Recherche avancée

Sur d’autres sites (7788)

• Using GDIgrab in FFmpeg with dshow Audio produces black screen

  16 mars 2016, par Spreadys

  here is my command :

  ffmpeg -f gdigrab -framerate 25 -offset_x 10 -offset_y 10 -show_region 1 -draw_mouse 1 -video_size 1280x720 -i desktop -f dshow -i audio="Microphone (2- ATR USB microphone)" -r 25 -threads 4 -c:v libx264 -pix_fmt yuv422p -preset superfast -tune fastdecode -x264opts keyint=25:min-keyint=1 -crf 4 -c:a aac -profile:a aac_low -async 25 "C:\Users\david\Desktop\%output%.mp4"

  The gdigrab video works great when it is on its own (no audio). The audio works fine when it is on its own (no video). When I join the two commands to capture both together, as soon as I move a window within my capture area, the area goes black.

  In Windows 7, I used to get around this by stopping the desktop composition service prior to capture, (SC stop uxsms), but this is now not possible in Win10.

  I thought it may be something graphics card related.
  My main monitor is on an Nvidia card, with my second running from the onboard Intel. This is setup for Quicksync H264 playback and encoding with my NLE.

  I know that I could use a dshow screen capture driver such as UScreen but am trying to avoid that as I need the capture area to be specified each time from a simple batch.

  Any help appreciated to solve this black area problem- its driving me crazy !
  David

• How to queue ffmpeg FIFO

  29 avril 2013, par Francois

  we build a service similar to youtube. Also converting runs fine with ffmpeg using this script from another post here :

  How do I set up an ffmpeg queue ?

  #!/bin/bash
  
  
  
  pipe=/tmp/ffmpeg
  
  
  
  trap "rm -f $pipe" EXIT
  
  
  
  # creating the FIFO    
  
  [[ -p $pipe ]] || mkfifo $pipe
  
  
  
  while true; do
  
  # can't just use "while read line" if we 
  
  # want this script to continue running.
  
  read line < $pipe
  
  
  
  # now implementing a bit of security,
  
  # feel free to improve it.
  
  # we ensure that the command is a ffmpeg one.
  
  [[ $line =~ ^ffmpeg ]] && bash <<< "$line"
  
  done

  This works pretty good when i send one by one to the named pipe. When i send more than one at same time the second one queues the terminal to the point the first one finished. if a try more than 2 the third one will not be transcoded.

  So i tried to workaround with background sending to get the terminal free (just drop the echo command and close the ssh connection) but this doesn't work, then i played around with screen -X but also no luck. Maybe someone has a good idea to deal this.

  What i wanna do is : Every uploaded video which is needed to transcode will send a echo to the named pipe. FIFO should match but not blocking the terminal. So i think i need something to really queue ffmpeg input.

  kindest regards
  Francois

• Merge commit ’00b62968d079e63bf22028f253ac297292436ebe’

  2 novembre 2015, par Hendrik Leppkes

  Merge commit ’00b62968d079e63bf22028f253ac297292436ebe’
  * commit ’00b62968d079e63bf22028f253ac297292436ebe’ :
  
    os_support : Don’t try to return the service name as a string in getnameinfo
  Merged-by : Hendrik Leppkes <h.leppkes@gmail.com>
  

  • [DH] configure
  • [DH] libavformat/os_support.c