Recherche avancée

Sur d’autres sites (9322)

• Generating MPEG DASH videos which are compilant with HbbTV 1.5 standard

  6 mai 2014, par Kai Mysliwiec

  I am looking for a the command line options for ffmpeg, DASHEncoder and MP4Box to generate HbbTV 1.5 compilant MPEG DASH videos.

  See http://www.hbbtv.org/pages/about_hbbtv/specification.php for more information on the HbbTV 1.5 standard and its DASH profile which is used there.

• Is there any way to compile ffmpeg library on Windows ?

  20 mars 2013, par AB1209

  Hi I have been working for weeks to compile ffmpeg library on Windows platform.But I found no way so far.Is it there any way to compile ffmpeg library on Windows 7.

  I am using Cygwin on Windows.
  Whenever I try to do ndk-build I get this message.

  Android NDK: ERROR:jni/yuv2rgb/Android.mk:ffmpeg-prebuilt: LOCAL_SRC_FILES points to a missing file
  
  Android NDK: Check that jni/ffmpeg-build/armeabi/libffmpeg.so exists  or that its path is correct
  
  /cygdrive/D/AndroidNDK/android-ndk-r8c/build/core/prebuilt-library.mk:43: *** Android NDK: Aborting    .  Stop.

  & while searching for solution of this error I found that I will have to compile the libraries first.For compiling by using ./build_android.sh. I got this error message.

  ./build_android.sh
  
  ./build_android.sh: line 72: ./configure: No such file or directory

  Please please guide me what should I do.

• (not) Decoding h264 nals using ffmpeg

  28 janvier 2013, par Zetlb

  Im trying to decode h264 file ( h264 bitstream, without containter ) I have generated and I have a problem.
  I think h264 file is correct because I'm playing it using ffplay.
  The problem is that in the third frame I get an assert a exit.
  I read the nals and create packets using this function ( probably here is the error ) :

  Here is my code :

  int encode_nals( AVPacket *pkt, x264_nal_t *nals, int nnal, x264_picture_t* pPicture )
  
  {
  
      uint8_t* sei = NULL;
  
      int sei_size = 0;
  
  
  
      // first search NAL with SEI info to put it first
  
      //
  
      for (int i = 0; i < nnal; i++)
  
      {
  
          if ( nals[i].i_type == NAL_SEI )
  
          {
  
              sei_size = nals[i].i_payload;
  
              sei = (uint8_t*) av_malloc( sei_size );
  
          }
  
      }
  
  
  
      uint8_t *p;
  
      int i, size = sei_size, ret;
  
  
  
      if (!nnal)
  
          return 0;
  
  
  
      // obtain the size of the new packet
  
      //
  
      for (i = 0; i < nnal; i++)
  
          size += nals[i].i_payload;
  
  
  
      pkt->data = ( uint8_t*) av_malloc( size );
  
      p = pkt->data;
  
  
  
      /* Write the SEI at the beginning of the packet */
  
      if (sei_size > 0 && nnal > 0) 
  
      {
  
          memcpy(p, sei, sei_size);
  
          p += sei_size;
  
          sei_size = 0;
  
          av_free( sei );
  
      }
  
  
  
      // write remaining NALs ( except NAL_SEI written first )
  
      //
  
      for (i = 0; i < nnal; i++)
  
      {
  
          if ( nals[i].i_type != NAL_SEI )
  
          {
  
              memcpy(p, nals[i].p_payload, nals[i].i_payload);
  
              p += nals[i].i_payload;
  
          }
  
      }
  
      // Complete packet info
  
      //
  
      pkt->pts = pPicture->i_pts;
  
      pkt->dts = pPicture->i_dts;
  
      pkt->size = size;
  
      pkt->flags |= AV_PKT_FLAG_KEY * pPicture->b_keyframe;
  
  
  
      return 1;
  
