Recherche avancée

Sur d’autres sites (9247)

• Create video file by mixing video and audio byte arrays FFmpeg & C++

  20 janvier 2021, par Sergey Zinovev

  I capture audio and video.

  


  Video is captured by using Desktop Duplication API and, as a result, I get Textures 2D.
These Textures 2D are char arrays.

  


  m_immediateContext->CopyResource(currTexture, m_acquiredDesktopImage.Get());&#xA;&#xA;D3D11_MAPPED_SUBRESOURCE* resource = new D3D11_MAPPED_SUBRESOURCE;&#xA;UINT subresource = D3D11CalcSubresource(0, 0, 0);&#xA;&#xA;m_immediateContext->Map(currTexture, subresource, D3D11_MAP_READ_WRITE, 0, resource);&#xA;&#xA;uchar * buffer = new uchar[(m_desc.Height * m_desc.Width * 4)];&#xA;const uchar * mappedData = static_cast<uchar>(resource->pData);&#xA;memcpy(buffer, mappedData, m_desc.Height * m_desc.Width * 4);&#xA;</uchar>

  


  Then the Textures 2D convert in cv::Mat and write Video using OpenCV.

  


  Audio captured by using WASAPI and, as a result, I get samples.

  


  BYTE * buffer = new BYTE[(numFramesAvailable * pwfx->nBlockAlign)];
memcpy(buffer, pData, numFramesAvailable * pwfx->nBlockAlign);


  


  These samples are byte arrays then write in WAV file.

  


  As a result, I get two files - video and audio, which merged by using FFmpeg.

  


  I want to skip the creation of video and audio files and promptly create one file compose of two strims (video and audio) from raw data.

  


  In order to make it I need help with FFmpeg code.
Specifically, in a way of creating and setting the correct output context and output streams, and how to encode raw data.

  


  I've already learned doc/examples FFmpeg, but still can't make the working code. So, I really need your help guys.

  


• Problem with processing FFMPEG video , the output video is totally black in first 3 seconds

  13 janvier 2021, par Trần Minh Tuấn

  I want to make a slideshow by using concat demuxer like this link :
https://trac.ffmpeg.org/wiki/Slideshow

  


  Here is the textfile preinputFiles.txt :

  


  file 'path a'
duration 2
file 'path b'
duration 2
file 'path c'
duration 2
file 'path c'


  


  and my command array is :

  


  String[] cmd = new String[]{
            "-f","concat","-i",
            textFile,
            "-vsync", "vfr", "-pix_fmt", "yuv420p",
            dest.getAbsolutePath()};


  


  the video has 3 seconds that it is totally black and run from 3s to 9s even the gallery shows that it is a 6 seconds long video .

  


  Thanks for the help !

  


• ffmpeg encoding a video with time_base Not equal to framerate does not work in HML5 video players

  1er juillet 2019, par Gilgamesh22

  I have a time_base of 90000 with a frame rate of 30. I can generate a h264 video and have it work in VLC but this video does not work in the browser player. If I change the time_base to 30 It works fine.

  Note : I am changing the frame->pts appropriately to match the time_base.
  Note : Video does not have audio stream

  //header.h
  
  AVCodecContext *cctx;
  
  AVStream* stream;

  Here is the non working example code

  //source.cpp
  
  stream->time_base = { 1, 90000 };
  
  stream->r_frame_rate = { fps, 1 };
  
  stream->avg_frame_rate = { fps, 1 };
  
  
  
  cctx->codec_id = codecId;
  
  cctx->time_base = { 1 ,  90000 };
  
  cctx->framerate = { fps, 1 };
  
  
  
  // ...... 
  
  // add frame code later on timestamp are in millisecond
  
  frame->pts = (timestamp - startTimeStamp)* 90;

  Here is the working example code

  //source.cpp
  
  stream->time_base = { 1, fps};
  
  stream->r_frame_rate = { fps, 1 };
  
  stream->avg_frame_rate = { fps, 1 };
  
  
  
  cctx->codec_id = codecId;
  
  cctx->time_base = { 1 ,  fps};
  
  cctx->framerate = { fps, 1 };
  
  
  
  // ...... 
  
  //  add frame code timestamp are in millisecond
  
  frame->pts = (timestamp - startTimeStamp)/(1000/fps);

  Any ideas on why the second example works and the first does not in the HTML5 video player.