Recherche avancée

Sur d’autres sites (11554)

• ffmpeg concat doesn't keep video speed/framerate ?

  28 août 2016, par Nick

  I have a bunch of small webm clips in directory vidchunks. These clips were generated using javascript MediaRecorder API.

  MediaRecorder code :

  var mRecorder;
  
  var chunks = [];
  
  navigator.mediaDevices.getUserMedia({
  
    audio: false,
  
    video: {
  
    width: 1280,
  
    height: 760,
  
    mozMediaSource: "screen",
  
    mediaSource: "screen"
  
  }
  
  }).then(function(stream) {
  
    mRecorder = new MediaRecorder(stream,{mimeType:"video/webm"});
  
    mRecorder.ondataavailable = function(event) {
  
      var blob = event.data;
  
      chunks.push(blob);
  
      var vidchunk = new Blob(chunks);
  
      // save to directory vid chunk.
  
    };
  
    mRecorder.start(5000);
  
  }).catch(function(error) {
  
    console.log(error.message);
  
  });

  Then, once I have a bunch of these 5 second clips, I merge them :

  ffmpeg -safe 0 \
  
  -f concat -i <(for f in `ls vidchunks/* | sort -V`; do echo file $f; done) \
  
  -c:v copy -r 30 -vsync drop test.webm

  When I play back these chunks individually, they play correctly. All 5 seconds of video are rendered smoothly.

  However, once I merge, some 5 second chunks are condensed to like 1 second. So the total video, which should be 50 seconds for 10 chunks, turns out to be like 38 seconds because three of the chunks got condensed to 1 second.

  Any ideas on how to fix this ?

  EDIT : I’ve tried :

  • without "-r 30 -vsync drop"
  • without "-c:v copy"

  No changes...

  EDIT2 (some ffprobe outputs) :

  ffprobe 1472343170-1.webm

  ffprobe version 2.8.6-1ubuntu2 Copyright (c) 2007-2016 the FFmpeg developers
  
    built with gcc 5.3.1 (Ubuntu 5.3.1-11ubuntu1) 20160311
  
    configuration: --prefix=/usr --extra-version=1ubuntu2 --build-suffix=-ffmpeg --toolchain=hardened --libdir=/usr/lib/x86_64-linux-gnu --incdir=/usr/include/x86_64-linux-gnu --cc=cc --cxx=g++ --enable-gpl --enable-shared --disable-stripping --disable-decoder=libopenjpeg --disable-decoder=libschroedinger --enable-avresample --enable-avisynth --enable-gnutls --enable-ladspa --enable-libass --enable-libbluray --enable-libbs2b --enable-libcaca --enable-libcdio --enable-libflite --enable-libfontconfig --enable-libfreetype --enable-libfribidi --enable-libgme --enable-libgsm --enable-libmodplug --enable-libmp3lame --enable-libopenjpeg --enable-libopus --enable-libpulse --enable-librtmp --enable-libschroedinger --enable-libshine --enable-libsnappy --enable-libsoxr --enable-libspeex --enable-libssh --enable-libtheora --enable-libtwolame --enable-libvorbis --enable-libvpx --enable-libwavpack --enable-libwebp --enable-libx265 --enable-libxvid --enable-libzvbi --enable-openal --enable-opengl --enable-x11grab --enable-libdc1394 --enable-libiec61883 --enable-libzmq --enable-frei0r --enable-libx264 --enable-libopencv
  
    libavutil      54. 31.100 / 54. 31.100
  
    libavcodec     56. 60.100 / 56. 60.100
  
    libavformat    56. 40.101 / 56. 40.101
  
    libavdevice    56.  4.100 / 56.  4.100
  
    libavfilter     5. 40.101 /  5. 40.101
  
    libavresample   2.  1.  0 /  2.  1.  0
  
    libswscale      3.  1.101 /  3.  1.101
  
    libswresample   1.  2.101 /  1.  2.101
  
    libpostproc    53.  3.100 / 53.  3.100
  
  Input #0, matroska,webm, from '1472343170-1.webm':
  
    Metadata:
  
      encoder         : Chrome
  
    Duration: N/A, start: 0.000000, bitrate: N/A
  
      Stream #0:0(eng): Video: vp8, yuv420p, 1280x760, SAR 1:1 DAR 32:19, 30 fps, 30 tbr, 1k tbn, 1k tbc (default)

  ffprobe 1472343245-16.webm

  ffprobe version 2.8.6-1ubuntu2 Copyright (c) 2007-2016 the FFmpeg developers
  
    built with gcc 5.3.1 (Ubuntu 5.3.1-11ubuntu1) 20160311
  
    configuration: --prefix=/usr --extra-version=1ubuntu2 --build-suffix=-ffmpeg --toolchain=hardened --libdir=/usr/lib/x86_64-linux-gnu --incdir=/usr/include/x86_64-linux-gnu --cc=cc --cxx=g++ --enable-gpl --enable-shared --disable-stripping --disable-decoder=libopenjpeg --disable-decoder=libschroedinger --enable-avresample --enable-avisynth --enable-gnutls --enable-ladspa --enable-libass --enable-libbluray --enable-libbs2b --enable-libcaca --enable-libcdio --enable-libflite --enable-libfontconfig --enable-libfreetype --enable-libfribidi --enable-libgme --enable-libgsm --enable-libmodplug --enable-libmp3lame --enable-libopenjpeg --enable-libopus --enable-libpulse --enable-librtmp --enable-libschroedinger --enable-libshine --enable-libsnappy --enable-libsoxr --enable-libspeex --enable-libssh --enable-libtheora --enable-libtwolame --enable-libvorbis --enable-libvpx --enable-libwavpack --enable-libwebp --enable-libx265 --enable-libxvid --enable-libzvbi --enable-openal --enable-opengl --enable-x11grab --enable-libdc1394 --enable-libiec61883 --enable-libzmq --enable-frei0r --enable-libx264 --enable-libopencv
  