  }

  Also I have tried to write all the nal's p_payload in the packet with a single memcpy, something like this :

  int encode_nals( AVPacket *pkt, x264_nal_t *nals, int nnal, x264_picture_t* pPicture )
  
  {
  
  
  
      uint8_t *p;
  
      int i, size = 0, ret;
  
  
  
      if (!nnal)
  
          return 0;
  
  
  
      // obtain the size of the new packet
  
      //
  
      for (i = 0; i < nnal; i++)
  
          size += nals[i].i_payload;
  
  
  
      pkt->data = ( uint8_t*) av_malloc( size );
  
      p = pkt->data;
  
      memcpy( pkt->data, nals[0].p_payload, size );    
  
  
  
      // Complete packet info
  
      //
  
      pkt->pts = pPicture->i_pts;
  
      pkt->dts = pPicture->i_dts;
  
      pkt->size = size;
  
      pkt->flags |= AV_PKT_FLAG_KEY * pPicture->b_keyframe;
  
  
  
      return 1;
  
  }

  Both cases obtain the same error.

  After having the packets, I use a code like this to decode ( I have the packets in a vector ) :

  avcodec_init();  
  
  avcodec_register_all();  
  
  AVCodec* pCodec;  
  
  pCodec=avcodec_find_decoder(CODEC_ID_H264);  
  
  AVCodecContext* pCodecContext;  
  
  pCodecContext=avcodec_alloc_context();  
  
  avcodec_open(pCodecContext,pCodec);  
  
  pCodecContext->extradata = NULL;
  
  
  
  pCodecContext->flags    |= CODEC_FLAG_4MV;
  
  pCodecContext->flags     |= CODEC_FLAG_PART;
  
  pCodecContext->flags     |=CODEC_FLAG_TRUNCATED;
  
  pCodecContext->flags2  |=  CODEC_FLAG2_CHUNKS;
  
  pCodecContext->pix_fmt   =   PIX_FMT_YUV420P  /*It can be RGB also*/;
  
  pCodecContext->skip_frame = AVDISCARD_DEFAULT;
  
  pCodecContext->error_concealment = 3;
  
  pCodecContext->error_recognition = 1;
  
  pCodecContext->skip_loop_filter = AVDISCARD_DEFAULT;
  
  pCodecContext->workaround_bugs = 1;
  
  pCodecContext->codec_type = AVMEDIA_TYPE_VIDEO;
  
  pCodecContext->codec_id = CODEC_ID_H264;
  
  
  
  int iNumFrames = 0;
  
  unsigned int nNumPackets = m_vPackets.size();
  
  int frameFinished = 0;
  
  for (   auto it = m_vPackets.begin();
  
          it != m_vPackets.end();
  
          it++ )
  
  {
  
      frameFinished = 0;
  
      AVFrame* pFrame;
  
      pFrame = avcodec_alloc_frame();
  
      AVPacket* pPacket = *it;
  
      int iReturn = avcodec_decode_video2( pCodecContext, pFrame, &frameFinished, pPacket );
  
      if ( frameFinished )
  
      {
  
          DrawFrame( pFrame ); // this function is not implemented now.
  
          iNumFrames++;
  
      }
  
      av_free_packet( pPacket );
  
  }

  The result of the execution is that on the first two iterations avcodec_decode_video2 returns a positive value, frameFinished is filled with a positive value and pFrame is filled ( I didn't show the frame on screen yet, is my next step ), and the third iteration, the avcodec_decode_video2 causes an assert and abort.
  The assert is on file mpegvideo.c, on the function :

  static void update_noise_reduction(MpegEncContext *s)

  The assert is :

  assert(s->last_picture_ptr==NULL || s->out_format != FMT_H264 || s->codec_id == CODEC_ID_SVQ3);

  Can anybody help me please ? what's wrong ?

  EDIT : The frames obtained on the first two calls of avcodec_decode_video2 aren't correct.

  Thanks
  Zetlb