Make LibavH264Encoder flush frames out at the right rate when necessary

This means they'll get passed to FfmpegOutput at the rate that
generates the correct timestamps. We avoid pacing the output like this
when the output classes don't require it.

Currently not doing this for LibavMjpegEncoder because MJPEG files
don't have timestamps.

Signed-off-by: David Plowman <[email protected]>

picamera2/encoders/libav_h264_encoder.py

@@ -1,5 +1,6 @@
 """This is a base class for a multi-threaded software encoder."""
 
+import time
 from fractions import Fraction
 from math import sqrt
 
@@ -24,6 +25,8 @@ def __init__(self, bitrate=None, repeat=True, iperiod=30, framerate=30, qp=None,
         self.qp = qp
         self.profile = profile
         self.preset = None
+        self.drop_final_frames = False
+        self._lasttimestamp = None
 
     def _setup(self, quality):
         # If an explicit quality was specified, use it, otherwise try to preserve any bitrate/qp
@@ -94,8 +97,18 @@ def _start(self):
         self._av_input_format = FORMAT_TABLE[self._format]
 
     def _stop(self):
-        for packet in self._stream.encode():
-            self.outputframe(bytes(packet), packet.is_keyframe, timestamp=packet.pts)
+        if not self.drop_final_frames:
+            # Annoyingly, libav still has lots of encoded frames internally which we must flush
+            # out. If the output(s) doesn't understand timestamps, we may need to "pace" these
+            # frames with correct time intervals. Unpleasant.
+            for packet in self._stream.encode():
+                if any(out.needs_pacing for out in self._output) and self._lasttimestamp is not None:
+                    time_system, time_packet = self._lasttimestamp
+                    delay_us = packet.pts - time_packet - (time.monotonic_ns() - time_system) / 1000
+                    if delay_us > 0:
+                        time.sleep(delay_us / 1000000)
+                self._lasttimestamp = (time.monotonic_ns(), packet.pts)
+                self.outputframe(bytes(packet), packet.is_keyframe, timestamp=packet.pts)
         self._container.close()
 
     def _encode(self, stream, request):
@@ -104,4 +117,5 @@ def _encode(self, stream, request):
             frame = av.VideoFrame.from_ndarray(m.array, format=self._av_input_format, width=self.width)
             frame.pts = timestamp_us
             for packet in self._stream.encode(frame):
+                self._lasttimestamp = (time.monotonic_ns(), packet.pts)
                 self.outputframe(bytes(packet), packet.is_keyframe, timestamp=packet.pts)

picamera2/outputs/ffmpegoutput.py

@@ -48,6 +48,8 @@ def __init__(self, output_filename, audio=False, audio_device="default", audio_s
         self.timeout = 1 if audio else None
         # A user can set this to get notifications of FFmpeg failures.
         self.error_callback = None
+        # We don't understand timestamps, so an encoder may have to pace output to us.
+        self.needs_pacing = True
 
     def start(self):
         general_options = ['-loglevel', 'warning',

picamera2/outputs/output.py

@@ -12,6 +12,7 @@ def __init__(self, pts=None):
         """
         self.recording = False
         self.ptsoutput = pts
+        self.needs_pacing = False
 
     def start(self):
         """Start recording"""