    libavutil      54. 31.100 / 54. 31.100
  
    libavcodec     56. 60.100 / 56. 60.100
  
    libavformat    56. 40.101 / 56. 40.101
  
    libavdevice    56.  4.100 / 56.  4.100
  
    libavfilter     5. 40.101 /  5. 40.101
  
    libavresample   2.  1.  0 /  2.  1.  0
  
    libswscale      3.  1.101 /  3.  1.101
  
    libswresample   1.  2.101 /  1.  2.101
  
    libpostproc    53.  3.100 / 53.  3.100
  
  Input #0, matroska,webm, from '1472343245-16.webm':
  
    Metadata:
  
      encoder         : Chrome
  
    Duration: N/A, start: 0.000000, bitrate: N/A
  
      Stream #0:0(eng): Video: vp8, yuv420p, 1280x760, SAR 1:1 DAR 32:19, 30 fps, 30 tbr, 1k tbn, 1k tbc (default)

  ffprobe 1472343350-37.webm

  ffprobe version 2.8.6-1ubuntu2 Copyright (c) 2007-2016 the FFmpeg developers
  
    built with gcc 5.3.1 (Ubuntu 5.3.1-11ubuntu1) 20160311
  
    configuration: --prefix=/usr --extra-version=1ubuntu2 --build-suffix=-ffmpeg --toolchain=hardened --libdir=/usr/lib/x86_64-linux-gnu --incdir=/usr/include/x86_64-linux-gnu --cc=cc --cxx=g++ --enable-gpl --enable-shared --disable-stripping --disable-decoder=libopenjpeg --disable-decoder=libschroedinger --enable-avresample --enable-avisynth --enable-gnutls --enable-ladspa --enable-libass --enable-libbluray --enable-libbs2b --enable-libcaca --enable-libcdio --enable-libflite --enable-libfontconfig --enable-libfreetype --enable-libfribidi --enable-libgme --enable-libgsm --enable-libmodplug --enable-libmp3lame --enable-libopenjpeg --enable-libopus --enable-libpulse --enable-librtmp --enable-libschroedinger --enable-libshine --enable-libsnappy --enable-libsoxr --enable-libspeex --enable-libssh --enable-libtheora --enable-libtwolame --enable-libvorbis --enable-libvpx --enable-libwavpack --enable-libwebp --enable-libx265 --enable-libxvid --enable-libzvbi --enable-openal --enable-opengl --enable-x11grab --enable-libdc1394 --enable-libiec61883 --enable-libzmq --enable-frei0r --enable-libx264 --enable-libopencv
  
    libavutil      54. 31.100 / 54. 31.100
  
    libavcodec     56. 60.100 / 56. 60.100
  
    libavformat    56. 40.101 / 56. 40.101
  
    libavdevice    56.  4.100 / 56.  4.100
  
    libavfilter     5. 40.101 /  5. 40.101
  
    libavresample   2.  1.  0 /  2.  1.  0
  
    libswscale      3.  1.101 /  3.  1.101
  
    libswresample   1.  2.101 /  1.  2.101
  
    libpostproc    53.  3.100 / 53.  3.100
  
  Input #0, matroska,webm, from '1472343350-37.webm':
  
    Metadata:
  
      encoder         : Chrome
  
    Duration: N/A, start: 0.000000, bitrate: N/A
  
      Stream #0:0(eng): Video: vp8, yuv420p, 1280x760, SAR 1:1 DAR 32:19, 30 fps, 30 tbr, 1k tbn, 1k tbc (default)

  I also noticed that the merged video plays normal when the mouse is moving around on the screen, but when the mouse is not moving, it moves 5x faster.

  I know this because I have a timer on the screen that I’m recording, and I can see that the timer speeds up significantly when mouse isn’t moving but when mouse is moving, the timer moves at normal speed.

  EDIT3 :
  The DTS/PTS looks like resets after each file is merged :

  DTS -9223363446915256, next:6866667 st:0 invalid dropping
  
  PTS -9223363446915256, next:6866667 invalid dropping st:0
  
  DTS -9223363446915222, next:6900000 st:0 invalid dropping
  
  PTS -9223363446915222, next:6900000 invalid dropping st:0
  
  DTS -9223363446915188, next:6933333 st:0 invalid dropping
  
  PTS -9223363446915188, next:6933333 invalid dropping st:0
  
  [concat @ 0x1cde3e0] DTS -9223363446920184 < 5006 out of order
  
  DTS -9223363446920184, next:5033333 st:0 invalid dropping
  
  PTS -9223363446920184, next:5033333 invalid dropping st:0
  
  DTS -9223363446920056, next:5066667 st:0 invalid dropping
  
  PTS -9223363446920056, next:5066667 invalid dropping st:0
  
  DTS -9223363446919956, next:5100000 st:0 invalid dropping
  
  PTS -9223363446919956, next:5100000 invalid dropping st:0

  EDIT4 : Tried ffmpeg -i "$f" -y -c copy -fflags +genpts "$f" and then merge again.

