Recherche avancée

Sur d’autres sites (14066)

• How to capture a timed screen recording on a Mac with ffmpeg

  28 septembre 2016, par wetjosh

  I’m on a Mac with MacOS Sierra installed. I’ve installed ffmpeg with homebrew. I list my devices via :

  ffmpeg -f avfoundation -list_devices true -i ""

  which returns :

  [AVFoundation input device @ 0x7fc2de40e840] AVFoundation video devices:
  
  [AVFoundation input device @ 0x7fc2de40e840] [0] FaceTime HD Camera
  
  [AVFoundation input device @ 0x7fc2de40e840] [1] Capture screen 0
  
  [AVFoundation input device @ 0x7fc2de40e840] AVFoundation audio devices:
  
  [AVFoundation input device @ 0x7fc2de40e840] [0] Built-in Microphone

  I don’t need audio so I start my 5 second screen recording via :

  ffmpeg -f avfoundation -t '5' -i '1' test.mov

  It creates an mov file in the working directory but doesn’t stop after 5 seconds. In fact, I can’t even stop the recording as it suggests by pressing ’q’. Ctl-C doesn’t work either, and I am left with force quitting via Activity Monitor. I’ve tried this same command but using device 0 (FaceTime camera) and it stops after 5 seconds.

  If someone can solve that riddle, my next question is how can I watch the newly created file in quicktime (I’m thinking I’ll need to encode or decode or something) because even the FaceTime video file would not open in QuickTime. It just says "The document could not be opened". It does, however, open with VLC.

  UPDATE : I’ve tried this on an older OS (Yosemite) and got the same results (thought it might be the new OS that broke it).

• youtube-dl python script postprocessing error : FFMPEG codecs aren't being recognized

  23 septembre 2016, par stackPusher

  My python script is trying to download youtube videos with youtube-dl.py. Works fine unless postprocessing is required. The code :

  import youtube_dl
  
  
  
  options = {
  
      'format':'bestaudio/best',
  
      'extractaudio':True,
  
      'audioformat':'mp3',
  
      'outtmpl':'%(id)s',     #name the file the ID of the video
  
      'noplaylist':True,
  
      'nocheckcertificate':True,
  
      'postprocessors': [{
  
          'key': 'FFmpegExtractAudio',
  
          'preferredcodec': 'mp3',
  
          'preferredquality': '192',
  
      }]
  
  }
  
  
  
  with youtube_dl.YoutubeDL(options) as ydl:
  
      ydl.download(['http://www.youtube.com/watch?v=BaW_jenozKc'])

  Below is the output I receive :

  I get a similar error if I try setting ’preferredcodec’ to ’opus’ or ’best’.
  I’m not sure if this is relevant, but I can run the command line counterpart fine :

  youtube-dl -o 'test2.%(ext)s' --extract-audio --audio-format mp3 --no-check-certificate https://www.youtube.com/watch?v=BaW_jenozKc

  I’ve gotten a few clues from the internet and other questions and from what i understand this is most likely an issue with my ffmpeg, which isn’t a python module right ? Here is my ffmpeg version and configuration :
  

  If the answer to my problem is to add some configuration setting to my ffmpeg please explain how i go about doing that.

• Converting YUV420P to BGRA ruins CPU

  21 septembre 2016, par Sun Dro

  I’m trying to write some media/stream player using chrome native-client (nacl) and ffmpeg. The problem is that nacl supports only BGRA and RGB formats and the stream I want to play is YUV420P.

  I solved this problem using FFMPEG

  int NCDecoder_DecodeVideo(NACLDecoder *pDec, AVFrame **ppFrame, AVPixelFormat pixFmt, int nWidth, int nHeight)
  
  {
  
      if (!pDec->bStarted) return 0;
  
  
  
      AVFrame *pFrame = *ppFrame;
  
      int nRetVal = avcodec_decode_video2(pDec->pCodecCtxV, pDec->pFrameV, &pDec->nGotFrameV, &pDec->vPkt);
  
      if(pDec->nGotFrameV)
  
      {
  
          if (!nWidth && !nHeight)
  
          {
  
              nWidth = pDec->pFrameV->width;
  
              nHeight = pDec->pFrameV->height;
  
          }
  
          else if (!nWidth || !nHeight)
  
          {
  
              if (nWidth) nHeight = nWidth * pDec->pFrameV->height / pDec->pFrameV->width;
  
              else nWidth = nHeight * pDec->pFrameV->width / pDec->pFrameV->height;
  
          }
  
  
  
          pFrame->width = nWidth;
  
          pFrame->height = nHeight;
  
          pFrame->format = pixFmt;
  
  
  
          pDec->pSwsCtx = sws_getCachedContext(pDec->pSwsCtx, pDec->pFrameV->width, pDec->pFrameV->height, 
  
              (AVPixelFormat)pDec->pFrameV->format, pFrame->width, pFrame->height, pixFmt, 0, NULL, NULL, NULL);
  
  
  
          sws_scale(pDec->pSwsCtx, (const uint8_t * const*)pDec->pFrameV->data, pDec->pFrameV->linesize, 0, 
  
              pDec->pFrameV->height, pFrame->data, pFrame->linesize);
  
  
  
          return 1;
  
      }
  
  
  
      return 0;
  
  } 

  Were argument pixFmt is AV_PIX_FMT_BGRA. Code works fine and I was happy until I try it on the full screen.

  When Im trying to scale image with full resolution, color format converting ruins my CPU memory and Its not even possible to watch video on the display.

  Is there any other fastest way to convert YUV to BGRA which not ruins my CPU usage or is there any way to display images using nacl-sdk with YUV color format ?