Recherche avancée

Sur d’autres sites (10949)

• OpenCV no longer opens video files VideoCapture

  15 février 2018, par gruffmeister

  I have a problem seemingly caused by OpenCV 3.xx - the problem does not manifest in OpenCV 2.xx

  The issue is reading video files. I’ve set my code up as follows :

  >#include 
  
  >#include 
  
  >#include 
  
  >#include 
  
  >#include 
  
  
  
  >int main()
  
  
  
  >    cv::VideoCapture cap;
  
  >    cv::Mat frame;
  
  >    if(!cap.open("Myfile.avi"))
  
  >        std::cout << "Open failed" << std::endl;
  
  >    else
  
  >        cap.read(frame);
  
  >
  
  >    cv::imshow("Frame", frame);
  
  >    cv::waitKey(5000);
  
  >    return 0;

  Now the problem is when the code gets to "cap.read(frame)" I get a "vector subscript is out of range" error with OpenCV 3.40 and this does not happen with my build of OpenCV 2.4.9. The format of the file is in avi, its not some weird codec, and clearly it works in previous versions of OpenCV.

  I’ve tried other OpenCV 3.xx builds and I get the same or similar problems with simply reading a file in.

  My question is twofold :
  How do I get OpenCV 3.xx to work with reading video files (or do I need to regress to 2.xx ?)

  Why has the major revision change completely screwed up video file reading ? That doesn’t make any sense for a computer vision API.

  As a guess it will be something to do with the FFMPEG implementation because various searches have turned up other people having issues with this.

  Any help is much appreciated.

  Thanks

• ffmpeg multiple filters need assistance

  16 mars 2018, par Tony Hanna

  I have one movie .mp4, one logo .png and one text (banner) sliding in the bottom, and I need to add subtitles. I’m trying this with complex filters but always with an error. Please help.

  ffmpeg.exe -re -i wonder.mp4 -i logo72.png -filter_complex "[0:v]drawtext=fontcolor=white:fontsize=32:fontfile='C:\Windows\Fonts\arial.ttf':textfile='d:\banner.txt':reload=1:y=h-line_h:x=w-(mod(n*3\,w+tw)-tw/40),drawbox=y=ih-40:color=black@0.4:width=iw:height=40:t=max[text];[text][1:v]overlay=15:15[filtered];[0:v][1:v]overlay=0:0,subtitles=subtitle.srt[out]" -map "[filtered]" - map "[out]" -map 0:a -codec:v libx264 -codec:a copy -f mpegts udp://10.10.10.4:5000?pkt_size=1316

  What am I doing wrong ?

• Force ffmpeg to generate WebVTT for each subtitle track

  14 juin 2018, par Rogue

  I use the following ffmpeg command to generate an HLS stream from a video :

  ffmpeg -i pipe:0 \
  
      -y -b:a 64k -acodec aac -vcodec copy \
  
      -hls_time 10 -hls_playlist_type vod -start_number 0 -hls_base_url http://127.0.0.1:5000/ -hls_list_size 0 \
  
      -f hls -crf 20 -hls_flags split_by_time -force_key_frames "expr:gte(t,n_forced*3)"\
  
      -threads 4 \
  
      target/stream.m3u8

  It works perfectly, and to my pleasure and surprise ffmpeg generates a second m3u8 file corresponding to the subtitles that are embeded in the video file, with the appropriate .vtt files.
  The problem is that when i input a video file with more than one subtitle tracks, i only get files for the first one.
  How can i force ffmpeg to output ALL tracks as m3u8 / .vtt files ?

  Thanks a lot