  Works a lot better, but now other files are getting skipped.

  Here’s ffprobe from a file that’s skipped :

  Metadata:
  
      encoder         : Lavf56.40.101
  
    Duration: 00:00:01.17, start: 0.000000, bitrate: 428 kb/s
  
      Stream #0:0(eng): Video: vp8, yuv420p, 1280x760, SAR 1:1 DAR 32:19, 30 fps, 30 tbr, 1k tbn, 1k tbc (default)
  
  [webm @ 0x1c01940] Codec for stream 0 does not use global headers but container format requires global headers
  
  Output #0, webm, to '1472343225-12.webm':
  
    Metadata:
  
      encoder         : Lavf56.40.101
  
      Stream #0:0(eng): Video: vp8, yuv420p, 1280x760 [SAR 1:1 DAR 32:19], q=2-31, 30 fps, 30 tbr, 1k tbn, 1k tbc (default)
  
  Stream mapping:
  
    Stream #0:0 -> #0:0 (copy)

  Here’s ffprobe from a file not skipped :

  Input #0, matroska,webm, from '1472343215-10.webm':
  
    Metadata:
  
      encoder         : Lavf56.40.101
  
    Duration: 00:00:05.03, start: 0.000000, bitrate: 80 kb/s
  
      Stream #0:0(eng): Video: vp8, yuv420p, 1280x760, SAR 1:1 DAR 32:19, 30 fps, 30 tbr, 1k tbn, 1k tbc (default)
  
  [webm @ 0x1349940] Codec for stream 0 does not use global headers but container format requires global headers
  
  Output #0, webm, to '1472343215-10.webm':
  
    Metadata:
  
      encoder         : Lavf56.40.101
  
      Stream #0:0(eng): Video: vp8, yuv420p, 1280x760 [SAR 1:1 DAR 32:19], q=2-31, 30 fps, 30 tbr, 1k tbn, 1k tbc (default)
  
  Stream mapping:
  
    Stream #0:0 -> #0:0 (copy)

• Decoding and playing audio with ffmpeg and XAudio2 - frequency raito wrong

  12 juillet 2016, par Brent de Carteret

  I’m using ffmpeg to decode audio and output it using the XAudio2 API, it works and plays synced with the video output using the pts. But it’s high pitched (i.e. sounds like chipmunks).

  Setting breakpoints I can see it has sets the correct sample rate from the audio codec in CreateSourceVoice. I’m stumped.

  Any help would be much appreciated.

  #include "DVDAudioDevice.h"
  
  
  
  HANDLE m_hBufferEndEvent;
  
  
  
  CDVDAudio::CDVDAudio()
  
  {
  
      m_pXAudio2 = NULL;
  
      m_pMasteringVoice = NULL;
  
      m_pSourceVoice = NULL;
  
      m_pWfx  = NULL;
  
  
  
      m_VoiceCallback = NULL;
  
  
  
      m_hBufferEndEvent = CreateEvent(NULL, false, false, "Buffer end event");
  
  }
  
  
  
  CDVDAudio::~CDVDAudio()
  
  {
  
      m_pXAudio2 = NULL;
  
      m_pMasteringVoice = NULL;
  
      m_pSourceVoice = NULL;
  
      m_pWfx  = NULL;
  
  
  
      m_VoiceCallback = NULL;
  
  
  
      CloseHandle(m_hBufferEndEvent);
  
      m_hBufferEndEvent = NULL;
  
  }
  
  
  
  bool CDVDAudio::Create(int iChannels, int iBitrate, int iBitsPerSample, bool bPasstrough)
  
  {
  
      CoInitializeEx(NULL, COINIT_MULTITHREADED);
  
  
  
      HRESULT hr = XAudio2Create( &m_pXAudio2, 0, XAUDIO2_DEFAULT_PROCESSOR);
  
  
  
      if (SUCCEEDED(hr))
  
      {
  
          m_pXAudio2->CreateMasteringVoice( &m_pMasteringVoice );
  
      }
  
  
  
      // Create source voice
  
      WAVEFORMATEXTENSIBLE wfx;
  
      memset(&wfx, 0, sizeof(WAVEFORMATEXTENSIBLE));
  
  
  
      wfx.Format.wFormatTag           = WAVE_FORMAT_PCM;
  
      wfx.Format.nSamplesPerSec       = iBitrate;//pFFMpegData->pAudioCodecCtx->sample_rate;//48000 by default
  
      wfx.Format.nChannels            = iChannels;//pFFMpegData->pAudioCodecCtx->channels;
  
      wfx.Format.wBitsPerSample       = 16;
  
      wfx.Format.nBlockAlign          = wfx.Format.nChannels*16/8;
  
      wfx.Format.nAvgBytesPerSec      = wfx.Format.nSamplesPerSec * wfx.Format.nBlockAlign;
  
      wfx.Format.cbSize               = sizeof(WAVEFORMATEXTENSIBLE)-sizeof(WAVEFORMATEX);
  
      wfx.Samples.wValidBitsPerSample = wfx.Format.wBitsPerSample;
  
