Nodejs performs live video streaming based on ffmpeg

This time I will bring you nodejs based on ffmpeg for video push and live broadcast. What are the precautions for nodejs based on ffmpeg for video push and live broadcast. The following is a practical case, let's take a look.

With ffmpeg as the core, it packages a client software that receives transcoding in the LAN and pushes it to the Internet. This article only uses the basic functions of ffmpeg, including streaming, transcoding, streaming and simple playback settings.

work process

1. Pull the remote video stream, the video stream format is rtsp

2. Convert to the common playback format rtmp

3. Push to the playback port rtmp://your push End address, users can directly play the content after using the playback software to connect to this address

Required tools and software

1. ffmpeg command line tool official website link, the advantages of choosing it are:

1. Free

2. No installation required, greatly reducing user operation complexity

3. Command line startup call

2. The nodejs version number is v6.11.3. (Electron is used in the actual project, but if there is no requirement to package it into a client, nodejs can run normally)

3. The tsc version number is v2.6.1. The project uses TypeScript as the main writing language, and there is no problem if youuse JavaScript.

If you use tsc, please use version 2.0 or above. The built-in @type tool will greatly improve coding efficiency

4. The fluent-ffmpeg version number is v2.1.2. This nodejs package encapsulates the command line calling part of ffmpeg, which enhances the readability of the code. If you are familiar with the ffmpeg command line manual, you do not need to use this package.

  npm install --save fluent-ffmpeg
  //使用js编码的用户,可以忽略下条命令
  npm install --save @types/fluent-ffmpeg

VLC playback software. Used to monitor whether streaming, transcoding, and playback are normal. Official website link

Implementation code

  const ffmpegPath = "./dist/ffmpegProgram/bin/ffmpeg.exe";
  const ffprobePath = "./dist/ffmpegProgram/bin/ffprobe.exe";
  const flvtoolPath = "./dist/ffmpegProgram/bin/ffplay.exe";
  export function startPushVideo():void{
    getCommands().then((commands:ffmpegPaths[])=>{
      for(let key in commands){
        let command = commands[key];
        //设置输入流地址
        let ffCommand = ffmpeg(command.inputPath)
        //设置输出流地址
        .output(command.outputPath)
        //因需要打包客户端软件,故而将ffmpeg打包进软件中
        //需设置各应用程序的对应路径
        //若仅在本机使用,可以跳过该步骤
        //设置环境变量,添加 PATH 即可
        .setFfmpegPath(ffmpegPath)
        .setFfprobePath(ffprobePath)
        .setFlvtoolPath(flvtoolPath)
        //为保证灵活性,非必须参数采用配置文件读取模式
        .size(command.size);
        for(let key in command.args){
          ffCommand.outputOption(command.args[key]);
        }
        ffCommand.on("start",(commandLine)=>{
          //commandLine 为实际上调用的命令行命令,拼接逻辑为
          //您的ffmpeg所在路径 -i inputOptions 您的拉流协议和路径 outputOptions 推送流协议和地址
          //ffmpeg -i "rtsp://yourPullUrl" -f flv -r 25 -s 640x480 -an "rtmp://yourPushUrl"
          console.log('[' + showTime() + '] Vedio is Pushing !');
          console.log('[' + showTime() + '] Spawned Ffmpeg with command !');
          console.log('[' + showTime() + '] Command: ' + commandLine);
        })
        .on('error', function(err, stdout, stderr) {
          console.log('error: ' + err.message);
          console.log('stdout: ' + stdout);
          console.log('stderr: ' + stderr);
        })
        .on('end', function() {
          console.log('[' + showTime() + '] Vedio Pushing is Finished !');
        })
        .run();
      }
    },(error)=>{
      console.log('error: ' + error);
    })
  }

Summary

The command obtained by listening to "start" can also be called through exec(yourCommandLine), but ffmpeg cannot be controlled at this time. operating results. After the program ends, the ffmpeg process is still running until the stream reports an error or the process is manually stopped. It’s not clear why fluent-ffmpeg You can notify the third-party process to close after the ontology process ends. The guess is to cut off the process through command line input, if only through ChildProcess.kill() It is impossible to close third-party processes.

When running on an I5 8G machine, single-stream push has occupied about 35% of the CPU. Multi-stream push requires other solutions. ​​​​​​

I believe you have mastered the method after reading the case in this article. For more exciting information, please pay attention to other related articles on the php Chinese website!

Recommended reading:

Detailed steps for setting global styles in Vue2.0

Use NodeJS to transcode videos

What are the steps required to implement a complete Angular4 FormText component

The above is the detailed content of Nodejs performs live video streaming based on ffmpeg. For more information, please follow other related articles on the PHP Chinese website!