Recherche avancée

Sur d’autres sites (6560)

• ffmpeg filter dev get_video_buffer function usage

  19 juin 2018, par Michael.Ma

  I’m writing a mosaic ffmpeg filter.
  copy code from vf_vflip.c, I register a callback function get_video_buffer
  and implement it as follows :

  static AVFrame *get_video_buffer(AVFilterLink *link, int w, int h)
  
  {
  
      MosaicContext *s = link->dst->priv;
  
      AVFrame *frame;
  
      int i;
  
      int j,k;
  
  
  
      frame = ff_get_video_buffer(link->dst->outputs[0], w, h);
  
      if (!frame)
  
          return NULL;
  
  
  
      if (s->w == 0 || s->h == 0)
  
          return frame;
  
  
  
      for (i = 0; i < 3; i ++) {
  
          // some trick code
  
      }
  
  
  
      return frame;
  
  }
  
  
  
  static const AVFilterPad avfilter_vf_mosaic_inputs[] = {
  
      {
  
          .name             = "default",
  
          .type             = AVMEDIA_TYPE_VIDEO,
  
          //.get_video_buffer = get_video_buffer,
  
          .filter_frame     = filter_frame,
  
          .config_props     = config_input,
  
      },
  
      { NULL }
  
  };

  But after read some filter dev guide, I found that only a few filter implement get_video_buffer funcition.The default function is f_default_get_video_buffer()
  I make some log find that ffmpeg call function like

  get_video_buffer
  
  filter_frame
  
  get_video_buffer
  
  filter_frame
  
  get_video_buffer
  
  filter_frame
  
  ...

  I’m confused about get_video_buffer do what job.
  Only know that some filter will hold frame cache.
  Can I juse comment out the get_video_buffer hook ?

  I see flip job both in get_video_buffer() and filter_frame() in vf_vflip.c
  Is it a waste of time ?

  static AVFrame *get_video_buffer(AVFilterLink *link, int w, int h)
  
  {
  
      FlipContext *flip = link->dst->priv;
  
      AVFrame *frame;
  
      int i;
  
  
  
      frame = ff_get_video_buffer(link->dst->outputs[0], w, h);
  
      if (!frame)
  
          return NULL;
  
  
  
      for (i = 0; i < 4; i ++) {
  
          int vsub = i == 1 || i == 2 ? flip->vsub : 0;
  
          int height = AV_CEIL_RSHIFT(h, vsub);
  
  
  
          if (frame->data[i]) {
  
              frame->data[i] += (height - 1) * frame->linesize[i];
  
              frame->linesize[i] = -frame->linesize[i];
  
          }
  
      }
  
  
  
      return frame;
  
  }
  
  
  
  static int filter_frame(AVFilterLink *link, AVFrame *frame)
  
  {
  
      FlipContext *flip = link->dst->priv;
  
      int i;
  
  
  
      for (i = 0; i < 4; i ++) {
  
          int vsub = i == 1 || i == 2 ? flip->vsub : 0;
  
          int height = AV_CEIL_RSHIFT(link->h, vsub);
  
  
  
          if (frame->data[i]) {
  
              frame->data[i] += (height - 1) * frame->linesize[i];
  
              frame->linesize[i] = -frame->linesize[i];
  
          }
  
      }
  
  
  
      return ff_filter_frame(link->dst->outputs[0], frame);
  
  }

• How can I best utilize an AWS service to segment a video into smaller chunks and then combine them back to together ? [on hold]

  19 avril 2018, par Justin Malin

  I am trying to do processing on videos uploaded to AWS S3 using an AWS Lambda function in Python. However, FFmpeg and ffmpeg-python (as far as I am aware) are unable to process objects and must do processing on stored files. Lambda only allows for 500 MB of storage in the /tmp/ folder, thus limiting the size of video that I can do processing on.

  If there is an alternative to FFmpeg that allows me to work on object files that I am unaware of, that would be a reasonable solution because I can scale up the memory of the Lambda function (although there is still a limit).

  Alternatively, I have looked into segmenting the video using AWS Elastic Transcoder, but I do not think I can dynamically segment the video using that service. If there is a service similar to this that could segment the video into individual frames (and back), that would be even better.

  I have also considered using AWS EC2, but I would only be using the EC2 service to segment videos sporadically, so it would be a waste to constantly have a server that capable running. If I use the AWS Elastic Beanstalk, would it automatically start a more powerful instance of EC2 to do the video segmentation (and reformation) when that is called and revert back to a much smaller instance when dormant ?

  Essentially, I would like to know if there are any services (preferably within AWS) that allow me to segment a video into shorter videos or into each frame at-will.

• On upload rename the file to input.mp4 not working

  7 mars 2018, par 57_Wolve

  Ok so i’m trying to make it to were the file is uploaded then renamed to input.mp4 so that it over rights the old one. But its not working ?

  But it still keeps uploading with the original file name.
  Or is there a way to just delete the files in the folder uploads after 30 min ? Both the uploaded file and output file.

  <?php
  
  $fileName = $_FILES["file1"]["name"]; // The file name
  
  $fileTmpLoc = $_FILES["file1"]["tmp_name"]; // File in the PHP tmp folder
  
  $fileType = $_FILES["file1"]["type"]; // The type of file it is
  
  $fileSize = $_FILES["file1"]["size"]; // File size in bytes
  
  $fileErrorMsg = $_FILES["file1"]["error"]; // 0 for false... and 1 for true
  
  if (!$fileTmpLoc) { // if file not chosen
  
      echo "ERROR: Please browse for a file before clicking the upload button.";
  
      exit();
  
  }
  
  
  
      $newfilename = $fileName . input.mp4';
  
      rename($fileName, $newfilename);
  
  
  
  if(move_uploaded_file($fileTmpLoc, "uploads/$newfilename")){
  
  
  
    echo "Starting ffmpeg... <br />";
  
    echo shell_exec("ffmpeg -y -i uploads/".$newfilename." uploads/output.mp3");
  
    echo "Done. <br /><br />";
  
  
  
  echo '<a href="http://test.tw-wcs.com:82/ffmpeg/MP4_To_MP3/uploads/output.mp3" target="_blank">Click Here</a> To open your file in a new tab.';
  
  
  
  } else {
  
      echo "move_uploaded_file function failed";
  
  }
  
  ?>