Recherche avancée

Sur d’autres sites (5701)

• lavf : make overlay_qsv work based on framesync

  3 avril 2018, par Ruiling Song

  lavf : make overlay_qsv work based on framesync
  The existing version which was cherry-picked from Libav does not work
  
  with FFmpeg framework, because ff_request_frame() was totally
  
  different between Libav (recursive) and FFmpeg (non-recursive).
  
  The existing overlay_qsv implementation depends on the recursive version
  
  of ff_request_frame to trigger immediate call to request_frame() on input pad.
  
  But this has been removed in FFmpeg since "lavfi : make request_frame() non-recursive."
  
  Now that we have handy framesync support in FFmpeg, so I make it work
  
  based on framesync. Some other fixing which is also needed to make
  
  overlay_qsv work are put in a separate patch.
  Signed-off-by : Ruiling Song <ruiling.song@intel.com>
  

  • [DH] libavfilter/Makefile
  • [DH] libavfilter/vf_overlay_qsv.c

• nodejs ffmpeg child-process stdio stream stops

  10 mars 2023, par BleedFeed

  I've been trying to make an internet radio stream using ffmpeg to encode the audio data in mp3

  


  function getAudioStream(url){

    return new Promise(async(resolve,reject)=>{

        const ytdlStream = await ytdl(url,{filter:'audioonly',quality:'highestaudio'});

        const ffmpegProcess = spawn('ffmpeg',[
        '-i','pipe:3',
        '-f','mp3',
        '-ar','44100',
        '-ac','2',
        '-b:a','128k',
        '-codec:a','libmp3lame',
        '-flush_packets', '1',
        'pipe:4'],{stdio:['ignore','ignore','pipe','pipe','pipe']});
        ytdlStream.pipe(ffmpegProcess.stdio[3]);
        ffmpegProcess.stderr.on('data',(chunk)=>{
            console.log('FFMPEG ERROR: ' + chunk.toString());
        })
        resolve(ffmpegProcess.stdio[4].pipe(new PassThrough({highWaterMark:4096})));
    });
}

async function setUpStream(fromQueue,client,shout){

    let readable;
    let videoDetails;

    if(fromQueue){
        readable = await getAudioStream(queue[0]);
        videoDetails = (await ytdl.getBasicInfo(queue[0])).videoDetails;
        queue.shift();
    }
    else{
        let song = songs[Math.floor(Math.random() * songs.length)];
        readable = await getAudioStream(song);
        videoDetails = (await ytdl.getBasicInfo(song)).videoDetails;
    }


    readable.on('data',async (chunk)=>{
        readable.pause();
        shout.send(chunk,chunk.length);
        await new Promise((resolve)=>{setTimeout(resolve,shout.delay())});
        readable.resume();
    });

    readable.on('end',()=>{
        readable.destroy();
        setUpStream(queue.length !==0,client,shout);
    });


    readable.on('error',(err)=>{
        console.log(err);
    });

}


  


  shout.send is from npm nodeshout library and it delivers the audio data to icecast server.
I get another ffmpeg audio stream once the current one ends. After 20 or 30 minutes ffmpeg logs stop and no further data comes from ffmpegStream

  


  FFMPEG ERROR: size=    2952kB time=00:03:08.89 bitrate= 128.0kbits/s speed=1.07x
FFMPEG ERROR: size=    3016kB time=00:03:13.00 bitrate= 128.0kbits/s speed=1.07x
FFMPEG ERROR: size=    3080kB time=00:03:17.10 bitrate= 128.0kbits/s speed=1.07x
FFMPEG ERROR: size=    3144kB time=00:03:21.17 bitrate= 128.0kbits/s speed=1.07x
FFMPEG ERROR: size=    3208kB time=00:03:25.27 bitrate= 128.0kbits/s speed=1.07x 


  


  these are the last lines printed at console. Maybe some sort of memory management problem with child process ? How can i find what causes this problem ?

  


• ffmpeg - convert movie AND show original input (as resized picture-in-picture, top-left corner) in the final output file

  2 octobre 2019, par raven

  this is my first post on this forum, so please be gentle in case I accidentally do trip over any forum rules that I would not know of yet :).

  I would like to apply some color-grading to underwater GoPro footage. To quicker gauge the effect of my color settings (trial-and-error, as of yet), would like to see the original input video stream as a PIP (e.g., scaled down 50%), in the top-left corner of the converted output movie.

  I have one input movie that is going to be color graded. The PIP should use the original as an input, just a scaled-down version of it.

  I would like to use ffmpeg’s -filter_complex option to do the PIP, but all examples I can find on "-filter_complex" would use two already existing movies. Instead, I would like to make the color-corrected stream an on-the-fly input to "-filter_complex", which then renders the PIP.

  Is that doable, all in one go ?

  Both the individual snippets below work fine, I now would like to combine these and skip the creation of an intermediate color-graded video. Your help combining these two steps into one single process is greatly appreciated !

  Thanks in advance,
  raven.

  [existing code snippets (M$ batch files)]
  
  
  
  ::declarations/defines::
  
  set "INPUT="
  
  set "TMP="
  
  set "OUTPUT="
  
  set "FFMPG="
  
  set "QU=9" :: quality settings
  
  
  
  set "CONV='"0 -1 0 -1 5 -1 0 -1 0:0 -1 0 -1 5 -1 0 -1 0:0 -1 0 -1 5 -1 
  
  0 -1 0:0 -1 0 -1 5 -1 0 -1 0'"" :: sharpening convolution filter
  
  
  
  ::color-grading part::
  
  %FFMPG% -i %INPUT% -vf convolution=%CONV%,colorbalance=rs=%rs%:gs=%gs%:bs=%bs%:rm=%rm%:gm=%gm%:bm=%bm%:rh=%rh%:gh=%gh%:bh=%bh% -q:v %QU% -codec:v mpeg4 %TMP%
  
  
  
  ::PIP part::
  
  %FFMPG% -i %INPUT% -i %TMP% -filter_complex "[1]scale=iw/3:ih/3 
  
  [pip]; [0][pip] overlay=main_w-overlay_w-10:main_h-overlay_h-10" -q:v 
  
  %QU% -codec:v mpeg4 %OUTPUT%
  
  
  
  [/existing code]