Recherche avancée

Sur d’autres sites (5959)

• How to grab individual frames from streaming video with VLC.DotNet and pass them to OpenCV

  12 février 2018, par user2219675

  I have a RTP video stream (MPEG TS & H264) that does not display well when opened with OpenCV ffmpeg. The video is not decoded well and contains artifacts while it is displayed correctly in VLC. So I thought to grab the video frames with VLC library (VLC DotNet) convert them to Mat and pass them to OpenCV.

  I couldn’t find any similar solutions. The only solution that seems non optimal is to transcode the stream in VLC to MJPEG stream and them open it in OpenCV.

  Does anyone have a better idea ?

• fftools/ffmpeg_enc : apply -top to individual encoded frames

  14 septembre 2023, par Anton Khirnov

  fftools/ffmpeg_enc : apply -top to individual encoded frames
  Fixes #9339.
  

  • [DH] fftools/ffmpeg_enc.c
  • [DH] tests/ref/fate/concat-demuxer-extended-lavf-mxf_d10
  • [DH] tests/ref/fate/concat-demuxer-simple1-lavf-mxf_d10
  • [DH] tests/ref/lavf/mxf_d10

• How to set individual image display durations with ffmpeg-python

  20 septembre 2022, par tompi

  I am using ffmpeg-python 0.2.0 with Python 3.10.0. Displaying videos in VLC 3.0.17.4.

  


  I am making an animation from a set of images. Each image is displayed for different amount of time.

  


  I have the basics in place with inputting images and concatenating streams, but I can't figure out how to correctly set frame duration.

  


  Consider the following example :

  


  stream1 = ffmpeg.input(image1_file)
stream2 = ffmpeg.input(image2_file)
combined_streams = ffmpeg.concat(stream1, stream2)
output_stream = ffmpeg.output(combined_streams, output_file)
ffmpeg.run(output_stream)


  


  With this I get a video with duration of a split second that barely shows an image before ending. Which is to be expected with two individual frames.

  


  For this example, my goal is to have a video of 5 seconds total duration, showing the image in stream1 for 2 seconds and the image in stream2 for 3 seconds.

  


  Attempt 1 : Setting t for inputs

  


  stream1 = ffmpeg.input(image1_file, t=2)
stream2 = ffmpeg.input(image2_file, t=3)
combined_streams = ffmpeg.concat(stream1, stream2)
output_stream = ffmpeg.output(combined_streams, output_file)
ffmpeg.run(output_stream)


  


  With this, I get a video with the duration of a split second and no image displayed.

  


  Attempt 2 : Setting frames for inputs

  


  stream1 = ffmpeg.input(image1_file, frames=48)
stream2 = ffmpeg.input(image2_file, frames=72)
combined_streams = ffmpeg.concat(stream1, stream2)
output_stream = ffmpeg.output(combined_streams, output_file, r=24)
ffmpeg.run(output_stream)


  


  In this case, I get the following error from ffmpeg :

  


  Option frames (set the number of frames to output) cannot be applied to input url ########## -- you are trying to apply an input option to an output file or vice versa. Move this option before the file it belongs to.


  


  I can't tell if this is a bug in ffmpeg-python or if I did it wrong.

  


  Attempt 3 : Setting framerate for inputs

  


  stream1 = ffmpeg.input(image1_file, framerate=1/2)
stream2 = ffmpeg.input(image2_file, framerate=1/3)
combined_streams = ffmpeg.concat(stream1, stream2)
output_stream = ffmpeg.output(combined_streams, output_file)
ffmpeg.run(output_stream)


  


  With this, I get a video with the duration of a split second and no image displayed. However, when I set both framerate values to 1/2, I get an animation of 4 seconds duration that displays the first image for two seconds and the second image for two seconds. This is the closest I got to a functional solution, but it is not quite there.

  


  I am aware that multiple images can be globbed by input, but that would apply the same duration setting to all images, and my images each have different durations, so I am looking for a different solution.

  


  Any ideas for how to get ffmpeg-python to do the thing is much appreciated.