             ost->st->codec->frame_size);
             assert_codec_experimental(ost->st->codec, 1);
             assert_avoptions(ost->opts);
             if (ost->st->codec->bit_rate && ost->st->codec->bit_rate < 1000)
@@ -3332,7 +3416,7 @@ static int check_keyboard_interaction(int64_t cur_time)
     return 0;
 }
 
-#if HAVE_THREADS
+#if HAVE_PTHREADS
 static void *input_thread(void *arg)
 {
     InputFile *f = arg;
@@ -3343,7 +3427,7 @@ static void *input_thread(void *arg)
         ret = av_read_frame(f->ctx, &pkt);
 
         if (ret == AVERROR(EAGAIN)) {
-            usleep(10000);
+            av_usleep(10000);
             ret = 0;
             continue;
         } else if (ret < 0)
@@ -3376,7 +3460,7 @@ static void free_input_threads(void)
         InputFile *f = input_files[i];
         AVPacket pkt;
 
-        if (f->joined)
+        if (!f->fifo || f->joined)
             continue;
 
         pthread_mutex_lock(&f->fifo_lock);
@@ -3444,7 +3528,7 @@ static int get_input_packet_mt(InputFile *f, AVPacket *pkt)
 
 static int get_input_packet(InputFile *f, AVPacket *pkt)
 {
-#if HAVE_THREADS
+#if HAVE_PTHREADS
     if (nb_input_files > 1)
         return get_input_packet_mt(f, pkt);
 #endif
@@ -3477,7 +3561,7 @@ static int transcode(void)
 
     timer_start = av_gettime();
 
-#if HAVE_THREADS
+#if HAVE_PTHREADS
     if ((ret = init_input_threads()) < 0)
         goto fail;
 #endif
@@ -3505,7 +3589,7 @@ static int transcode(void)
             if (no_packet_count) {
                 no_packet_count = 0;
                 memset(no_packet, 0, nb_input_files);
-                usleep(10000);
+                av_usleep(10000);
                 continue;
             }
             av_log(NULL, AV_LOG_VERBOSE, "No more inputs to read from, finishing.\n");
@@ -3627,7 +3711,7 @@ static int transcode(void)
         /* dump report by using the output first video and audio streams */
         print_report(0, timer_start, cur_time);
     }
-#if HAVE_THREADS
+#if HAVE_PTHREADS
     free_input_threads();
 #endif
 
@@ -3674,7 +3758,7 @@ static int transcode(void)
 
  fail:
     av_freep(&no_packet);
-#if HAVE_THREADS
+#if HAVE_PTHREADS
     free_input_threads();
 #endif
 
@@ -3990,6 +4074,7 @@ static int copy_metadata(char *outspec, char *inspec, AVFormatContext *oc, AVFor
             METADATA_CHECK_INDEX(index, context->nb_programs, "program")\
             meta = &context->programs[index]->metadata;\
             break;\
+        default: av_assert0(0);\
         }\
 
     SET_DICT(type_in, meta_in, ic, idx_in);
@@ -4094,7 +4179,7 @@ static void add_input_streams(OptionsContext *o, AVFormatContext *ic)
         ist->file_index = nb_input_files;
         ist->discard = 1;
         st->discard  = AVDISCARD_ALL;
-        ist->opts = filter_codec_opts(codec_opts, choose_decoder(o, ic, st), ic, st);
+        ist->opts = filter_codec_opts(codec_opts, ist->st->codec->codec_id, ic, st, choose_decoder(o, ic, st));
 
         ist->ts_scale = 1.0;
         MATCH_PER_STREAM_OPT(ts_scale, dbl, ist->ts_scale, ic, st);
@@ -4357,29 +4442,6 @@ static int opt_input_file(OptionsContext *o, const char *opt, const char *filena
     return 0;
 }
 
-static void parse_forced_key_frames(char *kf, OutputStream *ost)
-{
-    char *p;
-    int n = 1, i;
-
-    for (p = kf; *p; p++)
-        if (*p == ',')
-            n++;
-    ost->forced_kf_count = n;
-    ost->forced_kf_pts   = av_malloc(sizeof(*ost->forced_kf_pts) * n);
-    if (!ost->forced_kf_pts) {
-        av_log(NULL, AV_LOG_FATAL, "Could not allocate forced key frames array.\n");
-        exit_program(1);
-    }
-    p = kf;
-    for (i = 0; i < n; i++) {
-        char *next = strchr(p, ',');
-        if (next) *next++ = 0;
-        ost->forced_kf_pts[i] = parse_time_or_die("force_key_frames", p, 1);
-        p = next;
-    }
-}
-
 static uint8_t *get_line(AVIOContext *s)
 {
     AVIOContext *line;
@@ -4473,7 +4535,7 @@ static OutputStream *new_output_stream(OptionsContext *o, AVFormatContext *oc, e
     st->codec->codec_type = type;
     choose_encoder(o, oc, ost);
     if (ost->enc) {
-        ost->opts  = filter_codec_opts(codec_opts, ost->enc, oc, st);
+        ost->opts  = filter_codec_opts(codec_opts, ost->enc->id, oc, st, ost->enc);
     }
 
