Recherche avancée

Sur d’autres sites (9918)

• Optimizing Adaptive Streaming with FFMPEG

  25 octobre 2018, par Ramesh Navi

  I am working on a video on demand website, I am using laravel 5.7, FFMPEG and DASH player from (dashif.org). Got some questions.

  Extracting audio like :

  ffmpeg -i original.mp4 -vn -acodec libvorbis -ab 128k -dash 1 my_audio.webm

  Converting video like :

  ffmpeg -i original.mp4 -c:v libvpx-vp9 -keyint_min 150 \
  
  -g 150 -tile-columns 4 -frame-parallel 1  -f webm -dash 1 \
  
  -an -vf scale=160:190 -b:v 250k -dash 1 video_160x90_250k.webm \
  
  -an -vf scale=320:180 -b:v 500k -dash 1 video_320x180_500k.webm \
  
  -an -vf scale=640:360 -b:v 750k -dash 1 video_640x360_750k.webm \
  
  -an -vf scale=1280:720 -b:v 1500k -dash 1 video_1280x720_1500k.webm

  Creating manifest like :

  ffmpeg \
  
  -f webm_dash_manifest -i video_160x90_250k.webm \
  
  -f webm_dash_manifest -i video_320x180_500k.webm \
  
  -f webm_dash_manifest -i video_640x360_750k.webm \
  
  -f webm_dash_manifest -i video_1280x720_1500k.webm \
  
  -f webm_dash_manifest -i my_audio.webm \
  
  -c copy \
  
  -map 0 -map 1 -map 2 -map 3 -map 4 \
  
  -f webm_dash_manifest \
  
  -adaptation_sets "id=0,streams=0,1,2,3 id=1,streams=4" \
  
  my_video_manifest.mpd

  Now the problems :

  1. Video conversion takes a lot of time on the latest i5 8gb
     think-pad running Ubuntu 18. 4minute mp4 took more than 30minute. 10minute, 720p MP4 took forever, had to kill the process. Is
     that normal ? Any idea to optimize ?
  2. Need to find out original
     video’s dimension so that I can avoid dimension conversion more than
     the original. ffprob looks fine in command but produces too much of
     information than required, is there any simple function ?
  3. Converted webm file is bigger in size than original mp4, original
     mp4 of 720p was 33MB, but 640p webm is 76MB. Is that normal or
     something wrong ?

  Any suggestions to optimize the process are welcome.

• Can I pipe multiple ffmpeg outputs to different pipes ?

  4 mars 2021, par mr. rogers

  I have an AWS workflow that streams an audio or video file from S3 to ffmpeg inside a lambda. Which then streams the encoded output back to S3. I want to extend that flow to generate DASH content. On my local device, I can generate the MPD file and the encoded AV files. And there's the problem. Encoding to dash generates at least two files, even with the -single_file flag.

  


  Can I limit the dash export to just the video or MPD file ? Or can I somehow pipe the multiple outputs separately through some flag magic ?

  


  Thanks in advance.

  


• How to implement multiple video resolutions on front and back-end

  25 février 2021, par LanGuu

  I need a solution or hint on how should I handle multiple resolutions on front-end and back-end. I have been reading about hls, dash, and mse for last few days but the more information I read, the more lost I am.

  


  Right now I have only a microservice for downscaling video using FFmpeg. The FFmpeg returns mp4 with no hls or dash, 4 video resolutions with no sound, and 2 different audio quality.
On front-end, I use react-player. The hardest part is that I need to merge audio and video, synchronize and handle change quality events.
I have found Media source extensions but it won't work I am not sure, but probably because I am using raw mp4 files.

  


  I would like to ask you a few questions

  


  

  1. It is possible to combine video and audio sources without MSE and still keep video and audio separate

  


  2. Do I need HLS or Dash to use MSE ?

  


  3. What is a difference between progressive download and progressive streaming ?

  


  4. If I have to choose what is better right now hls or dash ? what is much easier to implement ?