  
  
      if(wfx.Format.nChannels == 1)
  
      {
  
          wfx.dwChannelMask = SPEAKER_MONO;
  
      }
  
      else if(wfx.Format.nChannels == 2)
  
      {
  
          wfx.dwChannelMask = SPEAKER_STEREO;
  
      }
  
      else if(wfx.Format.nChannels == 5)
  
      {
  
          wfx.dwChannelMask = SPEAKER_5POINT1;
  
      }
  
  
  
      wfx.SubFormat = KSDATAFORMAT_SUBTYPE_PCM;
  
  
  
      unsigned int flags = 0;//XAUDIO2_VOICE_NOSRC;// | XAUDIO2_VOICE_NOPITCH;
  
  
  
      //Source voice
  
  
  
      m_VoiceCallback = new StreamingVoiceCallback(this);
  
  
  
      hr = m_pXAudio2->CreateSourceVoice(&m_pSourceVoice,(WAVEFORMATEX*)&wfx, 0 , 1.0f, m_VoiceCallback);
  
  
  
      if(!SUCCEEDED(hr))
  
          return false;
  
  
  
      // Start sound
  
      hr = m_pSourceVoice->Start(0);
  
  
  
      if(!SUCCEEDED(hr))
  
          return false;
  
  
  
      return true;
  
  }
  
  
  
  DWORD CDVDAudio::AddPackets(unsigned char* data, DWORD len)
  
  {  
  
          memset(&m_SoundBuffer,0,sizeof(XAUDIO2_BUFFER));
  
  
  
          m_SoundBuffer.AudioBytes = len;
  
          m_SoundBuffer.pAudioData = data;
  
          m_SoundBuffer.pContext = NULL;//(VOID*)data;
  
  
  
          XAUDIO2_VOICE_STATE state;
  
  
  
          while(m_pSourceVoice->GetState( &state ), state.BuffersQueued > 60)
  
          {
  
              WaitForSingleObject( m_hBufferEndEvent, INFINITE );
  
          }
  
  
  
          m_pSourceVoice->SubmitSourceBuffer( &m_SoundBuffer );
  
  
  
      return 0;
  
  }
  
  
  
  void CDVDAudio::Destroy()
  
  {
  
      m_pMasteringVoice->DestroyVoice();
  
      m_pXAudio2->Release();
  
  
  
      m_pSourceVoice->DestroyVoice();
  
  
  
      delete m_VoiceCallback;
  
      m_VoiceCallback = NULL;
  
  }

  #include "DVDAudioCodecFFmpeg.h"
  
  #include "Log.h"
  
  
  
  CDVDAudioCodecFFmpeg::CDVDAudioCodecFFmpeg() : CDVDAudioCodec()
  
  {
  
      m_iBufferSize = 0;
  
      m_pCodecContext = NULL;
  
      m_bOpenedCodec = false;
  
  }
  
  
  
  CDVDAudioCodecFFmpeg::~CDVDAudioCodecFFmpeg()
  
  {
  
      Dispose();
  
  }
  
  
  
  bool CDVDAudioCodecFFmpeg::Open(AVCodecID codecID, int iChannels, int iSampleRate)
  
  {
  
      AVCodec* pCodec;
  
      m_bOpenedCodec = false;
  
  
  
      av_register_all();
  
  
  
      pCodec = avcodec_find_decoder(codecID);
  
  
  
      m_pCodecContext = avcodec_alloc_context3(pCodec);//avcodec_alloc_context();
  
      avcodec_get_context_defaults3(m_pCodecContext, pCodec);
  
  
  
      if (!pCodec)
  
      {
  
          CLog::Log(LOGERROR, "CDVDAudioCodecFFmpeg::Open() Unable to find codec");
  
          return false;
  
      }
  
  
  
      m_pCodecContext->debug_mv = 0;
  
      m_pCodecContext->debug = 0;
  
      m_pCodecContext->workaround_bugs = 1;
  
  
  
      if (pCodec->capabilities & CODEC_CAP_TRUNCATED)
  
          m_pCodecContext->flags |= CODEC_FLAG_TRUNCATED;
  
  
  
      m_pCodecContext->channels = iChannels;
  
      m_pCodecContext->sample_rate = iSampleRate;
  
      //m_pCodecContext->bits_per_sample = 24;
  
  
  
  /* //FIXME BRENT
  
    if( ExtraData && ExtraSize > 0 )
  
    {
  
      m_pCodecContext->extradata_size = ExtraSize;
  
      m_pCodecContext->extradata = m_dllAvCodec.av_mallocz(ExtraSize + FF_INPUT_BUFFER_PADDING_SIZE);
  
      memcpy(m_pCodecContext->extradata, ExtraData, ExtraSize);
  
    }
  
  */
  
      // set acceleration
  
      //m_pCodecContext->dsp_mask = FF_MM_FORCE | FF_MM_MMX | FF_MM_MMXEXT | FF_MM_SSE; //BRENT
  
  
  
      if (avcodec_open2(m_pCodecContext, pCodec, NULL) < 0)
  
