Recherche avancée

Sur d’autres sites (5335)

• FFMPEG, leaking memory

  8 juin 2016, par Nyaruko

  I use the following code to open a MP4 file. Each time the MP4 file is opened successfully by calling QVideoDecoder::openFile(QString filename).
  The initCode is called at the beginning of the program.

  However, each time I run the openFile(QString filename) function on the same file. There are some memory leak. What is wrong with my code ?

  void QVideoDecoder::InitVars()
  
  {
  
      //Input
  
      //-------------------
  
      ok = false;
  
      pFormatCtx = 0;
  
      pCodecCtx = 0;
  
      pCodec = 0;
  
      pFrame = 0;
  
      pFrameRGB = 0;
  
      buffer = 0;
  
      img_convert_ctx = 0;
  
      th = 0;
  
  }
  
  
  
  void QVideoDecoder::close()
  
  {
  
     // Free the RGB image
  
     if(buffer)
  
        delete [] buffer;
  
  
  
     // Free the YUV frame
  
     if(pFrame)
  
        av_free(pFrame);
  
  
  
     // Free the RGB frame
  
     if(pFrameRGB)
  
        av_free(pFrameRGB);
  
  
  
     // Close the codec
  
     if(pCodecCtx)
  
        avcodec_close(pCodecCtx);
  
  
  
     // Close the video file
  
     if(pFormatCtx)
  
         avformat_close_input(&pFormatCtx);
  
  
  
     // Close the timeout handler
  
     if (th)
  
         delete th;
  
  
  
     InitVars();
  
  }
  
  
  
  bool QVideoDecoder::initCodec()
  
  {
  
  
  
     ffmpeg::avcodec_register_all();
  
     ffmpeg::av_register_all();
  
     ffmpeg::avdevice_register_all();
  
     ffmpeg::avformat_network_init();
  
  
  
     printf("License: %s\n",ffmpeg::avformat_license());
  
     printf("AVCodec version %d\n",ffmpeg::avformat_version());
  
     printf("AVFormat configuration: %s\n",ffmpeg::avformat_configuration());
  
  
  
     return true;
  
  }
  
  
  
  bool QVideoDecoder::openFile(QString filename)
  
  {
  
      // Close the last video
  
      //----------------------------------------------
  
      close();
  
  
  
      // Initialize the params
  
      // ---------------------------------------------
  
      ffmpeg::AVInputFormat *iformat = NULL;
  
      ffmpeg::AVDictionary *opts = NULL;           
  
      ffmpeg::av_dict_set(&opts, "rtsp_transport", "tcp", NULL);
  
      ffmpeg::av_dict_set(&opts, "rtbufsize", "102400", NULL);
  
      ffmpeg::av_dict_set(&opts, "allowed_media_types", "video", NULL);
  
  
  
  
  
      //set the timeout handler for pFormatCtx
  
      //----------------------------------------------
  
      pFormatCtx = ffmpeg::avformat_alloc_context();
  
      th = new timeout_handler(3000000);
  
      pFormatCtx->interrupt_callback.opaque = (void*)th;
  
      pFormatCtx->interrupt_callback.callback = &timeout_handler::check_interrupt;
  
  
  
      // Open video file
  
      //---------------------------------------------
  
      th->reset(3000000);
  
      avformat_open_input(&pFormatCtx, filename.toStdString().c_str(), iformat, &opts) != 0)
  
      av_dict_free(&opts);
  
  
  
      // Retrieve stream information
  
      //---------------------------------------------
  
      th->reset(3000000);
  
      avformat_find_stream_info(pFormatCtx, NULL);
  
  
  
      // Dump information about file onto standard error
  
      //---------------------------------------------
  
      av_dump_format(pFormatCtx, 0, filename.toStdString().c_str(), false);
  
  
  
      // Find the first video stream
  
      //---------------------------------------------
  
      for (unsigned i = 0; i < pFormatCtx->nb_streams; i++)
  
      {
  
          if (pFormatCtx->streams[i]->codec->codec_type == ffmpeg::AVMEDIA_TYPE_VIDEO)
  
          {
  
              videoStream = i;
  
              break;
  
          }
  
      }
  
  
  
      // Get a pointer to the codec context for the video stream
  
      //--------------------------------------------
  
      pCodecCtx = pFormatCtx->streams[videoStream]->codec;
  
  
  
      // Find the decoder for the video stream
  
      //--------------------------------------------
  
      pCodec = avcodec_find_decoder(pCodecCtx->codec_id);
  
  
  
      // Set thread type to 0 seems decrease the amount of delay
  
      //--------------------------------------------
  
      pCodecCtx->thread_type = 0;
  
  // Open codec
  
  //--------------------------------------------
  
  avcodec_open2(pCodecCtx, pCodec, NULL);
  
  
  
  // Allocate video frame
  
  //--------------------------------------------
  
  pFrame = ffmpeg::avcodec_alloc_frame();
  
  
  
  // Allocate an AVFrame structure
  
  //--------------------------------------------
  
  pFrameRGB = ffmpeg::avcodec_alloc_frame();
  
  
  
  // Determine required buffer size and allocate buffer
  
  //--------------------------------------------
  
  numBytes = ffmpeg::avpicture_get_size(ffmpeg::PIX_FMT_RGB24, pCodecCtx->width, pCodecCtx->height);
  
  buffer = new uint8_t[numBytes];
  
  
  
