qtdemux: Add basic support for MPEG-A stereoscopic video

The MPEG-A format provides an extension to the ISO base media
file format to store stereoscopic content encoded with different
codecs like H.264 and MPEG-4:2. The stereo video media information(svmi)
atom declares the presence and storage method for the video.

Stereo video information for MPEG-A can also be supplied through
the 'stvi' atom (ref: ISO/IEC_14496-12, ISO/IEC_23000-11), which
is not implemented in this patch.

Also missing is support for stereo video encoded as separate video tracks
for now.

Based on a patch by Sreerenj Balachandran <sreerenj.balachandran@intel.com>

https://bugzilla.gnome.org/show_bug.cgi?id=611157

diff --git a/gst/isomp4/fourcc.h b/gst/isomp4/fourcc.h
index c361e98..70e327d 100644 (file)
--- a/gst/isomp4/fourcc.h
+++ b/gst/isomp4/fourcc.h
@@ -319,6 +319,12 @@ G_BEGIN_DECLS
 #define MS_WAVE_FOURCC(codecid)  GST_MAKE_FOURCC( \
         'm', 's', ((codecid)>>8)&0xff, ((codecid)&0xff))
 
+/* MPEG Application Format , Stereo Video */
+#define FOURCC_ss01     GST_MAKE_FOURCC('s','s','0','1')
+#define FOURCC_ss02     GST_MAKE_FOURCC('s','s','0','2')
+#define FOURCC_svmi     GST_MAKE_FOURCC('s','v','m','i')
+#define FOURCC_scdi     GST_MAKE_FOURCC('s','c','d','i')
+
 G_END_DECLS
 
 #endif /* __FOURCC_H__ */

diff --git a/gst/isomp4/qtdemux.c b/gst/isomp4/qtdemux.c
index 07b8b84..d4f129f 100644 (file)
--- a/gst/isomp4/qtdemux.c
+++ b/gst/isomp4/qtdemux.c
@@ -383,6 +383,10 @@ struct _QtDemuxStream
   guint32 def_sample_flags;
 
   gboolean disabled;
+
+  /* stereoscopic video streams */
+  GstVideoMultiviewMode multiview_mode;
+  GstVideoMultiviewFlags multiview_flags;
 };
 
 enum QtDemuxState
@@ -1744,6 +1748,8 @@ _create_stream (void)
   stream->sample_index = -1;
   stream->offset_in_sample = 0;
   stream->new_stream = TRUE;
+  stream->multiview_mode = GST_VIDEO_MULTIVIEW_MODE_NONE;
+  stream->multiview_flags = GST_VIDEO_MULTIVIEW_FLAGS_NONE;
   return stream;
 }
 
@@ -7694,6 +7700,7 @@ qtdemux_parse_trak (GstQTDemux * qtdemux, GNode * trak)
   GNode *pasp;
   GNode *tref;
   GNode *udta;
+  GNode *svmi;
 
   QtDemuxStream *stream = NULL;
   gboolean new_stream = FALSE;
@@ -7849,6 +7856,51 @@ qtdemux_parse_trak (GstQTDemux * qtdemux, GNode * trak)
   if (!(stbl = qtdemux_tree_get_child_by_type (minf, FOURCC_stbl)))
     goto corrupt_file;
 
+  /*parse svmi header if existing */
+  svmi = qtdemux_tree_get_child_by_type (stbl, FOURCC_svmi);
+  if (svmi) {
+    len = QT_UINT32 ((guint8 *) svmi->data);
+    version = QT_UINT32 ((guint8 *) svmi->data + 8);
+    if (!version) {
+      GstVideoMultiviewMode mode = GST_VIDEO_MULTIVIEW_MODE_NONE;
+      GstVideoMultiviewFlags flags = GST_VIDEO_MULTIVIEW_FLAGS_NONE;
+      guint8 frame_type, frame_layout;
+
+      /* MPEG-A stereo video */
+      if (qtdemux->major_brand == FOURCC_ss02)
+        flags |= GST_VIDEO_MULTIVIEW_FLAGS_MIXED_MONO;
+
+      frame_type = QT_UINT8 ((guint8 *) svmi->data + 12);
+      frame_layout = QT_UINT8 ((guint8 *) svmi->data + 13) & 0x01;
+      switch (frame_type) {
+        case 0:
+          mode = GST_VIDEO_MULTIVIEW_MODE_SIDE_BY_SIDE;
+          break;
+        case 1:
+          mode = GST_VIDEO_MULTIVIEW_MODE_ROW_INTERLEAVED;
+          break;
+        case 2:
+          mode = GST_VIDEO_MULTIVIEW_MODE_FRAME_BY_FRAME;
+          break;
+        case 3:
+          /* mode 3 is primary/secondary view sequence, ie
+           * left/right views in separate tracks. See section 7.2
+           * of ISO/IEC 23000-11:2009 */
+          GST_FIXME_OBJECT (qtdemux,
+              "Implement stereo video in separate streams");
+      }
+
+      if ((frame_layout & 0x1) == 0)
+        flags |= GST_VIDEO_MULTIVIEW_FLAGS_RIGHT_VIEW_FIRST;
+
+      GST_LOG_OBJECT (qtdemux,
+          "StereoVideo: composition type: %u, is_left_first: %u",
+          frame_type, frame_layout);
+      stream->multiview_mode = mode;
+      stream->multiview_flags = flags;
+    }
+  }
+
   /* parse stsd */
   if (!(stsd = qtdemux_tree_get_child_by_type (stbl, FOURCC_stsd)))
     goto corrupt_file;
@@ -11566,11 +11618,20 @@ qtdemux_video_caps (GstQTDemux * qtdemux, QtDemuxStream * stream,
 
