Recherche avancée

Sur d’autres sites (7099)

• ffmpeg watermarking processing is very slow

  25 février 2014, par Hitesh Rohilla

  I am working on a video processing project and using ffmpeg for watermarking. I achieve exactly what i want but the problem is that the process is very very slow.

  I am using a Intel Smart 2nd gen family Core-i5 Processor with 4GB RAM on a 32 bit operating system Microsoft Windows-8 on a 64-bit CPU.

  I tried watermarking a video [mp4] of length 1:30 Min. size of file is 1.5GB

  Whole process accomplish in 3 Hrs to watermark my video file and what i noticed while process is that ffmpeg was processing 28 frames per sec first and then later it slow down up to 20 frames per second. a normal human watch video with frame rate of 30 frame per second and process was even slower then this that's why it took more time [3 Hrs] then the actual length of video itself [1:33]

  What i think to make process efficient is to use ffmpeg Watermarking Source Code and modify it somewhat...

  I want to ask if someone have achieve fast watermarking before by any other way or have modified this code to achieve faster process in order to save my time...

• create a timelapse video using MediaRecorder API ( and ffmpeg ? )

  24 août 2022, par The Blind Hawk

  Summary

  


  I have a version of my code already working on Chrome and Edge (Mac Windows and Android), but I need some fixes for it to work on IOS (Safari/Chrome).
  
My objective is to record around 25 minutes and download a timelapse version of the recording.
  
final product requirements :

  


  speed: 3fps
length: ~25s

(I need to record one frame every 20 seconds for 25 mins)


  


  this.secondStream settings :

  


  this.secondStream = await navigator.mediaDevices.getUserMedia({
    audio: false,
    video: {width: 430, height: 430, facingMode: "user"}
});


  


  My code for IOS so far :

  


          startIOSVideoRecording: function() {
            console.log("setting up recorder");
            var self = this;
            this.data = [];

            if (MediaRecorder.isTypeSupported('video/mp4')) {
                // IOS does not support webm, so I will be using mp4
                var options = {mimeType: 'video/mp4', videoBitsPerSecond : 1000000};
            } else {
                console.log("ERROR: mp4 is not supported, trying to default to webm");
                var options = {mimeType: 'video/webm'};
            }
            console.log("options settings:");
            console.log(options);

            this.recorder = new MediaRecorder(this.secondStream, options);

            this.recorder.ondataavailable = function(evt) {
                if (evt.data && evt.data.size > 0) {
                    self.data.push(evt.data);
                    console.log('chunk size: ' + evt.data.size);
                }
            }

            this.recorder.onstop = function(evt) {
                console.log('recorder stopping');
                var blob = new Blob(self.data, {type: "video/mp4"});
                self.download(blob, "mp4");
                self.sendMail(videoBlob);
            }

            console.log("finished setup, starting")
            this.recorder.start(1200);

            function sleep(ms) { return new Promise(resolve => setTimeout(resolve, ms));}

            async function looper() {
                // I am trying to pick one second every 20 more or less
                await sleep(500);
                self.recorder.pause();
                await sleep(18000);
                self.recorder.resume();
                looper();
            }
            looper();
        },


  


  Issues

  


  Only one call to getUserMedia()

  


  I am already using this.secondstream elsewhere, and I need the settings to stay as they are for the other functionality.
  
On Chrome and Edge, I could just call getUserMedia() again with different settings, and the issue would be solved, but on IOS calling getUserMedia() a second time kills the first stream.
  
The settings that I was planning to use (works for Chrome and Edge) :

  


  navigator.mediaDevices.getUserMedia({
    audio: false,
    video: { 
        width: 360, height: 240, facingMode: "user", 
        frameRate: { min:0, ideal: 0.05, max:0.1 } 
    },
}


  


  The timelapse library I am using does not support mp4 (ffmpeg as alternative ?)

  


  I am forced to use mp4 on IOS apparently, but this does not allow me to use the library I was relying on so I need an alternative.
  
I am thinking of using ffmpeg but cannot find any documentation to make it interact with the blob before the download.
  
I do not want to edit the video after downloading it, but I want to be able to download the already edited version, so no terminal commands.

  


  MediaRecorder pause and resume are not ideal

  


  On Chrome and Edge I would keep one frame every 20 seconds by setting the frameRate to 0.05, but this does not seem to work on IOS for two reasons.
  
First one is related to the first issue of not being able to change the settings of getUserMedia() without destroying the initial stream in the first place.
  
And even after changing the settings, It seems that setting the frame rate below 1 is not supported on IOS. Maybe I wrote something else wrong, but I was not able to open the downloaded file.
  
Therefore I tried relying on pausing and resuming the MediaRecorder, but this brings forth another two issues :
  
I am currently saving 1 second every 20 seconds and not 1 frame every 20 seconds, and I cannot find any workarounds.
  
Pause and Resume take a little bit of time, making the code unreliable, as I sometimes pick 2/20 seconds instead of 1/20, and I have no reliability that the loop is actually running every 20 seconds (might be 18 might be 25).

  


  My working code for other platforms

  


  This is my code for the other platforms, hope it helps !
  
Quick note : you will need to give it a bit of time between setup and start.
  
The timelapse library is here

  


  
        setupVideoRecording: function() {
            let video  = { 
                width: 360, height: 240, facingMode: "user", 
                frameRate: { min:0, ideal: 0.05, max:0.1 } 
            };
            navigator.mediaDevices.getUserMedia({
                audio: false,
                video: video,
            }).then((stream) => {
                // this is a video element
                const recVideo = document.getElementById('self-recorder');
                recVideo.muted = true;
                recVideo.autoplay = true;
                recVideo.srcObject = stream;
                recVideo.play();
            });
        },

        startVideoRecording: function() {
            console.log("setting up recorder");
            var self = this;
            this.data = [];

            var video = document.getElementById('self-recorder');

            if (MediaRecorder.isTypeSupported('video/webm; codecs=vp9')) {
                var options = {mimeType: 'video/webm; codecs=vp9'};
            } else  if (MediaRecorder.isTypeSupported('video/webm')) {
                var options = {mimeType: 'video/webm'};
            }
            console.log("options settings:");
            console.log(options);

            this.recorder = new MediaRecorder(video.captureStream(), options);

            this.recorder.ondataavailable = function(evt) {
                self.data.push(evt.data);
                console.log('chunk size: ' + evt.data.size);
            }

            this.recorder.onstop = function(evt) {
                console.log('recorder stopping');
                timelapse(self.data, 3, function(blob) {
                    self.download(blob, "webm");
                });
            }

            console.log("finished setup, starting");
            this.recorder.start(40000);
        }


  


• Limit the cpu usage when converting youtube videos to mp3 using youtube-dl

  30 novembre 2015, par Kamal

  I am using youtube-dl & avconv with php on nginx server to convert youtube video into mp3. Using htop I noticed the usage of cpus peak at 100% when couple of avconv processes are running at the same time, and once my server crushed down because of that.

  The first solution that I encountered cpulimit but after some research,I’ve found cpu limit is not smart enough to handle multiple simultaneous instances.

  From this answer I learned that I can limit the number of cpu threads using -threads option.
  My basic command : youtube-dl --extract-audio --audio-format mp3 <video url="url"></video>

  I wonder if I can add -threads option to the youtube-dl command, I am not sure if that is possible.

  I have 2 cores cpu server, I am thinking of upgrading to 4 cores cpu and limit the avconv to use just 2 cores, what do you think ? Is that the best way to go (using thread option) ?