     avcodec_get_context_defaults3(st->codec, ost->enc);
@@ -4591,7 +4653,7 @@ static OutputStream *new_video_stream(OptionsContext *o, AVFormatContext *oc, in
 
     if (!ost->stream_copy) {
         const char *p = NULL;
-        char *forced_key_frames = NULL, *frame_size = NULL;
+        char *frame_size = NULL;
         char *frame_aspect_ratio = NULL, *frame_pix_fmt = NULL;
         char *intra_matrix = NULL, *inter_matrix = NULL;
         const char *filters = "null";
@@ -4689,9 +4751,9 @@ static OutputStream *new_video_stream(OptionsContext *o, AVFormatContext *oc, in
             }
         }
 
-        MATCH_PER_STREAM_OPT(forced_key_frames, str, forced_key_frames, oc, st);
-        if (forced_key_frames)
-            parse_forced_key_frames(forced_key_frames, ost);
+        MATCH_PER_STREAM_OPT(forced_key_frames, str, ost->forced_keyframes, oc, st);
+        if (ost->forced_keyframes)
+            ost->forced_keyframes = av_strdup(ost->forced_keyframes);
 
         MATCH_PER_STREAM_OPT(force_fps, i, ost->force_fps, oc, st);
 
@@ -4903,7 +4965,8 @@ static void init_output_filter(OutputFilter *ofilter, OptionsContext *o,
 {
     OutputStream *ost;
 
-    switch (ofilter->out_tmp->filter_ctx->output_pads[ofilter->out_tmp->pad_idx].type) {
+    switch (avfilter_pad_get_type(ofilter->out_tmp->filter_ctx->output_pads,
+                                  ofilter->out_tmp->pad_idx)) {
     case AVMEDIA_TYPE_VIDEO: ost = new_video_stream(o, oc, -1); break;
     case AVMEDIA_TYPE_AUDIO: ost = new_audio_stream(o, oc, -1); break;
     default:
@@ -4923,6 +4986,9 @@ static void init_output_filter(OutputFilter *ofilter, OptionsContext *o,
                "cannot be used together.\n", ost->file_index, ost->index);
         exit_program(1);
     }
+    if (o->recording_time != INT64_MAX)
+        av_log(NULL, AV_LOG_WARNING,
+               "-t does not work with -filter_complex (yet).\n");
 
     if (configure_output_filter(ofilter->graph, ofilter, ofilter->out_tmp) < 0) {
         av_log(NULL, AV_LOG_FATAL, "Error configuring filter.\n");
@@ -4965,7 +5031,8 @@ static void opt_output_file(void *optctx, const char *filename)
             if (!ofilter->out_tmp || ofilter->out_tmp->name)
                 continue;
 
-            switch (ofilter->out_tmp->filter_ctx->output_pads[ofilter->out_tmp->pad_idx].type) {
+            switch (avfilter_pad_get_type(ofilter->out_tmp->filter_ctx->output_pads,
+                                          ofilter->out_tmp->pad_idx)) {
             case AVMEDIA_TYPE_VIDEO:    o->video_disable    = 1; break;
             case AVMEDIA_TYPE_AUDIO:    o->audio_disable    = 1; break;
             case AVMEDIA_TYPE_SUBTITLE: o->subtitle_disable = 1; break;
@@ -5227,6 +5294,7 @@ loop_end:
                      AV_DICT_DONT_OVERWRITE);
         if(o->recording_time != INT64_MAX)
             av_dict_set(&oc->metadata, "duration", NULL, 0);
+        av_dict_set(&oc->metadata, "creation_time", NULL, 0);
     }
     if (!o->metadata_streams_manual)
         for (i = output_files[nb_output_files - 1]->ost_index; i < nb_output_streams; i++) {
@@ -5260,7 +5328,6 @@ loop_end:
                 } else if (ret < 0)
                     exit_program(1);
             }
-            printf("ret %d, stream_spec %s\n", ret, stream_spec);
         }
         else {
             switch (type) {
@@ -5354,6 +5421,7 @@ static int opt_help(const char *opt, const char *arg)
     show_help_children(avcodec_get_class(), flags);
     show_help_children(avformat_get_class(), flags);
     show_help_children(sws_get_class(), flags);
+    show_help_children(swr_get_class(), AV_OPT_FLAG_AUDIO_PARAM);
 
     return 0;
 }