Recherche avancée

Sur d’autres sites (8377)

• ffmpeg : add per-stream input option drop_changed

  15 mars, par Gyan Doshi

  ffmpeg : add per-stream input option drop_changed
  This is a replacement in ffmpeg for the deprecated avcodec flag AV_CODEC_FLAG_DROPCHANGED.
  This option is meant to be used when the filtergraph should not be
  
  reinited upon input parameter changes as that leads to loss of state
  
  in the filtergraph potentially leading to broken or aborted output,
  
  e.g. inserting of silence with first_pts specified in aresample.
  Generally useful to avoid corrupted yet decodable packets in live
  
  streaming inputs.
  This option when enabled takes precedence over reinit_filters
  

  • [DH] doc/ffmpeg.texi
  • [DH] fftools/ffmpeg.h
  • [DH] fftools/ffmpeg_demux.c
  • [DH] fftools/ffmpeg_filter.c
  • [DH] fftools/ffmpeg_opt.c

• FFMPEG merge continous input stream with intermittent input stream without waiting

  18 mai 2024, par gian848396

  My goal is to take two audio inputs and merge them. The below works for this, however, I want a continuous output stream even when only one of the two inputs are being streamed into FFmpeg. The below only produces an output when the stdin pipe input is receiving audio.

  


  Here I have an FFmpeg child process where one input is a stream from the device's microphone, the other comes from wav audio buffers sent to the server in clips by a client.

  


  @Injectable()
export class AppService {
  ffmpegProcess: ChildProcessWithoutNullStreams;

  constructor() {
    this.start();
  }

  start() {
    this.ffmpegProcess = spawn(ffmpegPath, [
      '-f', 'avfoundation', // mac os media devices
      '-i', ':1', // mac os microphone input
      '-f', 'wav',
      '-i', 'pipe:', // stdin input source
      '-codec:a', 'aac',
      '-b:a', '128k',
      // '-ab', '32k',
      '-f', 'hls',
      '-hls_time', '4', // Segment duration (in seconds)
      '-hls_list_size', '3', // Number of HLS segments to keep in playlist
      '-hls_flags', 'delete_segments', // Automatically delete old segments
      '-filter_complex', 'amerge=inputs=2',
      'public/output.m3u8' // HLS playlist file name
    ]);
  }
}



  


  Here I'm piping wav audio buffers posted to the server and streamed as the stdin source to the FFMepg process, to be merged into the output.

  


  @Controller()
export class AppController {
  constructor(
    private readonly appService: AppService,
  ) { }

  @Post('broadcast')
  @UseInterceptors(FileInterceptor('audio_data'))
  uploadFile(@UploadedFile() file: Express.Multer.File) {
    console.log(file);
    Readable.from(file.buffer).pipe(this.appService.ffmpegProcess.stdin);
  }
}


  


  I've considered trying to pass in a null stream for the stdin when no submitted wav audio is present though I'm spinning my wheels. How can I produce a continuous output while only one of the two inputs are streaming, and while both are streaming ?

  


• avfilter/vf_scale : add optional "ref" input

  24 avril 2024, par Niklas Haas

  avfilter/vf_scale : add optional "ref" input
  This is automatically enabled if the width/height expressions reference
  
  any ref_* variable. This will ultimately serve as a more principled
  
  replacement for the fundamentally broken scale2ref.
  See-Also : https://trac.ffmpeg.org/ticket/10795
  

  • [DH] Changelog
  • [DH] doc/filters.texi
  • [DH] libavfilter/vf_scale.c