Recherche avancée

Sur d’autres sites (7572)

• streaming video is playing, when voice input enable the video change to silence,how to fix this bug ?

  18 avril 2015, par CrespoXiao

  Please check the image, I add the following code when voide input finish, but it doesn’t work. Once the voice input finish, the video would have sound after a short while, but sometimes not, don’t know why. Any solution ?

      /**
  
   *  back from other audio app or use voice input
  
   */
  
  - (void)enableAudio {
  
  //    AudioSessionInitialize(NULL, NULL, NULL, NULL);
  
  //    UInt32 sessionCategory = kAudioSessionCategory_MediaPlayback;
  
  //    AudioSessionSetProperty(kAudioSessionProperty_AudioCategory,sizeof(sessionCategory),&sessionCategory);
  
  //    AudioSessionSetActive(YES);
  
  
  
      [[AVAudioSession sharedInstance] setCategory:AVAudioSessionCategoryPlayback error: nil];
  
      [[AVAudioSession sharedInstance] setActive: YES error: nil];
  
  }

• Copy PTS from original video stream to processed video stream using ffmpeg and opencv

  25 janvier 2019, par Krishna Prasad Yellapragada

  I have two video files in webm format. First is the original video, captured using Chrome’s MediaRecorder API. The second is a processed video generated by processing each frame of the original video using OpenCV. How do I override the PTS of frames in processed video with PTS from original video. The number of frames between the two is same.

  The processed video is generated from a sequence of images using ffmpeg.

  ffmpeg -i frames/frame_%d.png -pix_fmt yuv420p ffmpeg-cfr.mp4

  This results in a CFR of 25fps. I have tried using the steps from the link, Can I create a VFR video from timestamped images ?. This only works on mp4 videos, and not on webm. The process of encoding to MP4 and converting back to webm results in very bad video quality and takes 3x time for each video.

  Along with the sequence of images, I am also writing the PTS of the original video for each frame in a timecodes.txt file. Are there any options in ffmpeg similar to mp4fpsmod (https://github.com/nu774/mp4fpsmod) ?
  I expect that when I dump block-wise timestamps in webm for original video and processed video, the PTS for each block match between the two.

• Cut AVI video via FFMPEG results in black screen video, but audio is OK

  25 décembre 2017, par mipi

  I want to trim a AVI video (H264 codec) via ffmpeg. The time interval for the result is available as START_TIME_ORIG and DURATION_ORIG (both in microseconds). To make sure that the resulting video starts with an IDR frame, I determine START_TIME and DURATION via ffprobe by executing

  ffprobe -show_frames -pretty -read_intervals [TIME_FROM%TIME_TO] input.avi

  twice to get the IDR frames which are (1st call) closest to START_TIME_ORIG and (2nd call) closest to START_TIME_ORIG+DURATION_ORIG. TIME_FROM and TIME_TO is an interval of 5 seconds plus/minus around (1st call) START_TIME_ORIG and (2nd call) START_TIME_ORIG+DURATION_ORIG. To identify a frame as IDR frame I verify that key_frame=1 and pict_type=I. START_TIME is then set to pkt_dts_time of that frame. In a similar way I calculate DURATION.

  Then ffmpeg is called :

  ffmpeg -ss [START_TIME] -i input.avi -t [DURATION] -codec copy -reset_timestamps 1 -async 1 -map 0 -y output.avi

  Unfortunately the resulting video has a black screen only, audio is OK. What is wrong with my approach ?
  Thanks, mipi