Recherche avancée

Sur d’autres sites (5038)

• rtsp-simple-server store video during publish step in byte

  23 juin 2022, par zt1983811

  We are using rtsp-simple-server as our video streaming publisher. In the current repo they provided a step to store the video by command line after the stream is ready like this. However, we don't want to re-stream again to store the video, we would like to store the video during the video bytes publish to rtsp step.

  


  I found the code is here to publish RTP package bytes to RTSP stream. I actually create a function here to store the RTP payload bytes to a file. like following code changes to store video in byte. The output of byte file is here. However when I try to use ffmpeg to convert it to mp4 I got error like following
cmd I use

  


  ffmpeg -i video/output_video_06_23_2022_13.264 -c:v libx264 video/fromh246.mp4


  


  Err :

  


  [h264 @ 0x55c7b595f280] Format h264 detected only with low score of 1, misdetection possible!
[h264 @ 0x55c7b5960dc0] missing picture in access unit with size 16263103
[AVBSFContext @ 0x55c7b5968000] No start code is found.
video/output_video_06_23_2022_13.264: could not find codec parameters
Input #0, h264, from 'video/output_video_06_23_2022_13.264':
  Duration: N/A, bitrate: N/A
    Stream #0:0: Video: h264, none, 1200k tbn
Output #0, mp4, to 'video/fromh246.mp4':
Output file #0 does not contain any stream



  


  I also tried these are not working as well. The file generated is not playable

  


  ffmpeg  -f rawvideo -pix_fmt yuyv422 -r 25 -s:v 1280x720 -i video/output_video_06_23_2022_13.264 -c:v libx264 abcd.mp4
ffmpeg  -f rawvideo -pix_fmt yuv420p -r 25 -video_size 1280x720 -i video/output_video_06_23_2022_13.264 -c:v libx264 abcdd.mp4


  


  You can check out my stackover_save_video_byte branch. and start server then run following CMD to publish and store the stream to byte file

  


  ffmpeg -re -nostdin -i video/inputVideo.mp4 -vcodec libx264 -preset:v ultrafast -acodec aac -f flv rtmp://localhost:1935/mystream


  


  I read a lot of question here and there, some of the post saying I need add separator (0000x1) I tried it also, not really working. I am wondering few things

  


  

  1. The byte I am saving, is that format in h264 ? (from the byte file header it seems like it)

  


  2. For diff RTP packet do I need to add separator ?

  


  3. Before I am saving do I need to do decoder or encoder again ? I see the repo is actually did the decode from ACC -> NALU -> encode to RTP. I actually took the RTP payload. Do I need some special things to make the bytes to h264 standard ?

  


  


• Upgrading Transcooding system, Ideal database to store chunk metadata [closed]

  11 mai 2022, par Vitoo

  I try to update my transcoding system, to make it more efficient and failure resistant. Right now service transcode the whole video at once generating multiply resolution it consumes a lot of resources and when transcoding crash it has to start from beginning.

  


  I think the better method is to chop a video into small 5-10 sec pieces and encode it separately and at the end merge it, but the main problem is to delegate pieces to different encoder workers. I need to store chunks in some database and every worker should lock that it is already working on it and if successful change state to true.

  


  What do you think about this ?
What kind of database should I use ?
Do you have some additional ideas or propositions ?

  


• fftools/ffmpeg : store the output file index in OutputFile

  11 décembre 2021, par Anton Khirnov

  fftools/ffmpeg : store the output file index in OutputFile
  Use it to simplify check_init_output_file(). Will allow further
  
  simplifications in the following commits.
  

  • [DH] fftools/ffmpeg.c
  • [DH] fftools/ffmpeg.h
  • [DH] fftools/ffmpeg_opt.c