  // Assign appropriate parts of buffer to image planes in pFrameRGB
  
  //--------------------------------------------
  
  avpicture_fill((ffmpeg::AVPicture *)pFrameRGB, buffer, ffmpeg::PIX_FMT_RGB24,
  
      pCodecCtx->width, pCodecCtx->height);
  
  
  
  return true;
  
  }

• libopusenc : Add channel mapping family argument

  15 juin 2016, par Michael Graczyk

  libopusenc : Add channel mapping family argument
  The default value of -1 indicates that ffmpeg should determine the channel
  
  mapping automatically, which was the behavior before this commit.
  Unless the -mapping_family argument is provided, behavior is unchanged.
  Signed-off-by : Michael Niedermayer <michael@niedermayer.cc>
  

  • [DH] doc/encoders.texi
  • [DH] libavcodec/libopusenc.c

• Could not open encoder using ffmpeg C APi for MOV format

  22 juin 2016, par lupod

  Context

  I am writing a program for doing some video processing on an input file. I wrote two classes for handling the "reading/writing frames" part that essentially wrap the functions of ffmpeg. These classes may be instantiated by providing an input and output file name, and in their constructor I initialize everything that is needed (or at least I hope so).

  This are the two routines that are called inside the constructors :

  // InputVideoHandler.cpp
  
  void InputVideoHandler::init(char* name) {
  
    streamIndex = -1;
  
    int numStreams;
  
  
  
    if (avformat_open_input(&formatCtx, name, NULL, NULL) != 0)
  
      throw std::exception("Invalid input file name.");
  
  
  
    if (avformat_find_stream_info(formatCtx, NULL)<0)
  
      throw std::exception("Could not find stream information.");
  
  
  
    numStreams = formatCtx->nb_streams;
  
  
  
    if (numStreams < 0)
  
      throw std::exception("No streams in input video file.");
  
  
  
    for (int i = 0; i < numStreams; i++) {
  
      if (formatCtx->streams[i]->codec->codec_type == AVMEDIA_TYPE_VIDEO) {
  
        streamIndex = i;
  
        break;
  
      }
  
    }
  
  
  
    if (streamIndex < 0)
  
      throw std::exception("No video stream in input video file.");
  
  
  
    // find decoder using id
  
    codec = avcodec_find_decoder(formatCtx->streams[streamIndex]->codec->codec_id);
  
    if (codec == nullptr)
  
      throw std::exception("Could not find suitable decoder for input file.");
  
  
  
    // copy context from input stream
  
    codecCtx = avcodec_alloc_context3(codec);
  
    if (avcodec_copy_context(codecCtx, formatCtx->streams[streamIndex]->codec) != 0)
  
      throw std::exception("Could not copy codec context from input stream.");
  
  
  
    if (avcodec_open2(codecCtx, codec, NULL) < 0)
  
      throw std::exception("Could not open decoder.");
  
  
  
    codecCtx->refcounted_frames = 1;
  
  }
  
  
  
  // OutputVideoBuilder.cpp
  
  void OutputVideoBuilder::init(char* name, AVCodecContext* inputCtx) {
  
    if (avformat_alloc_output_context2(&formatCtx, NULL, NULL, name) < 0)
  
      throw std::exception("Could not determine file extension from provided name.");
  
  
  
    codec = avcodec_find_encoder(inputCtx->codec_id);
  
    if (codec == nullptr) {
  
      throw std::exception("Could not find suitable encoder.");
  
    }
  
  
  
    codecCtx = avcodec_alloc_context3(codec);
  
    if (avcodec_copy_context(codecCtx, inputCtx) < 0)
  
      throw std::exception("Could not copy output codec context from input");
  
  
  
    codecCtx->time_base = inputCtx->time_base;
  
  
  
    if (avcodec_open2(codecCtx, codec, NULL) < 0)
  
      throw std::exception("Could not open encoder.");
  
  
  
    stream = avformat_new_stream(formatCtx, codec);
  
    if (stream == nullptr) {
  
      throw std::exception("Could not allocate stream.");
  
    }
  
  
  
    stream->id = formatCtx->nb_streams - 1;
  
    stream->codec = codecCtx;
  
    stream->time_base = codecCtx->time_base;
  
  
  
    av_dump_format(formatCtx, 0, name, 1);
  
    if (!(formatCtx->oformat->flags & AVFMT_NOFILE)) {
  
      if (avio_open(&formatCtx->pb, name, AVIO_FLAG_WRITE) < 0) {
  
        throw std::exception("Could not open output file.");
  
      }
  
    }
  
  
  
    if (avformat_write_header(formatCtx, NULL) < 0) {
  
      throw std::exception("Error occurred when opening output file.");
  
    }
  
  
  
  }

  As you see, for the Init function of the output handler I require that an AVCodecContext must be provided. In my code, I pass to the constructor the AVCodecContext that is stored in the input handler and that was previously created.

  Question :

  The two functions work fine when I test my program with some video formats, like .mpg or .avi. When I try to process .mov/.mp4 files, however, my code throws the exception that I labeled "Could not open encoder." in OutputVideoBuilder::Init(). Why is this happening ? I read from the General documentation of ffmpeg that that format should be supported as well by ffmpeg. I am assuming that I am doing something wrong in my code, which I do not completely understand because it was created by trying to adapt the tutorials of the documentation to my specific case. Also for this reason, any comments on things that are useless or things that are missing will be greatly appreciated.