Recherche avancée

Sur d’autres sites (4969)

• Scheduled ffmpeg function gives thread.error and also fires ffmpeg too many times

  16 février 2016, par user2192778

  I want to record a clip of a radio stream every hour. Below is the code I am using to accomplish this so far.

      def sched(): # schedules a recording every hour
  
          def stream_record ():
  
              timeinfo = datetime.now().strftime('%Y%m%d_%H%M_%S%f')
  
              ffmpegEXE = "C:/path/to/ffmpeg.exe" 
  
              subprocess.call([ffmpegEXE, '-i', url, '-t', '00:07:00', 
  
              output_folder + timeinfo + '_' + str(start_minute) + 'url.mp3'], shell=True)
  
  
  
          i = 0
  
          while True:
  
  
  
          x = datetime.today()
  
          y=x.replace(day=x.day+1, hour=i, minute= start_minute, second=0, microsecond=0) 
  
          i = (i + 1) % 24
  
          delta_t=y-x
  
          secs=delta_t.seconds+1
  
          t = Timer(secs,stream_record)
  
          t.start()
  
  
  
  sched()

  Two things go wrong. (1) It will run, however an error reads :

  line X in (module)

  sched()

  line Y in sched

  t.start()

  line Z in start

  _start_new_thread(self.__bootstrap, ())

  thread.error : can’t start new thread

  And (2) when it runs, ffmpeg will initialize a recording anywhere from 5-15 times, saving many clips when I only want it to save one.

  How do I fix these errors and get ffmpeg to connect and record only one clip every hour ?

  I know this is an issue with the scheduling function ; the ffmpeg command works fine, as does the python script calling it.

• Setting up time triggers on video player

  28 juillet 2022, par gPats

  Background :

  



  I have a some text that I want to display in a textbox whenever the video is playing in a certain interval, say (00:02:00-00:04:00 "Hello there").

  



  These textbox should only be displayed whenever the video is playing in the particular interval. If I rewind, or skip the video to a different time, the textbox should be generated/destroyed depending whether the current time falls within the interval or not.

  



  I have searched the documentation for ffplay and vlc for implementing this functionality.

  



  What I have tried :

  



  ffplay : my strategy would be to keep polling the current time of the video which ffplay dumps in STDERR and check if the current time falls within the interval.

  



  LibVLC :
LibVLC supports asynchronous events, but not the kind which would depend on the time at which video is playing.

  



  My question :

  



  1.Is it advisable to keep polling the current video time or setup callbacks/trigger functions ?

  



  2.Is it possible to get the current playing time of the video as a variable and create callback functions accordingly ?

  



  NOTE : I am aware that subtitles do the same thing, however I want the text to be in a different window. The textbox should be decoupled from the video player I use.

  


• Controlling end time in video player via AVPacket information / setting pts/dts properly

  26 janvier 2017, par SyntheticGio

  I’m currently working in C/C++ using the FFMPEG core (libavcodec, etc.). I’m capturing a stream and writing it in chunks to different files. So imagine the stream is 5 minutes in length and I’m writing five files of one minute in length each. I’m able to do this successfully.

  Currently, each file after the first file has a start time equal to the time it would have been in the un-chunked stream. So the second video file starts at 1 minute, the third starts at 2 minutes, etc. This was inadvertent but as it turns out is beneficial in my particular use case.

  VLC or other video players that I’ve tried report this start time ’properly’, but the end time shows as the duration (not start time + duration). My gut feeling is that the player simply is making the assumption all videos start at 0 and it shows the length as the ’end time’ but I don’t actually know this so I’d like to know if there is anyway to set the AVPacket information so the player for the third video would start at 2 minutes and end at 3 minutes (for a 1 minute length video) - as an example ?

  As an alternative, if I wanted to do this the traditional way (reset each chunk to starting at time 0), I assume I’d normalize the AVPacket.pts and AVPacket.dts by subtracting the values of the final packet in the previous chunk ? This seems like this strategy would work for pts but I’m less sure about it working for dts. I feel like it would generally work for dts but there might be times when this fails, so I’d like to know if this is a safe method (or if there is a better method I should use in this case).