Recherche avancée

Sur d’autres sites (3960)

• Inconsistent behavior when placing adjacent video overlays

  24 juillet 2023, par abingham

  This script uses ffmpeg to overlay a red [1] and blue [2] stream atop a yellow [0] stream. The intent is that the red stream will be exactly adjacent to the blue stream ; at frame 1731 the output video should transition immediately from pure red to pure blue :

  


  # These values result in a blank frame where the videos join at around 58s
LENGTH=1800
SPLIT_AT=1731

# These values work just fine.
# LENGTH=1800
# SPLIT_AT=2

ffmpeg \
-f lavfi -r 30 -i "color=c=yellow:s=1920x1080" \
-f lavfi -r 30 -i "color=c=990000:s=1920x1080" \
-f lavfi -r 30 -i "color=c=000099:s=1920x1080" \
-filter_complex \
"[1]concat=n=1[red_concat];"\
"[red_concat]tpad=start=0[red_pad];"\
"[0][red_pad]overlay=enable=between(n\,0\,$SPLIT_AT):eof_action=repeat[red_overlay];"\
"[2]concat=n=1[blue_concat];"\
"[blue_concat]tpad=start=$SPLIT_AT[blue_pad];"\
"[red_overlay][blue_pad]overlay=enable=between(n\,$SPLIT_AT\,$LENGTH):eof_action=repeat[blue_overlay];"\
"[blue_overlay]trim=end_frame=$LENGTH[full];" \
 -map "[full]" video.mp4 -y


  


  However, what I see is that there is a single black frame between the pure red and pure blue. This black seems to be coming from the frames that tpad adds.

  


  What I don't understand is why I'm seeing that black frame. As far as I can tell, my uses of tpad and overlay/enable-between should produce a seamless transition from red to blue. And indeed, if I change the SPLIT_AT value to almost anything else I get exactly that ; the value 1731 is "special" in some way.

  


  Can anyone explain what's going on ? Maybe more importantly, can anyone reproduce this problem ?

  


  My first thought was that I've got some form of fencepost error in how I'm using tpad or enable-between. However, that doesn't explain why the script works for most values but not 1731. This difference in results feels like there's a rounding error somewhere, but since I'm dealing purely in terms of integer frame numbers, I don't see how I could be introducing them.

  


  A few notes :

  


  


  • This is a drastically reduced example from a larger video compositing system. Things like the use of concat may appear unnecessary, but they're representative of what the larger system is doing.

    


  


  • I'm using ffmpeg-6.0 installed via homebrew on an M1 macbook

    


  


  


• Why aren't the videos in my S3 bucket buffering to html video tag ?

  2 juin 2019, par Michael Cain

  I have so far successfully programmed a node script on a Udoo x86 advanced plus that captures an Ethernet connected IP cam’s RTSP stream. I use ffmpeg to trans-code the stream into 5 second mp4 files. As soon as the files show up in the folder they are uploaded/synced to my AWS S3 Bucket. Next I have a Node server that GET’s the most recently created mp4 file from the S3 bucket and runs it through mediasource extension and finally to an html video tag.

  The videos are playing on the browser but not in any kind of synchronous manner. No buffering seems to be taking place. one video plays then another and so on. Video is skipping all over the place.

  I would really appreciate any guidance with this bug.

  export function startlivestream() {
  
    const videoElement = document.getElementById("my-video");
  
    const myMediaSource = new MediaSource();
  
    const url = URL.createObjectURL(myMediaSource);
  
    videoElement.src = url;
  
    myMediaSource.addEventListener("sourceopen", sourceOpen);
  
  }
  
  function sourceOpen() {
  
    if (window.MediaSource.isTypeSupported(
  
        'video/mp4; codecs="avc1.42E01E, mp4a.40.2"'
  
      )
  
    ) 
  
  {
  
         console.log("YES");
  
    }
  
  
  