      {
  
          CLog::Log(LOGERROR, "CDVDAudioCodecFFmpeg::Open() Unable to open codec");
  
          Dispose();
  
          return false;
  
      }
  
  
  
      m_bOpenedCodec = true;
  
      return true;
  
  }
  
  
  
  void CDVDAudioCodecFFmpeg::Dispose()
  
  {
  
      if (m_pCodecContext)
  
      {
  
          if (m_bOpenedCodec) avcodec_close(m_pCodecContext);
  
          m_bOpenedCodec = false;
  
          av_free(m_pCodecContext);
  
          m_pCodecContext = NULL;
  
      }
  
      m_iBufferSize = 0;
  
  }
  
  
  
  int CDVDAudioCodecFFmpeg::Decode(BYTE* pData, int iSize)
  
  {
  
      int iBytesUsed;
  
      if (!m_pCodecContext) return -1;
  
  
  
      //Copy into a FFMpeg AVPAcket again
  
      AVPacket packet;
  
      av_init_packet(&packet);
  
  
  
      packet.data=pData;
  
      packet.size=iSize;
  
  
  
      int iOutputSize = AVCODEC_MAX_AUDIO_FRAME_SIZE; //BRENT
  
  
  
      iBytesUsed = avcodec_decode_audio3(m_pCodecContext, (int16_t *)m_buffer, &iOutputSize/*m_iBufferSize*/, &packet);
  
  
  
      m_iBufferSize = iOutputSize;//BRENT
  
  
  
      return iBytesUsed;
  
  }
  
  
  
  int CDVDAudioCodecFFmpeg::GetData(BYTE** dst)
  
  {
  
      *dst = m_buffer;
  
      return m_iBufferSize;
  
  }
  
  
  
  void CDVDAudioCodecFFmpeg::Reset()
  
  {
  
      if (m_pCodecContext) avcodec_flush_buffers(m_pCodecContext);
  
  }
  
  
  
  int CDVDAudioCodecFFmpeg::GetChannels()
  
  {
  
      if (m_pCodecContext) return m_pCodecContext->channels;
  
      return 0;
  
  }
  
  
  
  int CDVDAudioCodecFFmpeg::GetSampleRate()
  
  {
  
      if (m_pCodecContext) return m_pCodecContext->sample_rate;
  
      return 0;
  
  }
  
  
  
  int CDVDAudioCodecFFmpeg::GetBitsPerSample()
  
  {
  
      if (m_pCodecContext) return 16;
  
      return 0;
  
  }
  #include "DVDPlayerAudio.h"
  
  #include "DVDDemuxUtils.h"
  
  #include "Log.h"
  
  
  
  #include 
  
  #include "DVDAudioCodecFFmpeg.h" //FIXME Move to a codec factory!!
  
  
  
  CDVDPlayerAudio::CDVDPlayerAudio(CDVDClock* pClock) : CThread()
  
  {
  
      m_pClock = pClock;
  
      m_pAudioCodec = NULL;
  
      m_bInitializedOutputDevice = false;
  
      m_iSourceChannels = 0;
  
      m_audioClock = 0;
  
  
  
  //  m_currentPTSItem.pts = DVD_NOPTS_VALUE;
  
  //  m_currentPTSItem.timestamp = 0;
  
  
  
      SetSpeed(DVD_PLAYSPEED_NORMAL);
  
  
  
      InitializeCriticalSection(&m_critCodecSection);
  
      m_messageQueue.SetMaxDataSize(10 * 16 * 1024);
  
  //  g_dvdPerformanceCounter.EnableAudioQueue(&m_packetQueue);
  
  }
  
  
  
  CDVDPlayerAudio::~CDVDPlayerAudio()
  
  {
  
  //  g_dvdPerformanceCounter.DisableAudioQueue();
  
  
  
      // close the stream, and don't wait for the audio to be finished
  
      CloseStream(true);
  
      DeleteCriticalSection(&m_critCodecSection);
  
  }
  
  
  
  bool CDVDPlayerAudio::OpenStream( CDemuxStreamAudio *pDemuxStream )                                 
  
  {
  
      // should alway's be NULL!!!!, it will probably crash anyway when deleting m_pAudioCodec here.
  
      if (m_pAudioCodec)
  
      {
  
          CLog::Log(LOGFATAL, "CDVDPlayerAudio::OpenStream() m_pAudioCodec != NULL");
  
          return false;
  
      }
  
  
  
      AVCodecID codecID = pDemuxStream->codec;
  
  
  
      CLog::Log(LOGNOTICE, "Finding audio codec for: %i", codecID);
  
      //m_pAudioCodec = CDVDFactoryCodec::CreateAudioCodec( pDemuxStream ); 
  
      m_pAudioCodec = new CDVDAudioCodecFFmpeg; //FIXME BRENT Codec Factory needed!
  
