Recherche avancée

Sur d’autres sites (9876)

• How do I use ffmpeg to bulk scale videos of different sizes to the same size (some potentially with black bar padding) ?

  9 décembre 2020, par Tumbleweed53

  I have a bunch of videos I want to stitch together. I want to scale them all to the same size, maintaining aspect ratio (thus adding black bar padding if necessary). Some videos will be downscaled, but most upscaled.

  


  Does anyone have a clever ffmpeg scale formula to accomplish this ? My alternative is to iterate through every video and calculate the scale formula based on its size, but I'd prefer not to do that if I don't have to.

  


• libav / FFMPEG streaming in one encoding, saving as another

  25 février 2018, par user2333829

  I’m using libav on an embedded device. After some effort, I got a fragmented MPEG4 livestream going with it. I am very happy with this now.

  What I’d like to do now is have two outputs from the same input. The first output is what I currently have, which is a web-quality video (lossy @ video rate) being processed in a callback and streamed somewhere. The second output should be a lossless video saved to a file at a much higher framerate.

  I know I can produce the second output independently, but is there a clever way for libav to do these together ? Or do I just do them both independently ?

• C++ Parse data from FFMPEG pipe output

  5 janvier 2017, par Simon

  I want to play around with data coming from an RTSP stream (e.g., do motion detection). Instead of using the cumbersome ffmpeg C API for decoding the stream, I decided to try something different. FFmpeg offers the possibility to pipe its output to stdout. Therefore, I wrote a C++ program which calls popen() to make an external call to ffmpeg and grabs the output. My aim is now to extract the YUV images from the resulting stream. As a first test, I decoded an h264 video file and wrote the stdout output coming from the ffmpeg call to a file.

  #include <iostream>
  
  #include <fstream>
  
  #include 
  
  
  
  using namespace std;
  
  
  
  int main()
  
  {
  
    FILE *in;
  
    char buff[512];
  
  
  
    if(!(in = popen("ffmpeg -i input.mp4 -c:v rawvideo -an -r 1 -f rawvideo pipe:1", "r")))
  
    {
  
      return 1;
  
    }
  
  
  
    ofstream outputFile("output.yuv");
  
    while(fgets(buff, sizeof(buff), in) != NULL)
  
    {
  
      outputFile &lt;&lt; buff;
  
    }
  
  
  
    outputFile.close();
  
    pclose(in);
  
    return 0;
  
  }
  
  </fstream></iostream>

  The resulting raw video can be played with vlc afterwards :

  vlc --rawvid-fps 1 --rawvid-width 1280 --rawvid-height 544 --rawvid-chroma I420 output.yuv

  Here, I chose the width and height from the video (a trailer of the Simpsons movie from http://www.dvdloc8.com/clip.php?movieid=12167&clipid=3). This first test worked very well. The resulting file is the same as when calling the ffmpeg binary directly with

  ffmpeg -i input.mp4 -c:v rawvideo -an -f output_ffmpeg.yuv

  Now I want to do some processing with the images coming from the ffmpeg output instead of dumping it to a file. My question is : Is there a clever way of parsing the stdout data from ffmpeg ? Ideally, I want to parse the stream into a sequence of instances of a YUVImage class (for example).