Recherche avancée

Sur d’autres sites (9688)

• mpegtsenc : Add support for muxing Opus in MPEG-TS

  5 novembre 2015, par Sebastian Dröge

  mpegtsenc : Add support for muxing Opus in MPEG-TS
  Signed-off-by : Sebastian Dröge <sebastian@centricular.com>
  
  Previous version reviewed-by : Kieran Kunhya <kierank@obe.tv>
  
  Signed-off-by : Michael Niedermayer <michael@niedermayer.cc>
  

  • [DH] libavformat/mpegtsenc.c

• ffmpeg demux concat video output blinks/stuttering between joined clips

  11 juin 2024, par hieroshima

  I'm experiencing a strange issue with FFmpeg after partially joining a few MP4 clips into one output file (output.mp4). The problem is that the video between clips randomly blinks or stutters, showing frames from the previous scene in the next scene. This happens in a completely random manner.

  


  All my MP4 clips have the same FPS, TBR, and TBN values, except for the bitrate (bolded below), which varies for each clip. These clips were prepared and encoded using AWS MediaConvert to ensure they have consistent parameters.

  


  Here’s the ffprobe result for one of the clips :

  


  Stream #0:0[0x1](und): Video: h264 (High) (avc1 / 0x31637661),
yuv420p(progressive), 640x360 [SAR 1:1 DAR 16:9], **84 kb/s**, 29.97 fps,
29.97 tbr, 30k tbn (default)


  


  Below is my input.txt file for concatenation :

  


  file 'clip1_360.mp4' 
outpoint 00:00:04.00
file 'clip2_360.mp4' 
outpoint 00:00:03.00
file 'clip3_360.mp4' 
outpoint 00:00:04.00


  


  The FFmpeg command I'm using to concatenate the clips is :

  


  ffmpeg -y -f concat -safe 0 -protocol_whitelist
file,http,https,tcp,tls -fflags +igndts -i input.txt -c copy output/output.mp4


  


  I use the -fflags +igndts flag because I receive warnings about non-monotonic DTS. I’m employing the concat demuxer to avoid re-encoding the video, as the time required to generate the output is crucial. This is why I have prepared and encoded all clips in advance to have consistent values.

  


  Here is screen recording from my video player to see this problem :

  


  https://youtu.be/SfWnCbMeOfw

  


  Any assistance or suggestions on how to resolve this issue would be greatly appreciated. Because I don't have any more idea what I can do with this. If you need some more details which I didn't provide, please let me know.

  


• OpenCV reading from live camera creates a short video that moves quickly

  17 novembre 2022, par user19019404

  I am reading in a live vide stream from a CCTV camera. The camera is set to 5 fps, another is set to 25fps and another to 30fps. Irrespective of the FPS that the camera is set, I can record 5 minutes but end up with a 30 second recorded clip where everyone is running around the scene.

  


  My code is the 'typical' read in video and write video code that you would find online such as (code below simplified for readability) :

  


  import cv2

video = cv2.VideoCapture(live RTSP address of camera)

if (video.isOpened() == False):
    print("Error reading video file")
else:
    frame_width = video.get(cv2.CAP_PROP_FRAME_WIDTH)
    frame_height = video.get(cv2.CAP_PROP_FRAME_HEIGHT)
    frame_fps = video.get(cv2.CAP_PROP_FPS)
    size = (frame_width, frame_height)
    result = cv2.VideoWriter('filename.avi',cv2.VideoWriter_fourcc(*'MJPG'),frame_fps , size)

    while(True):
        ret, frame = video.read()
        if ret == True:
            result.write(frame)
            cv2.imshow('Frame', frame)
        if cv2.waitKey(1) & 0xFF == ord('s'):
            break
        else:
            break
    video.release()
    result.release()
    cv2.destroyAllWindows()
print("The video was successfully saved with new fps")


  


  I have tried playing with the FPS by reading in the FPS from the live camera and using the same FPS in the video write, but all that results is a video that is a fraction of the real time and with people zooming around the scene. So watching a 5 minute smooth video results in a 20 second recorded video with everyone zooming around.

  


  Is this something that I need to fix in the writing of the video or do I need a second pass with ffmpeg to readjust the video ?

  


  Much appreciated

  


  Update, corrected the code above and :
When printing the frames read and the frame written the numbers are the same, showing that each frame that is read is being written (so I am not losing frames along the way thereby writing half the amount of frames).