  
  
      if (!m_pAudioCodec->Open(pDemuxStream->codec, pDemuxStream->iChannels, pDemuxStream->iSampleRate))
  
      {
  
          m_pAudioCodec->Dispose();
  
          delete m_pAudioCodec;
  
          m_pAudioCodec = NULL;
  
          return false;
  
      }
  
  
  
      if( !m_pAudioCodec )
  
      {
  
          CLog::Log(LOGERROR, "Unsupported audio codec");
  
          return false;
  
      }
  
  
  
      m_codec = pDemuxStream->codec;
  
      m_iSourceChannels = pDemuxStream->iChannels;
  
  
  
      m_messageQueue.Init();
  
  
  
      CLog::Log(LOGNOTICE, "Creating audio thread");
  
      Create();
  
  
  
      return true;
  
  }
  
  
  
  void CDVDPlayerAudio::CloseStream(bool bWaitForBuffers)
  
  {
  
      // wait until buffers are empty
  
      if (bWaitForBuffers) m_messageQueue.WaitUntilEmpty();
  
  
  
      // send abort message to the audio queue
  
      m_messageQueue.Abort();
  
  
  
      CLog::Log(LOGNOTICE, "waiting for audio thread to exit");
  
  
  
      // shut down the adio_decode thread and wait for it
  
      StopThread(); // will set this->m_bStop to true
  
      this->WaitForThreadExit(INFINITE);
  
  
  
      // uninit queue
  
      m_messageQueue.End();
  
  
  
      CLog::Log(LOGNOTICE, "Deleting audio codec");
  
      if (m_pAudioCodec)
  
      {
  
          m_pAudioCodec->Dispose();
  
          delete m_pAudioCodec;
  
          m_pAudioCodec = NULL;
  
      }
  
  
  
      // flush any remaining pts values
  
      //FlushPTSQueue(); //FIXME BRENT
  
  }
  
  
  
  void CDVDPlayerAudio::OnStartup()
  
  {
  
      CThread::SetName("CDVDPlayerAudio");
  
      pAudioPacket = NULL;
  
      m_audioClock = 0;
  
      audio_pkt_data = NULL;
  
      audio_pkt_size = 0;
  
  
  
  //  g_dvdPerformanceCounter.EnableAudioDecodePerformance(ThreadHandle());
  
  }
  
  
  
  void CDVDPlayerAudio::Process()
  
  {
  
      CLog::Log(LOGNOTICE, "running thread: CDVDPlayerAudio::Process()");
  
  
  
      int result;
  
  
  
      // silence data
  
      BYTE silence[1024];
  
      memset(silence, 0, 1024);
  
  
  
      DVDAudioFrame audioframe;
  
  
  
      __int64 iClockDiff=0;
  
      while (!m_bStop)
  
      {
  
          //Don't let anybody mess with our global variables
  
          EnterCriticalSection(&m_critCodecSection);
  
          result = DecodeFrame(audioframe, m_speed != DVD_PLAYSPEED_NORMAL); // blocks if no audio is available, but leaves critical section before doing so
  
          LeaveCriticalSection(&m_critCodecSection);
  
  
  
          if( result & DECODE_FLAG_ERROR ) 
  
          {      
  
              CLog::Log(LOGERROR, "CDVDPlayerAudio::Process - Decode Error. Skipping audio frame");
  
              continue;
  
          }
  
  
  
          if( result & DECODE_FLAG_ABORT )
  
          {
  
              CLog::Log(LOGDEBUG, "CDVDPlayerAudio::Process - Abort recieved, exiting thread");
  
              break;
  
          }
  
  
  
          if( result & DECODE_FLAG_DROP ) //FIXME BRENT
  
          {
  
  /*          //frame should be dropped. Don't let audio move ahead of the current time thou
  
              //we need to be able to start playing at any time
  
              //when playing backwords, we try to keep as small buffers as possible
  
  
  
              // set the time at this delay
  
              AddPTSQueue(audioframe.pts, m_dvdAudio.GetDelay());
  
  */
  
              if (m_speed > 0)
  
              {
  
                  __int64 timestamp = m_pClock->GetAbsoluteClock() + (audioframe.duration * DVD_PLAYSPEED_NORMAL) / m_speed;
  
                  while( !m_bStop && timestamp > m_pClock->GetAbsoluteClock() ) Sleep(1);
  
              }
  
              continue;
  
          }
  
  
  
          if( audioframe.size > 0 ) 
  
          {
  
              // we have succesfully decoded an audio frame, openup the audio device if not already done
  
              if (!m_bInitializedOutputDevice)
  
              {
  
                  m_bInitializedOutputDevice = InitializeOutputDevice();
  
              }
  
  
  
              //Add any packets play
  
              m_dvdAudio.AddPackets(audioframe.data, audioframe.size);
  
  
  
              // store the delay for this pts value so we can calculate the current playing
  
              //AddPTSQueue(audioframe.pts, m_dvdAudio.GetDelay() - audioframe.duration);//BRENT
  
          }
  
  
  
          // if we where asked to resync on this packet, do so here
  
          if( result & DECODE_FLAG_RESYNC )
  
          {      
  
              CLog::Log(LOGDEBUG, "CDVDPlayerAudio::Process - Resync recieved.");
  
              //while (!m_bStop && (unsigned int)m_dvdAudio.GetDelay() > audioframe.duration ) Sleep(5); //BRENT
  
              m_pClock->Discontinuity(CLOCK_DISC_NORMAL, audioframe.pts);
  
          }
  
  
  
  #ifdef USEOLDSYNC
  
          //Clock should be calculated after packets have been added as m_audioClock points to the 
  
