vp{8,9}enc: Tell the encoder about actual timestamps and durations of frames

... instead of just counting frames. The values are supposed to be in timebase
units, not frame units. This fixes various quality problems with VP8/VP9
encoding and in general makes the encoder behave better.

Thanks to Nirbheek Chauhan for noticing this bug.

ext/vpx/gstvp8enc.c

@@ -805,7 +805,6 @@ gst_vp8_enc_finalize (GObject * object)
   g_mutex_clear (&gst_vp8_enc->encoder_lock);
 
   G_OBJECT_CLASS (parent_class)->finalize (object);
-
 }
 
 static void
@@ -1877,15 +1876,19 @@ gst_vp8_enc_drain (GstVideoEncoder * video_encoder)
   int flags = 0;
   vpx_codec_err_t status;
   gint64 deadline;
+  vpx_codec_pts_t pts;
 
   encoder = GST_VP8_ENC (video_encoder);
 
   g_mutex_lock (&encoder->encoder_lock);
   deadline = encoder->deadline;
 
-  status =
+  pts =
-      vpx_codec_encode (&encoder->encoder, NULL, encoder->n_frames, 1, flags,
+      gst_util_uint64_scale (encoder->last_pts,
-      deadline);
+      encoder->cfg.g_timebase.num * (GstClockTime) GST_SECOND,
+      encoder->cfg.g_timebase.den);
+
+  status = vpx_codec_encode (&encoder->encoder, NULL, pts, 1, flags, deadline);
   g_mutex_unlock (&encoder->encoder_lock);
 
   if (status != 0) {
@@ -1917,11 +1920,18 @@ gst_vp8_enc_drain (GstVideoEncoder * video_encoder)
 static GstFlowReturn
 gst_vp8_enc_finish (GstVideoEncoder * video_encoder)
 {
+  GstVP8Enc *encoder;
   GstFlowReturn ret;
 
   GST_DEBUG_OBJECT (video_encoder, "finish");
 
+  encoder = GST_VP8_ENC (video_encoder);
+
+  if (encoder->inited) {
     ret = gst_vp8_enc_drain (video_encoder);
+  } else {
+    ret = GST_FLOW_OK;
+  }
 
   return ret;
 }
@@ -1954,13 +1964,13 @@ gst_vp8_enc_handle_frame (GstVideoEncoder * video_encoder,
   vpx_image_t *image;
   GstVP8EncUserData *user_data;
   GstVideoFrame vframe;
+  vpx_codec_pts_t pts;
+  unsigned long duration;
 
   GST_DEBUG_OBJECT (video_encoder, "handle_frame");
 
   encoder = GST_VP8_ENC (video_encoder);
 
-  encoder->n_frames++;
-
   GST_DEBUG_OBJECT (video_encoder, "size %d %d",
       GST_VIDEO_INFO_WIDTH (&encoder->input_state->info),
       GST_VIDEO_INFO_HEIGHT (&encoder->input_state->info));
@@ -1979,8 +1989,25 @@ gst_vp8_enc_handle_frame (GstVideoEncoder * video_encoder,
   }
 
   g_mutex_lock (&encoder->encoder_lock);
+  pts =
+      gst_util_uint64_scale (frame->pts,
+      encoder->cfg.g_timebase.num * (GstClockTime) GST_SECOND,
+      encoder->cfg.g_timebase.den);
+  encoder->last_pts = frame->pts;
+
+  if (frame->duration != GST_CLOCK_TIME_NONE) {
+    duration =
+        gst_util_uint64_scale (frame->duration,
+        encoder->cfg.g_timebase.num * (GstClockTime) GST_SECOND,
+        encoder->cfg.g_timebase.den);
+    encoder->last_pts += frame->duration;
+  } else {
+    duration = 1;
+  }
+
   status = vpx_codec_encode (&encoder->encoder, image,
-      encoder->n_frames, 1, flags, encoder->deadline);
+      pts, duration, flags, encoder->deadline);
+
   g_mutex_unlock (&encoder->encoder_lock);
   gst_video_frame_unmap (&vframe);
 

ext/vpx/gstvp8enc.h

@@ -102,7 +102,7 @@ struct _GstVP8Enc
 
   vpx_image_t image;
 
