Recherche avancée

Sur d’autres sites (6968)

• FFMPEG Determine average color of an area of a video

  12 novembre 2019, par Naved Khan

  I have a use case where I’d want to insert one of two watermarks - one designed for a dark-ish background, the other for a light background into a video. Let’s say that I’d want to do this on the top right corner of the video.

  How do I determine the average color of the top right section of the video ? Post this, how do I determine which watermark to use by looking at the average color ?

  I have a solution right now where I am taking equally spaced screenshots and then measuring the average color, but it’s excruciatingly slow, especially for longer videos.

  # Calculate average color
  
      black_distances = []
  
      white_distances = []
  
  
  
      movie = FFMPEG::Movie.new(video_file)
  
      (0..movie.duration / 10).each do |second|
  
  
  
        # extract a frame
  
        filename = "tmp/watermark/#{SecureRandom.uuid}.jpg"
  
        movie.screenshot filename.to_s, seek_time: second
  
  
  
        # analyse frame for color distance
  
        frame = MiniMagick::Image.open(filename)
  
        frame.crop('20%x20%+80%+0')
  
        frame.resize('1x1')
  
        pixel = frame.get_pixels.flatten
  
  
  
        distance_from_black = Math.sqrt(((black[0] - pixel[0])**2 + (black[1] - pixel[1])**2 + (black[2] - pixel[2])**2))
  
        distance_from_white = Math.sqrt(((white[0] - pixel[0])**2 + (white[1] - pixel[1])**2 + (white[2] - pixel[2])**2))
  
  
  
        black_distances.push distance_from_black
  
        white_distances.push distance_from_white
  
  
  
        File.delete(filename) if File.exist?(filename)
  
      end
  
  
  
      average_black_distance = black_distances.reduce(:+).to_f / black_distances.size
  
      average_white_distance = white_distances.reduce(:+).to_f / white_distances.size

  I am also confused about how to use the resulting average_black_distance and average_white_distance to determine which watermark to use.

• libavfilter/dnn : determine dnn output during execute_model instead of set_input_output

  25 avril 2019, par Guo, Yejun

  libavfilter/dnn : determine dnn output during execute_model instead of set_input_output
  Currently, within interface set_input_output, the dims/memory of the tensorflow
  
  dnn model output is determined by executing the model with zero input,
  
  actually, the output dims might vary with different input data for networks
  
  such as object detection models faster-rcnn, ssd and yolo.
  This patch moves the logic from set_input_output to execute_model which
  
  is suitable for all the cases. Since interface changed, and so dnn_backend_native
  
  also changes.
  In vf_sr.c, it knows it's srcnn or espcn by executing the model with zero input,
  
  so execute_model has to be called in function config_props
  Signed-off-by : Guo, Yejun <yejun.guo@intel.com>
  
  Signed-off-by : Pedro Arthur <bygrandao@gmail.com>
  

  • [DH] libavfilter/dnn_backend_native.c
  • [DH] libavfilter/dnn_backend_native.h
  • [DH] libavfilter/dnn_backend_tf.c
  • [DH] libavfilter/dnn_backend_tf.h
  • [DH] libavfilter/dnn_interface.h
  • [DH] libavfilter/vf_sr.c

• ffmpeg trim mp3 - determine precisely the start and end times of section to be trimmed

  4 février 2019, par Ahmed Khalil

  I have a long mp3 track of an audio book (more than 9 hours long) that I would like to trim using ffmpeg.

  The sample code below is used to trim an mp3 section by providing the start and end times. However, when I determine the start and end times, then checking the output file, it’s not as precisely as I want, sometimes several minutes ahead/before the desired point.

  import subprocess
  
  file = r'audio book.mp3'
  
  track_name = "trimmed section"
  
  output = r'D:\{0}'.format(track_name)
  
  start = '01:26:04'
  
  end = '01:33:17'
  
  
  
  d = subprocess.getoutput('ffmpeg -i "{0}" -ss {1} -to {2} -c copy {3}.mp3"'
  
                       .format(file, start, end, output))
  
  
  
  print(d)

  Is there a way to determine with accuracy the real start and end time of an mp3 audio track, to be given afterwards as inputs to the code...to trim the desired sections all at once, without the need to adjust/fine-tune the start and end time manually ??