Recherche avancée

Sur d’autres sites (8478)

• 🧵 [QUESTION][QUESTION] Reuse PlainTransport for new FFmpeg stream without full reinit

  29 mai, par Sang Vo

  I'm building a backend-driven streaming application using Mediasoup + FFmpeg + PlainTransport.
My goal is to switch between different media sources (e.g. welcome.mp4, waiting.mp4, streaming.ts) from backend, while keeping the same PlainTransport alive to avoid the overhead of tearing down and rebuilding the pipeline.

  


  ✅ Current Setup :
Backend : NestJS server with Mediasoup

  


  Media ingest : FFmpeg sends RTP stream to PlainTransport

  


  Producer : created after FFmpeg starts

  


  Frontend : React client that consumes via Consumer after signaling

  


  🔄 What I want :
When the backend Start streaming a file (e.g., welcome.mp4), I want to :

  


  Start FFmpeg again with welcome.mp4 using the same RTP ports (same PlainTransport)
After welcome is wailting.mp4. When has event streaming start play streaming, end event streaming play wailting

  


  Create a new Producer with the new stream

  


  Notify the frontend to create a new Consumer

  


  All this without having to destroy and recreate the PlainTransport for all event

  


  🧠 Questions :
Is this a valid and recommended workflow in Mediasoup ?

  


  If so, does Mediasoup allow reusing the same PlainTransport across multiple Producer instances (one at a time) ?

  


  Will RTP stream re-sync correctly if FFmpeg restarts and sends new RTP packets ?

  


  Is it necessary to explicitly configure SSRC and payloadType to match, or will Mediasoup auto-detect again per new producer ?

  


  Any insights or recommendations on best practices for this dynamic switching scenario would be very helpful !

  


  Thanks in advance 🙏

  


• Writing metadata (Artist Name, Song Title, Year, Album, Duration, Genre) to mp3/m4a audio file using youtube-dl (subsequent AtomicParsely error)

  10 décembre 2016, par IRNotSmart

  I am extracting audio only from youtube videos using youtube-dl. I would like to write the metadata (i.e. Artist Name and Song Title, Year, Album, Duration, Genre) into the mp3/m4a file after downloading. My attempt to accomplish this starts with this code :

  @echo off
  
  youtube-dl --format m4a/mp3 --youtube-skip-dash-manifest --embed-thumbnail -o "%%(title)s.%%(ext)s" --metadata-from-title "%%(artist)s - %%(title)s" --add-metadata 2Y6Nne8RvaA
  
  pause

  This code produces the following output :

  [youtube] 2Y6Nne8RvaA: Downloading webpage
  
  [youtube] 2Y6Nne8RvaA: Extracting video information
  
  [youtube] 2Y6Nne8RvaA: Downloading thumbnail ...
  
  [youtube] 2Y6Nne8RvaA: Writing thumbnail to: Kungs vs Cookin' on 3 Burners - Thi
  
  s Girl.jpg
  
  [download] Destination: Kungs vs Cookin' on 3 Burners - This Girl.m4a
  
  [download] 100% of 2.99MiB in 00:01
  
  [ffmpeg] Correcting container in "Kungs vs Cookin' on 3 Burners - This Girl.m4a"
  
  
  
  [fromtitle] parsed title: This Girl
  
  [fromtitle] parsed artist: Kungs vs Cookin' on 3 Burners
  
  [ffmpeg] Adding metadata to 'Kungs vs Cookin' on 3 Burners - This Girl.m4a'
  
  ERROR: AtomicParsley was not found. Please install.
  
  Press any key to continue . . .

  As you can see, I am able to successfully able to add a few of the tags from the video, but not all of them and the Year is royally screwed up.
  

  What is this AtomicParsely error and how do I remedy it ? Do I need this program to correctly add all the Metadata to the file that I want, or can this be accomplished in another way ?

  Referencing Steven Penny’s post, FFmpeg metadata not showing in Windows ?, is solving this problem as simple as using an ffmpeg command ?

  When I do a google search for this song, the first link that shows is the exact link I’m using on YouTube, and the search shows pertinent metadata (see below). I’m not sure if this data is input manually by users, or if Google mined this from the video :
  

  I admit that I’m new to using youtube-dl and ffmpeg, but with the help of the commenters on StackOverflow, I’m learning more each day. This post is a follow-up to my previous question : Downloading YouTube to mp3 and writing metadata (artist/song title) to mp3 file using youtube-dl

• Merge individual frame to video file using Opencv

  18 août 2022, par Rohit

  I am trying to stack a individual frame to a video file using Opencv. I want to combine two different code together to make the individual frame.
Following code help me extract the individual frame,

  


  fourcc = cv2.VideoWriter_fourcc(*'mp4v')
out = cv2.VideoWriter('file_data.mp4',fourcc,20 (1920,1080),False)
while True:
    ret, frame=cap.read()
    mask = object_detector.apply(frame)
    _, mask  = cv2.threshold(mask,254,255,cv2.THRESH_BINARY)       
    contours,_ = cv2.findContours(mask, cv2.RETR_TREE, cv2.CHAIN_APPROX_SIMPLE)
    res = cv2.bitwise_and(frame,frame,mask=mask)
    for cnt in contours:
        area = cv2.contourArea(cnt)
        if area>1000:   
            #print("Area of contour:", area)
            cv2.drawContours(frame, [cnt], -1, (0,255,0),2)
            cv2.imwrite("file%d.jpg"%count, frame)
            out.write(frame)
    if cv2.waitKey(1) and 0xFF == ord('q'):
        break


  


  I tried storing the individual frame in array, but it didn't work. It doesn't show any error, but pc crash.

  


  


  fps = 20,
,width = 1920
,height = 1080