Recherche avancée

Sur d’autres sites (9463)

• Webcam Serverless Live stream

  23 juillet 2021, par curiouscoder

  I'm trying to live stream my webcam in a serverless way in the following flow :

  


  webcam browser >> s3 bucket >> lambda/ffmpeg encoding >> s3 output bucket >> dash player

  


  This is working really good so far but I'm facing the following problem :

  


  ffmpeg will only encode those seconds received (I stream the webcam to s3 each X seconds with some 300kb .webm file). So the .mpd file generated by ffmpeg encoder will have the type 'static' when ffmpeg finishes encoding and not the 'dynamic' type desired. Therefore, the dash player won't request the other files from s3 and the streaming will stop. For example, if I let the webcam streaming running for 15 seconds, the viewer is able to watch the 15 minutes. But if I keep sending the streams each 2 seconds the viewer will be able to watch only the first 2 seconds because browser won't request any other .m4s files.

  


  So, I have the following question :

  


  Is there a way to force the dash player to reload the .mpd file that is stored in s3 even when the type is static instead of dynamic ?

  


  Thanks in advance !

  


• FFmpeg option to generate a .m4s file every second

  29 janvier 2020, par Adrian Angkajaya

  I am trying to stream my live recording from a camera (web cam/ ip cam) to my web application. The streaming technique I use is MPEG-DASH, which has manifest in MPD format. To generate an MPD format from the web-cam, I use FFmpeg tool in shell command line :
  ffmpeg -re -y -f dshow -i video="Logitech HD Webcam C525" -c:v libx264 -c:a libfdk_aac -f dash "manifest.mpd". This code will generate a video chunk in .m4s format every 5-8 seconds.

  Question is, what FFmpeg option can I use to generate a .m4s file every second instead of every 5-8 seconds ? I suppose it has something to do with segment ?

• How to segment audio and video to the same segments number ?

  24 décembre 2019, par julian zapata

  I’ve extracted audio and video from the same mp4 file and created different variants for different resolutions. When I segment the videos and the audio file with MP4Box, the video segments are the same segments number but the audio segments have one more. I need the video and audio segments to be the same number to use with dash.

  I’m using the next commands to produce the video and audio files :

  Command to extract the audio from mp4 file :

  ffmpeg -y -i "transformers.mp4" -c:a aac -b:a 192k -vn "transformers_audio.m4a"

  Produce each variant of the video with the next command modifying the resolution and bitrate parameters :

  ffmpeg -i transformers.mp4 -c:v libx264 -r 24 -x264opts "keyint=48:min-keyint=48:no-scenecut" -an -vf scale=640:360 -b:v 750k -dash 1 transformers_640x360_750k.mp4

  Generate the mpd file and segment the videos and audio each 4 seconds with :

  mp4box -dash 4000 -profile "dashavc264:live" -bs-switching no -sample-groups-traf \ 
  
  -out output4\ 
  
  transformers.mpd \ 
  
  transformers_480x270_400k.mp4 \ 
  
  transformers_640x360_800k.mp4 \ 
  
  transformers_960x540_1200k.mp4 \
  
  transformers_1280x720_1500k.mp4 \ 
  
  transformers_1920x1080_4000k.mp4 \
  
  transformers_audio.m4a

  This produces 36 segments for each variant of the video and 37 for the audio. How to solve this little variation ? How to make the audio segments number exactly the number of video segments ?