Recherche avancée

Sur d’autres sites (10757)

• AVFormatContext Transcode : Input or Output ?

  21 août 2016, par Victor.dMdB

  I have 2 functions running in 2 threads. One is written with opencv to parse images then write them to avformatcontext :

  VideoCapture vidSrc = cv::VideoCapture(vidURL);
  
  int vidWidth = vidSrc.get(CV_CAP_PROP_FRAME_WIDTH);
  
  int vidHeight = vidSrc.get(CV_CAP_PROP_FRAME_HEIGHT);
  
  AVRational vidFPS = {(int) vidSrc.get(CV_CAP_PROP_FPS), 1};
  
  AVFormatContext* av_fmt_ctx =  avformat_alloc_context();
  
  
  
  av_fmt_ctx->flags = AVFMT_NOFILE;
  
  cmd_data_ptr->video_data_conf.av_fmt_ctx = av_fmt_ctx;
  
  av_fmt_ctx->oformat = av_guess_format( NULL, props.vidURL.c_str(), NULL );
  
  AVCodec* vcodec = avcodec_find_encoder(av_fmt_ctx->oformat->video_codec);
  
  AVStream* vstrm = avformat_new_stream(av_fmt_ctx, vcodec);
  
  
  
  vstrm->codec = avcodec_alloc_context3(vcodec);
  
  vstrm->codec->width = vidWidth;
  
  vstrm->codec->height = vidHeight;
  
  vstrm->codec->pix_fmt = vcodec->pix_fmts[0];
  
  vstrm->codec->time_base = vstrm->time_base = av_inv_q(vidFPS);
  
  vstrm->r_frame_rate = vstrm->avg_frame_rate = vidFPS;
  
  if (av_fmt_ctx->oformat->flags & AVFMT_GLOBALHEADER)
  
    vstrm->codec->flags |= AV_CODEC_FLAG_GLOBAL_HEADER;
  
  avcodec_open2(vstrm->codec, vcodec, nullptr);
  
  
  
  int got_pkt = 0;
  
  AVFrame* frame = av_frame_alloc();
  
  std::vector framebuf(av_image_get_buffer_size(vstrm->codec->pix_fmt, vidWidth, vidHeight,32));
  
  avpicture_fill(reinterpret_cast(frame), framebuf.data(), vstrm->codec->pix_fmt, vidWidth, vidHeight);
  
  frame->width = vidWidth;
  
  frame->height = vidHeight;
  
  frame->format = static_cast<int>(vstrm->codec->pix_fmt);
  
  while(true){
  
      vidSrc >> in;
  
      cv::Mat towrite;
  
      const int stride[] = { static_cast<int>(towrite.step[0]) };
  
      sws_scale(swsctx, &amp;towrite.data, stride, 0, towrite.rows, frame->data, frame->linesize);
  
      frame->pts = frame_pts++;
  
      AVPacket pkt;
  
      pkt.data = nullptr;
  
      pkt.size = 0;
  
      pkt.stream_index = vstrm->id;
  
      av_init_packet(&amp;pkt);
  
      avcodec_encode_video2(vstrm->codec, &amp;pkt, end_of_stream ? nullptr : frame, &amp;got_pkt);
  
      if (got_pkt) {
  
            pkt.duration = 1;
  
            av_packet_rescale_ts(&amp;pkt, vstrm->codec->time_base, vstrm->time_base);
  
            av_write_frame(av_fmt_ctx, &amp;pkt);
  
       }
  
       av_packet_unref(&amp;pkt);
  
   }
  
  av_frame_free(&amp;frame);
  
  avcodec_close(vstrm->codec);
  
  </int></int>

  And another thread reads those frames

  while(1){
  
      AVPacket packet;
  
      av_read_frame(av_fmt_ctx, &packet);
  
      ...
  
  }

  The first function doesn’t seem to be working properly, as there is a segmentation error on av_write_frame or even withav_dump_format. Is it correct that the first function is created as an output context ? Or should I be setting up as an input context for the other thread ? I’m a bit confused. I’m still trying to wrap my head around ffmpeg.

• Input/output error sending RTMP to ffmpeg

  18 juin 2021, par greg

  I'm trying to use the "re-stream" feature of Vimeo (i.e. Vimeo send an rtmp stream to me).
When it tries to connect, I have this :

  


  Proto = rtmp, path = /src/live, app = src, fname = live
Window acknowledgement size = 2500000
Deleting stream...
Input/output error


  


  I cannot figure out what's wrong here...

  


  Any idea ?

  


  Thanks in advance

  


• ffmpeg pipe input and output in Python

  4 juin 2021, par MinasCham

  I want to use ffmpeg to read an RTSP stream, extract frames via a pipe, do some processing on them with Python and afterwards combine the processed frames via another pipe with the original audio. I'm using the subprocess module in Python to execute the ffmpeg command as well as read and write the frames from and to ffmpeg.

  


  Questions :

  


  

  1. Is it possible to pipe both stdin and stdout so as to extract the frames and then feed them back in after the processing ?

  


  2. Do i also have to pipe the audio separately and feed it with the processed frames or can i simply copy the audio stream when mapping the output ?