lavf/movenc: add sdtp (sample dependency) box

The sdtp is required by the AppleTV 4K in order to play 2160p60 video.

Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>

libavcodec/avcodec.h

@@ -1444,6 +1444,12 @@ typedef struct AVPacket {
  * outside the packet may be followed.
  */
 #define AV_PKT_FLAG_TRUSTED   0x0008
+/**
+ * Flag is used to indicate packets that contain frames that can
+ * be discarded by the decoder.  I.e. Non-reference frames.
+ */
+#define AV_PKT_FLAG_DISPOSABLE 0x0010
+
 
 enum AVSideDataParamChangeFlags {
     AV_SIDE_DATA_PARAM_CHANGE_CHANNEL_COUNT  = 0x0001,

libavformat/isom.h

@@ -318,6 +318,11 @@ void ff_mp4_parse_es_descr(AVIOContext *pb, int *es_id);
 #define MOV_TKHD_FLAG_IN_PREVIEW    0x0004
 #define MOV_TKHD_FLAG_IN_POSTER     0x0008
 
+#define MOV_SAMPLE_DEPENDENCY_UNKNOWN 0x0
+#define MOV_SAMPLE_DEPENDENCY_YES     0x1
+#define MOV_SAMPLE_DEPENDENCY_NO      0x2
+
+
 #define TAG_IS_AVCI(tag)                    \
     ((tag) == MKTAG('a', 'i', '5', 'p') ||  \
      (tag) == MKTAG('a', 'i', '5', 'q') ||  \

libavformat/movenc.c

@@ -252,6 +252,30 @@ static int mov_write_stss_tag(AVIOContext *pb, MOVTrack *track, uint32_t flag)
     return update_size(pb, pos);
 }
 
+/* Sample dependency atom */
+static int mov_write_sdtp_tag(AVIOContext *pb, MOVTrack *track)
+{
+    int i;
+    uint8_t leading, dependent, reference, redundancy;
+    int64_t pos = avio_tell(pb);
+    avio_wb32(pb, 0); // size
+    ffio_wfourcc(pb, "sdtp");
+    avio_wb32(pb, 0); // version & flags
+    for (i = 0; i < track->entry; i++) {
+        dependent = MOV_SAMPLE_DEPENDENCY_YES;
+        leading = reference = redundancy = MOV_SAMPLE_DEPENDENCY_UNKNOWN;
+        if (track->cluster[i].flags & MOV_DISPOSABLE_SAMPLE) {
+            reference = MOV_SAMPLE_DEPENDENCY_NO;
+        }
+        if (track->cluster[i].flags & MOV_SYNC_SAMPLE) {
+            dependent = MOV_SAMPLE_DEPENDENCY_NO;
+        }
+        avio_w8(pb, (leading << 6)   | (dependent << 4) |
+                    (reference << 2) | redundancy);
+    }
+    return update_size(pb, pos);
+}
+
 static int mov_write_amr_tag(AVIOContext *pb, MOVTrack *track)
 {
     avio_wb32(pb, 0x11); /* size */
@@ -2353,6 +2377,8 @@ static int mov_write_stbl_tag(AVFormatContext *s, AVIOContext *pb, MOVMuxContext
          track->par->codec_tag == MKTAG('r','t','p',' ')) &&
         track->has_keyframes && track->has_keyframes < track->entry)
         mov_write_stss_tag(pb, track, MOV_SYNC_SAMPLE);
+    if (track->par->codec_type == AVMEDIA_TYPE_VIDEO && track->has_disposable)
+        mov_write_sdtp_tag(pb, track);
     if (track->mode == MODE_MOV && track->flags & MOV_TRACK_STPS)
         mov_write_stss_tag(pb, track, MOV_PARTIAL_SYNC_SAMPLE);
     if (track->par->codec_type == AVMEDIA_TYPE_VIDEO &&
@@ -5316,6 +5342,10 @@ int ff_mov_write_packet(AVFormatContext *s, AVPacket *pkt)
         if (trk->cluster[trk->entry].flags & MOV_SYNC_SAMPLE)
             trk->has_keyframes++;
     }
+    if (pkt->flags & AV_PKT_FLAG_DISPOSABLE) {
+        trk->cluster[trk->entry].flags |= MOV_DISPOSABLE_SAMPLE;
+        trk->has_disposable++;
+    }
     trk->entry++;
     trk->sample_count += samples_in_chunk;
     mov->mdat_size    += size;

libavformat/movenc.h

@@ -53,6 +53,7 @@ typedef struct MOVIentry {
     int          cts;
 #define MOV_SYNC_SAMPLE         0x0001
 #define MOV_PARTIAL_SYNC_SAMPLE 0x0002
+#define MOV_DISPOSABLE_SAMPLE   0x0004
     uint32_t     flags;
 } MOVIentry;
 
@@ -89,6 +90,7 @@ typedef struct MOVTrack {
     long        sample_size;
     long        chunkCount;
     int         has_keyframes;
+    int         has_disposable;
 #define MOV_TRACK_CTTS         0x0001
 #define MOV_TRACK_STPS         0x0002
 #define MOV_TRACK_ENABLED      0x0004