Recherche avancée

Sur d’autres sites (7718)

• avformat/hls : Set AVFMT_TS_DISCONT flag on HLS input format

  27 octobre 2019, par Philip Langdale

  avformat/hls : Set AVFMT_TS_DISCONT flag on HLS input format
  There have been many reports over the years about problems when
  
  taking an HLS stream as input to `ffmpeg` where there are timestamp
  
  discontinuities present. This is explicitly supported in the
  
  HLS spec (EXT-X-DISCONTINUITY) and often used for ad injection.
  Various fixes and work-arounds have been proposed over the years,
  
  but one step that seems obvious, even if it's not a complete fix,
  
  is to mark the HLS input format as supporting discontinuities. This
  
  will prevent timestamp fixup logic in ffmpeg.c kicking in that ends
  
  up mangling the timestamps unnecessarily.
  I've tested this out with an example provided by Joe Koberg early
  
  last year, and it is sufficient to allow `ffmpeg` to download and
  
  mux the stream correctly. Joe had briefly suggested that other
  
  situations can still be handled incorrectly, but this seems like
  
  a strict improvement.
  Joe's example :
  https://s3.amazonaws.com/playon-test-videos/discont_test_new/discont_test.m3u8
  Reviewed-by : Steven Liu <lq@onvideo.cn>
  
  Reviewed-by : Dennis Mungai <dmngaie@gmail.com>
  

  • [DH] libavformat/hls.c

• What's the best way to get video metadata from a MP4 file in ASP.Net MVC using C# ?

  23 septembre 2019, par Maddhacker24

  I’ve been searching on Google and StackOverflow for a good couple of hours. There seems to be a lot of similar questions on StackOverflow but they are all about 3-5 years old.

  Is using FFMPEG still the best way these days to pull metadata from a video file in a .NET web application ? And if so, what’s the best C# wrapper out there ?

  I’ve tried MediaToolkit, MediaFile.dll without any luck. I saw ffmpeg-csharpe but that looks like it hasn’t been touched in a few years.

  I haven’t found any current data on this subject. Is the ability to pull metadata from a video built into the latest version of .NET now ?

  I’m basically looking for any direction at this point.

  I should add that whatever I use could be invoked thousands of times per hour so it will need to be efficient.

• RTP Timestamps Are Not Monotonically increasing

  25 août 2019, par Fr0sty

  I am finding it a bit difficult trying to understand whether or not the hack around with FFmpeg and OpenCV really provided a RTP timestamp. My last post helped a little bit but got me stuck in trying to validate the timestamps obtained through this work around by modifying ffmpeg and opencv.

  FFmpeg version : 4.1.0
  OpenCV version : 3.4.1

  import cv2
  
  import time
  
  from datetime import datetime, date
  
  
  
  uri = 'rtsp://admin:password@192.168.1.66:554/Streaming/Channels/101'
  
  cap = cv2.VideoCapture(uri)
  
  '''One is the offset between the two epochs. Unix uses an epoch located at 1/1/1970-00:00h (UTC) and NTP uses 1/1/1900-00:00h.
  
  This leads to an offset equivalent to 70 years in seconds (there are 17 leap years between the two dates so the offset is'''
  
  time_offset = 2208988800 # (70*365 + 17)*86400 = 2208988800 (in seconds)
  
  # offset = 3775484294
  
  days = 43697
  
  pdat = "1900-01-01 00:00:00:00"
  
  mdat = "2019-08-23 22:02:44:00" # str(datetime.now()) + str(datetime.now().time())
  
  pdate = datetime.strptime(pdat, "%Y-%m-%d %H:%M:%S:%f").date()
  
  mdate = datetime.strptime(mdat, "%Y-%m-%d %H:%M:%S:%f").date()
  
  delta = (mdate - pdate).days
  
  offset = delta * 86400
  
  def time_delta(s):
  
      return (s - time_offset)
  
  
  
  while True:
  
      frame_exists, curr_frame = cap.read()
  
      if frame_exists:
  
          seconds = cap.getRTPTimeStampSeconds()
  
          fraction = cap.getRTPTimeStampFraction()
  
          timestamp = cap.getRTPTimeStampTs()
  
          unix_offset = seconds - time_offset
  
          msec = int((int(fraction) / 0xFFFFFFFF) * 1000.0)
  
          ts = float(str(unix_offset) + "." + str(msec))
  
          # print("Timestamp per Frame:%i" % timestamp)
  
          print((datetime.fromtimestamp(float(ts) + offset)))
  
  cap.release()

  My Output :

  On August 23, 2019 at 22:02

  ...
  
  2019-08-23 13:59:52.781000
  
  2019-08-23 13:59:52.726000
  
  2019-08-23 13:59:52.671000
  
  2019-08-23 13:59:52.616000
  
  2019-08-23 13:59:52.561000
  
  2019-08-23 13:59:52.506000
  
  2019-08-23 13:59:52.451000
  
  2019-08-23 13:59:52.396000
  
  2019-08-23 13:59:52.342000
  
  2019-08-23 13:59:52.287000
  
  2019-08-23 13:59:52.232000
  
  2019-08-23 13:59:52.177000
  
  2019-08-23 13:59:52.122000
  
  2019-08-23 13:59:52.067000
  
  2019-08-23 13:59:52.012000
  
  2019-08-23 13:59:53.570000
  
  2019-08-23 13:59:53.020000
  
  2019-08-23 13:59:53.847000
  
  2019-08-23 13:59:53.792000

  I’ve noticed how the time increments weirdly (that’s not suppose to happen in the real, current time), such as the last two lines and a few others in between in the output. A bit flabbergasted as to what went wrong. Also trying this out on multiple IP cameras, with each showing a different timestamp probably related to when they were turned on.