Aren/haiku
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

FFMPEG Plugin: Fix video start_time handling

Browse Source 
- I misinterpreted the semantics of reordered_opaque. I thought it would
  establish the correct relationship between the start_time returned by
  GetNextChunk() and the next video frame successfully decoded. But for this
  to work reordered_opaque expects to be filled with presentation time stamps.
  A series of presentation time stamps may be jumping back in time due to the
  presence of B-frames. The decoded frame presentation time stamps series would
  then be ordered in a monotonically increased way.
  But actually GetNextChunk() always returns monotonically increasing start
  times. Mapping this behaviour to FFMPEG's expectations means labeling those
  start times as decoding time stamps (dts). Though for those start times to be
  related to the correct decoded video frames you have to assign the start time
  with the AVPacket containing the data to be decoded.

- This commit finally makes DVB video playback working for me with the TV app.
  Though no audio yet.

- The documentation was updated accordingly.
This commit is contained in:
Colin Günther 2014-08-10 13:31:45 +02:00
parent b77f1724a2
commit ed9de7dfca
2 changed files with 18 additions and 23 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

39
src/add-ons/media/plugins/ffmpeg/AVCodecDecoder.cpp
View File

@ -741,7 +741,7 @@ AVCodecDecoder::_DecodeNextVideoFrame()
while (true) {
status_t loadingChunkStatus
= _LoadNextVideoChunkIfNeededAndUpdateStartTime();
= _LoadNextVideoChunkIfNeededAndAssignStartTime();
if (loadingChunkStatus == B_LAST_BUFFER_ERROR)
return _FlushOneVideoFrameFromDecoderBuffer();
@ -822,21 +822,21 @@ AVCodecDecoder::_DecodeNextVideoFrame()
}
/*! \brief Loads the next video chunk into fVideoChunkBuffer and assigns it to
fTempPacket accordingly only if fTempPacket is empty. Updates fContext
with the start time of the new data chunk.
/*! \brief Loads the next video chunk into fVideoChunkBuffer and assigns it
(including the start time) to fTempPacket accordingly only if
fTempPacket is empty.
\returns B_OK
1. meaning: Next video chunk is loaded and fContext is updated.
2. meaning: No need to load and update anything. Proceed as usual.
1. meaning: Next video chunk is loaded.
2. meaning: No need to load and assign anything. Proceed as usual.
\returns B_LAST_BUFFER_ERROR No more video chunks available.
fVideoChunkBuffer, fTempPacket and fContext are left untouched.
\returns Other errors Caller should bail out because fVideoChunkBuffer,
fTempPacket and fContext are in unknown states. Normal operation cannot
be guaranteed.
fVideoChunkBuffer and fTempPacket are left untouched.
\returns Other errors Caller should bail out because fVideoChunkBuffer and
fTempPacket are in unknown states. Normal operation cannot be
guaranteed.
*/
status_t
AVCodecDecoder::_LoadNextVideoChunkIfNeededAndUpdateStartTime()
AVCodecDecoder::_LoadNextVideoChunkIfNeededAndAssignStartTime()
{
// TODO: Rename fVideoChunkBuffer to fChunkBuffer, once the audio path is
// responsible for releasing the chunk buffer, too.
@ -863,16 +863,11 @@ AVCodecDecoder::_LoadNextVideoChunkIfNeededAndUpdateStartTime()
fTempPacket.data = fVideoChunkBuffer;
fTempPacket.size = fChunkBufferSize;
fContext->reordered_opaque = chunkMediaHeader.start_time;
// Let FFMPEG handle the relationship between start_time and
// decoded video frame.
//
// Explanation:
// The received chunk buffer may not contain the next video
// frame to be decoded, due to frame reordering (e.g. MPEG1/2
// provides encoded video frames in a different order than the
// decoded video frame).
fTempPacket.dts = chunkMediaHeader.start_time;
// Let FFMPEG handle the correct relationship between start_time and
// decoded video frame. By doing so we are simply copying the way how
// it is implemented in ffplay.c
// \see http://git.videolan.org/?p=ffmpeg.git;a=blob;f=ffplay.c;h=09623db374e5289ed20b7cc28c262c4375a8b2e4;hb=9153b33a742c4e2a85ff6230aea0e75f5a8b26c2#l1502
//
// FIXME: Research how to establish a meaningful relationship
// between start_time and decoded video frame when the received
@ -1039,7 +1034,7 @@ AVCodecDecoder::_UpdateMediaHeaderForVideoFrame()
fHeader.type = B_MEDIA_RAW_VIDEO;
fHeader.file_pos = 0;
fHeader.orig_size = 0;
fHeader.start_time = fRawDecodedPicture->reordered_opaque;
fHeader.start_time = fRawDecodedPicture->pkt_dts;
fHeader.size_used = fDecodedDataSizeInBytes;
fHeader.u.raw_video.display_line_width = fRawDecodedPicture->width;
fHeader.u.raw_video.display_line_count = fRawDecodedPicture->height;

2
src/add-ons/media/plugins/ffmpeg/AVCodecDecoder.h
View File

@ -61,7 +61,7 @@ private:
media_header* mediaHeader,
media_decode_info* info);
status_t _DecodeNextVideoFrame();
status_t _LoadNextVideoChunkIfNeededAndUpdateStartTime();
status_t _LoadNextVideoChunkIfNeededAndAssignStartTime();
// TODO: Remove the "Video" word once
// the audio path is responsible for
// freeing the chunk buffer, too.