-  int n_frames;
+  GstClockTime last_pts;
   int keyframe_distance;
 
   GstVideoCodecState *input_state;

ext/vpx/gstvp9enc.c

@@ -780,7 +780,6 @@ gst_vp9_enc_finalize (GObject * object)
   g_mutex_clear (&gst_vp9_enc->encoder_lock);
 
   G_OBJECT_CLASS (parent_class)->finalize (object);
-
 }
 
 static void
@@ -1816,15 +1815,19 @@ gst_vp9_enc_drain (GstVideoEncoder * video_encoder)
   int flags = 0;
   vpx_codec_err_t status;
   gint64 deadline;
+  vpx_codec_pts_t pts;
 
   encoder = GST_VP9_ENC (video_encoder);
 
   g_mutex_lock (&encoder->encoder_lock);
   deadline = encoder->deadline;
 
-  status =
+  pts =
-      vpx_codec_encode (&encoder->encoder, NULL, encoder->n_frames, 1, flags,
+      gst_util_uint64_scale (encoder->last_pts,
-      deadline);
+      encoder->cfg.g_timebase.num * (GstClockTime) GST_SECOND,
+      encoder->cfg.g_timebase.den);
+
+  status = vpx_codec_encode (&encoder->encoder, NULL, pts, 0, flags, deadline);
   g_mutex_unlock (&encoder->encoder_lock);
 
   if (status != 0) {
@@ -1856,11 +1859,18 @@ gst_vp9_enc_drain (GstVideoEncoder * video_encoder)
 static GstFlowReturn
 gst_vp9_enc_finish (GstVideoEncoder * video_encoder)
 {
+  GstVP9Enc *encoder;
   GstFlowReturn ret;
 
   GST_DEBUG_OBJECT (video_encoder, "finish");
 
+  encoder = GST_VP9_ENC (video_encoder);
+
+  if (encoder->inited) {
     ret = gst_vp9_enc_drain (video_encoder);
+  } else {
+    ret = GST_FLOW_OK;
+  }
 
   return ret;
 }
@@ -1892,13 +1902,13 @@ gst_vp9_enc_handle_frame (GstVideoEncoder * video_encoder,
   int flags = 0;
   vpx_image_t *image;
   GstVideoFrame vframe;
+  vpx_codec_pts_t pts;
+  unsigned long duration;
 
   GST_DEBUG_OBJECT (video_encoder, "handle_frame");
 
   encoder = GST_VP9_ENC (video_encoder);
 
-  encoder->n_frames++;
-
   GST_DEBUG_OBJECT (video_encoder, "size %d %d",
       GST_VIDEO_INFO_WIDTH (&encoder->input_state->info),
       GST_VIDEO_INFO_HEIGHT (&encoder->input_state->info));
@@ -1912,8 +1922,25 @@ gst_vp9_enc_handle_frame (GstVideoEncoder * video_encoder,
   }
 
   g_mutex_lock (&encoder->encoder_lock);
+  pts =
+      gst_util_uint64_scale (frame->pts,
+      encoder->cfg.g_timebase.num * (GstClockTime) GST_SECOND,
+      encoder->cfg.g_timebase.den);
+  encoder->last_pts = frame->pts;
+
+  if (frame->duration != GST_CLOCK_TIME_NONE) {
+    duration =
+        gst_util_uint64_scale (frame->duration,
+        encoder->cfg.g_timebase.num * (GstClockTime) GST_SECOND,
+        encoder->cfg.g_timebase.den);
+    encoder->last_pts += frame->duration;
+  } else {
+    duration = 1;
+  }
+
   status = vpx_codec_encode (&encoder->encoder, image,
-      encoder->n_frames, 1, flags, encoder->deadline);
+      pts, duration, flags, encoder->deadline);
+
   g_mutex_unlock (&encoder->encoder_lock);
   gst_video_frame_unmap (&vframe);
 

ext/vpx/gstvp9enc.h

@@ -102,8 +102,7 @@ struct _GstVP9Enc
 
   vpx_image_t image;
 
-  int n_frames;
+  GstClockTime last_pts;
-
   GstVideoCodecState *input_state;
 };