          //time after they have been played
  
  
  
          const __int64 iCurrDiff = (m_audioClock - m_dvdAudio.GetDelay()) - m_pClock->GetClock();
  
          const __int64 iAvDiff = (iClockDiff + iCurrDiff)/2;
  
  
  
          //Check for discontinuity in the stream, use a moving average to
  
          //eliminate highfreq fluctuations of large packet sizes
  
          if( ABS(iAvDiff) > 5000 ) // sync clock if average diff is bigger than 5 msec 
  
          {
  
              //Wait untill only the new audio frame wich triggered the discontinuity is left
  
              //then set disc state
  
              while (!m_bStop && (unsigned int)m_dvdAudio.GetBytesInBuffer() > audioframe.size ) Sleep(5);
  
  
  
              m_pClock->Discontinuity(CLOCK_DISC_NORMAL, m_audioClock - m_dvdAudio.GetDelay());
  
              CLog::("CDVDPlayer:: Detected Audio Discontinuity, syncing clock. diff was: %I64d, %I64d, av: %I64d", iClockDiff, iCurrDiff, iAvDiff);
  
              iClockDiff = 0;
  
          }
  
          else
  
          {      
  
              //Do gradual adjustments (not working yet)
  
              //m_pClock->AdjustSpeedToMatch(iClock + iAvDiff);
  
              iClockDiff = iCurrDiff;
  
          }
  
  #endif
  
      }
  
  }
  
  
  
  void CDVDPlayerAudio::OnExit()
  
  {
  
      //g_dvdPerformanceCounter.DisableAudioDecodePerformance();
  
  
  
      // destroy audio device
  
      CLog::Log(LOGNOTICE, "Closing audio device");
  
      m_dvdAudio.Destroy();
  
      m_bInitializedOutputDevice = false;
  
  
  
      CLog::Log(LOGNOTICE, "thread end: CDVDPlayerAudio::OnExit()");
  
  }
  
  
  
  // decode one audio frame and returns its uncompressed size
  
  int CDVDPlayerAudio::DecodeFrame(DVDAudioFrame &audioframe, bool bDropPacket)
  
  {
  
      CDVDDemux::DemuxPacket* pPacket = pAudioPacket;
  
      int n=48000*2*16/8, len;
  
  
  
      //Store amount left at this point, and what last pts was
  
      unsigned __int64 first_pkt_pts = 0;
  
      int first_pkt_size = 0; 
  
      int first_pkt_used = 0;
  
      int result = 0;
  
  
  
      // make sure the sent frame is clean
  
      memset(&audioframe, 0, sizeof(DVDAudioFrame));
  
  
  
      if (pPacket)
  
      {
  
          first_pkt_pts = pPacket->pts;
  
          first_pkt_size = pPacket->iSize;
  
          first_pkt_used = first_pkt_size - audio_pkt_size;
  
      }
  
  
  
      for (;;)
  
      {
  
          /* NOTE: the audio packet can contain several frames */
  
          while (audio_pkt_size > 0)
  
          {
  
              len = m_pAudioCodec->Decode(audio_pkt_data, audio_pkt_size);
  
              if (len < 0)
  
              {
  
                  /* if error, we skip the frame */
  
                  audio_pkt_size=0;
  
                  m_pAudioCodec->Reset();
  
                  break;
  
              }
  
  
  
              // fix for fucked up decoders //FIXME BRENT
  
              if( len > audio_pkt_size )
  
              {        
  
                  CLog::Log(LOGERROR, "CDVDPlayerAudio:DecodeFrame - Codec tried to consume more data than available. Potential memory corruption");        
  
                  audio_pkt_size=0;
  
                  m_pAudioCodec->Reset();
  
                  assert(0);
  
              }
  
  
  
              // get decoded data and the size of it
  
              audioframe.size = m_pAudioCodec->GetData(&audioframe.data);
  
  
  
              audio_pkt_data += len;
  
              audio_pkt_size -= len;
  
  
  
              if (audioframe.size <= 0) continue;
  
  
  
              audioframe.pts = m_audioClock;
  
  
  
              // compute duration.
  
              n = m_pAudioCodec->GetChannels() * m_pAudioCodec->GetBitsPerSample() / 8 * m_pAudioCodec->GetSampleRate();
  
              if (n > 0)
  
              {
  
                  // safety check, if channels == 0, n will result in 0, and that will result in a nice devide exception
  
                  audioframe.duration = (unsigned int)(((__int64)audioframe.size * DVD_TIME_BASE) / n);
  
  
  
                  // increase audioclock to after the packet
  
                  m_audioClock += audioframe.duration;
  
              }
  
  
  
              //If we are asked to drop this packet, return a size of zero. then it won't be played
  
              //we currently still decode the audio.. this is needed since we still need to know it's 
  
              //duration to make sure clock is updated correctly.
  
              if( bDropPacket )
  
              {
  
                  result |= DECODE_FLAG_DROP;
  
              }
  
              return result;
  
          }
  
  
  
          // free the current packet
  
          if (pPacket)
  
          {
  
              CDVDDemuxUtils::FreeDemuxPacket(pPacket); //BRENT FIXME
  
              pPacket = NULL;
  
              pAudioPacket = NULL;
  
          }
  
  
  
          if (m_messageQueue.RecievedAbortRequest()) return DECODE_FLAG_ABORT;
  
  
  