     gst_video_info_init (&info);
     gst_video_info_set_format (&info, format, stream->width, stream->height);
+    GST_VIDEO_INFO_MULTIVIEW_MODE (&info) = stream->multiview_mode;
+    GST_VIDEO_INFO_MULTIVIEW_FLAGS (&info) = stream->multiview_flags;
+
     caps = gst_video_info_to_caps (&info);
     *codec_name = gst_pb_utils_get_codec_description (caps);
 
     /* enable clipping for raw video streams */
     stream->need_clip = TRUE;
+  } else if (stream->multiview_mode != GST_VIDEO_MULTIVIEW_MODE_NONE) {
+    gst_caps_set_simple (caps,
+        "multiview-mode", G_TYPE_STRING,
+        gst_video_multiview_mode_to_caps_string (stream->multiview_mode),
+        "multiview-flags", GST_TYPE_VIDEO_MULTIVIEW_FLAGSET,
+        stream->multiview_flags, GST_FLAG_SET_MASK_EXACT, NULL);
   }
 
   return caps;

diff --git a/gst/isomp4/qtdemux_dump.c b/gst/isomp4/qtdemux_dump.c
index a73812e..ea99996 100644 (file)
--- a/gst/isomp4/qtdemux_dump.c
+++ b/gst/isomp4/qtdemux_dump.c
@@ -837,6 +837,36 @@ qtdemux_dump_sdtp (GstQTDemux * qtdemux, GstByteReader * data, int depth)
 }
 
 gboolean
+qtdemux_dump_svmi (GstQTDemux * qtdemux, GstByteReader * data, int depth)
+{
+  guint32 version;
+  guint stereo_mono_change_count;
+  guint i;
+
+  version = GET_UINT32 (data);
+  GST_LOG ("%*s  version/flags: %08x", depth, "", version);
+
+  if (!version) {
+    /* stereoscopic visual type information */
+    GST_LOG ("%*s     stereo_composition_type: %d", depth, "",
+        GET_UINT8 (data));
+    GST_LOG ("%*s     is_left_first: %d", depth, "",
+        ((guint8) GET_UINT8 (data)) & 0x01);
+
+    /* stereo_mono_change information */
+    stereo_mono_change_count = GET_UINT32 (data);
+    GST_LOG ("%*s     stereo_mono_change_count: %d", depth, "",
+        stereo_mono_change_count);
+    for (i = 1; i <= stereo_mono_change_count; i++) {
+      GST_LOG ("%*s     sample_count: %d", depth, "", GET_UINT32 (data));
+      GST_LOG ("%*s     stereo_flag: %d", depth, "",
+          ((guint8) GET_UINT8 (data)) & 0x01);
+    }
+  }
+  return TRUE;
+}
+
+gboolean
 qtdemux_dump_unknown (GstQTDemux * qtdemux, GstByteReader * data, int depth)
 {
   int len;

diff --git a/gst/isomp4/qtdemux_dump.h b/gst/isomp4/qtdemux_dump.h
index 6b12874..162b411 100644 (file)
--- a/gst/isomp4/qtdemux_dump.h
+++ b/gst/isomp4/qtdemux_dump.h
@@ -81,6 +81,8 @@ gboolean qtdemux_dump_tfdt (GstQTDemux * qtdemux, GstByteReader * data,
     int depth);
 gboolean qtdemux_dump_unknown (GstQTDemux * qtdemux, GstByteReader * data,
     int depth);
+gboolean qtdemux_dump_svmi (GstQTDemux *qtdemux, GstByteReader *data,
+    int depth);
 
 gboolean qtdemux_node_dump (GstQTDemux * qtdemux, GNode * node);
 

diff --git a/gst/isomp4/qtdemux_types.c b/gst/isomp4/qtdemux_types.c
index 651a85f..6672efe 100644 (file)
--- a/gst/isomp4/qtdemux_types.c
+++ b/gst/isomp4/qtdemux_types.c
@@ -184,6 +184,10 @@ static const QtNodeType qt_node_types[] = {
   {FOURCC_frma, "Audio codec format", 0},
   {FOURCC_name, "name", 0},
   {FOURCC_mean, "mean", 0},
+  {FOURCC_svmi, "Stereoscopic Video Media Information", 0,
+      qtdemux_dump_svmi},
+  {FOURCC_scdi, "Stereoscopic Camera and Display Information", 0,
+      qtdemux_dump_unknown},
   {0, "unknown", 0,},
 };