  // 1. add source buffers
  
  
  
    const mediaCodec = 'video/mp4; codecs="avc1.4D601F"';
  
    var mediasource = this;
  
    const videoSourceBuffer = mediasource.addSourceBuffer(mediaCodec);
  
  
  
  // 2. download and add our audio/video to the SourceBuffers
  
  
  
  function checkVideo(url) {
  
    var oReq = new XMLHttpRequest();
  
      oReq.open("GET", url, true);
  
      oReq.responseType = "arraybuffer";
  
  
  
       oReq.onload = function(oEvent) {
  
        var arrayBuffer = oReq.response; // Note: not oReq.responseText
  
        if (arrayBuffer) {
  
           videoSourceBuffer.addEventListener("updateend", function(_) {
  
             mediasource.endOfStream();
  
             document.getElementById("my-video").play();
  
     });
  
             videoSourceBuffer.appendBuffer(arrayBuffer);
  
        }
  
      };
  
  
  
      oReq.send(null);
  
    }
  
  
  
    setInterval(function() {
  
      checkVideo("http://localhost:8080");
  
    }, 5000);

  My ffmpeg tags :

  const startRecording = () => {
  
    const args = [
  
      "-rtsp_transport",
  
      "tcp",
  
      "-i",
  
      inputURL,
  
      "-f",
  
      "segment",
  
      "-segment_time",
  
      "5",
  
      "-segment_format",
  
      "mp4",
  
      "-segment_format_options",
  
      "movflags=frag_keyframe+empty_moov+default_base_moof",
  
      "-segment_time",
  
      "5",
  
      "-segment_list_type",
  
      "m3u8",
  
      "-c:v",
  
      "copy",
  
      "-strftime",
  
      "1",
  
      `${path.join(savePath, "test-%Y-%m-%dT%H-%M-%S.mp4")}`
  
    ];

  From what I have learned about Mediasource extensions they allow multiple videos to be taken in and allow the client to buffer them so it looks like one longer video. In simple terms.

• Implementing Live Streaming Webinar Architecture - Need Guidance

  21 septembre 2023, par vishnugpl

  Question : I'm working on a project to implement a live streaming webinar platform, and I'm looking for guidance on the architecture and technologies to use. I have some specific requirements and constraints, and I'd appreciate any advice or pointers on how to get started.

  


  Context :
I'm tasked with building a live streaming webinar platform from scratch for our organization. Here are some key requirements and constraints :

  


  


  1. Live Video Streaming : We need to support live video streaming for
webinars, workshops, and conferences. Low latency is crucial for
real-time interaction between presenters and attendees.

     


  


  2. Scalability : The platform should be able to handle a large number of concurrent attendees, potentially thousands or more,
without compromising performance.

     


  


  3. Security : Security is a top priority. We need to ensure that only authorized users can access the webinars, and we must protect
against unauthorized recording or distribution of the content.

     


  


  4. User Interaction : Attendees should be able to ask questions, participate in polls, and interact with presenters in real-time.

     


  


  5. Recording and Playback : We also want to provide recorded versions of past webinars for on-demand viewing (HLS).

     


  


  


  Current Tech Stack :

  


  

  • Frontend : React

  


  • Backend : Node.js with Express

  


  • Database : MongoDB

  


  • Cloud Hosting : AWS

  


  


  Questions :

  


  


  • What technologies or services are best suited for handling live video streaming with low latency ?

    


  


  • How can I ensure scalability in terms of both video streaming and handling a large number of users ?

    


  


  • What security measures should I consider to protect the content and user data ?

    


  


  • Are there any recommended frameworks or libraries for building interactive features like real-time chat and polling ?

    


  


  • What's the best approach for recording and serving on-demand webinar content ?

    


  


  


  Additional Information :

  


  


  • Budget is a consideration, so cost-effective solutions are preferred.

    


  


  • Any open-source or third-party tools that can simplify development are welcome.

    


  


  • I have experience with AWS, but I'm open to other cloud providers if they offer advantages for this project.