          // read next packet and return -1 on error
  
          LeaveCriticalSection(&m_critCodecSection); //Leave here as this might stall a while
  
  
  
          CDVDMsg* pMsg;
  
          MsgQueueReturnCode ret = m_messageQueue.Get(&pMsg, INFINITE);
  
  
  
          EnterCriticalSection(&m_critCodecSection);
  
          if (MSGQ_IS_ERROR(ret) || ret == MSGQ_ABORT) return DECODE_FLAG_ABORT;
  
  
  
          if (pMsg->IsType(CDVDMsg::DEMUXER_PACKET))
  
          {
  
              CDVDMsgDemuxerPacket* pMsgDemuxerPacket = (CDVDMsgDemuxerPacket*)pMsg;
  
              pPacket = pMsgDemuxerPacket->GetPacket();
  
              pMsgDemuxerPacket->m_pPacket = NULL; // XXX, test
  
              pAudioPacket = pPacket;
  
              audio_pkt_data = pPacket->pData;
  
              audio_pkt_size = pPacket->iSize;
  
          }
  
          else
  
          {
  
              // other data is not used here, free if
  
              // msg itself will still be available
  
              pMsg->Release();
  
          }
  
  
  
          // if update the audio clock with the pts
  
          if (pMsg->IsType(CDVDMsg::DEMUXER_PACKET) || pMsg->IsType(CDVDMsg::GENERAL_RESYNC))
  
          {
  
              if (pMsg->IsType(CDVDMsg::GENERAL_RESYNC))
  
              { 
  
                  //player asked us to sync on this package
  
                  CDVDMsgGeneralResync* pMsgGeneralResync = (CDVDMsgGeneralResync*)pMsg;
  
                  result |= DECODE_FLAG_RESYNC;
  
                  m_audioClock = pMsgGeneralResync->GetPts();
  
              }
  
              else if (pPacket->pts != DVD_NOPTS_VALUE) // CDVDMsg::DEMUXER_PACKET, pPacket is already set above
  
              {
  
                  if (first_pkt_size == 0) 
  
                  { 
  
                      //first package
  
                      m_audioClock = pPacket->pts;        
  
                  }
  
                  else if (first_pkt_pts > pPacket->pts)
  
                  { 
  
                      //okey first packet in this continous stream, make sure we use the time here        
  
                      m_audioClock = pPacket->pts;        
  
                  }
  
                  else if((unsigned __int64)m_audioClock < pPacket->pts || (unsigned __int64)m_audioClock > pPacket->pts)
  
                  {
  
                      //crap, moved outsided correct pts
  
                      //Use pts from current packet, untill we find a better value for it.
  
                      //Should be ok after a couple of frames, as soon as it starts clean on a packet
  
                      m_audioClock = pPacket->pts;
  
                  }
  
                  else if(first_pkt_size == first_pkt_used)
  
                  {
  
                      //Nice starting up freshly on the start of a packet, use pts from it
  
                      m_audioClock = pPacket->pts;
  
                  }
  
              }
  
          }
  
          pMsg->Release();
  
      }
  
  }
  
  
  
  void CDVDPlayerAudio::SetSpeed(int speed)
  
  { 
  
      m_speed = speed;
  
  
  
      //if (m_speed == DVD_PLAYSPEED_PAUSE) m_dvdAudio.Pause(); //BRENT FIXME
  
      //else m_dvdAudio.Resume();
  
  }
  
  
  
  bool CDVDPlayerAudio::InitializeOutputDevice()
  
  {
  
      int iChannels = m_pAudioCodec->GetChannels();
  
      int iSampleRate = m_pAudioCodec->GetSampleRate();
  
      int iBitsPerSample = m_pAudioCodec->GetBitsPerSample();
  
      //bool bPasstrough = m_pAudioCodec->NeedPasstrough(); //BRENT
  
  
  
      if (iChannels == 0 || iSampleRate == 0 || iBitsPerSample == 0)
  
      {
  
          CLog::Log(LOGERROR, "Unable to create audio device, (iChannels == 0 || iSampleRate == 0 || iBitsPerSample == 0)");
  
          return false;
  
      }
  
  
  
      CLog::Log(LOGNOTICE, "Creating audio device with codec id: %i, channels: %i, sample rate: %i", m_codec, iChannels, iSampleRate);
  
      if (m_dvdAudio.Create(iChannels, iSampleRate, iBitsPerSample, /*bPasstrough*/0)) // always 16 bit with ffmpeg ? //BRENT Passthrough needed?
  
      {
  
          return true;
  
      }
  
  
  
      CLog::Log(LOGERROR, "Failed Creating audio device with codec id: %i, channels: %i, sample rate: %i", m_codec, iChannels, iSampleRate);
  
      return false;
  
  }

• ffmpeg - scaling and stacking 2 videos ?

  25 mai 2016, par Gambit2007

  I have 2 inputs and i want to scale, crop and put them on top of each other at the same time. My command should look something like this :

  ffmpeg -i input1 -i input2 -filter_complex crop=10000:5000:1000:0,scale=3840:1536 vstack output.mp4

  I know i need to use chaining (?) but i tried to look it up online and couldn’t really get it to work.

  So what would be the correct syntax the scale and crop both inputs and then put them vertically on top of each other while using ’-filter_complex’ only once ?

  Thanks !