Recherche avancée

Sur d’autres sites (7381)

• How to record all dynamic videos in the process of capturing screen video with ffmpeg from a python program ?

  4 décembre 2020, par fengnix

  I have many robotframework test cases and in the first case, a ffmpeg command like the following is invoked to record the whole running process :

  


  ffmpeg -framerate 30 -f gdigrab -i desktop -c:v libx264rgb -crf 0 -preset ultrafast output.mkv


  


  Whenever I firstly run all cases and then manuually run the above command from an addition command console, the recorded video always looks fine, it looks like all contents on the screen can be correctly captured.

  


  However, once I execute the command the same as the above one in the first case by call the following code :

  


  p=subprocess.Popen(cmd, shell=True, stdin=subprocess.PIPE, stdout=subprocess.PIPE, stderr=subprocess.PIPE)


  


  and then in the final test case the record process is stopped by calling the following code to tell ffmpeg that we want to stop the recording :

  


  p.stdin.write(bytes("q",'UTF-8'))  


  


  the final result video only contain correct contents of the "start" and the "end" of the whole process, but all other contents no longer changed and seemd just a static image, which means all the dynamic effects on the screen cannot be captured.

  


  Could anyone be so kind as to let me know what the matter is and how to solve it ?

  


• Crop video into a 4x4 grid/tiles/matrix efficiently via command-line ffmpeg ?

  22 avril 2017, par Dylan

  Hello Stackoverflow community !

  I dread having to ask questions, but there seems to be no efficient way to take a single input video and apply a matrix transformation/split the video into equal sized pieces, preferably 4x4=16 segments per input.

  I tried using all the libraries such as ffmpeg and mencoder, but having 16 outputs can be as slow as 0.15x. The goal of my project is the split the video into 16 segments, rearrange those segments and combine back into a final video ; later reversing the process in HTML5 canvas. Here is a picture to help you understand what I am talking about :
  the source but also the final destination after reorganizing the pieces

  after the segments are exported, I will recombine the pieces in a puzzle manner (they are also inverted and rotated but unrelated)

  I do not believe you can do this all in one command, so my goal is to crop into 16 mapped outputs quickly, then reassemble them in a different order. But I can do the other parts myself. Ideally there would be a way to move pixel blocks eg 100x100 and just move them around. My math is not strong enough..
  I really appreciate the work you guys do !

  admin@dr.com

• FFMPEG - Overlaying video with transparency, alpha has strange threshold

  6 mars 2019, par tanker_

  I’m attempting to overlay an Apple ProRes 4444 video with a alpha/transparency onto a normal video. However, upon inspecting the final output from FFMPEG, compared to the same files overlaid on top of one another and rendered in Final Cut Pro, there is a discrepancy in how they render the edges around the object.

  Screenshot comparison :

  

  Additional closeup :

  

  Here is my input :

  ffmpeg \
  
      -i background.MOV -x264opts colormatrix=bt709 \
  
      -i alpha_object.MOV -x264opts colormatrix=bt709 \
  
      -filter_complex " \
  
          [0:v]setpts=PTS-STARTPTS, scale=1920x1080[top]; \
  
          [1:v]setpts=PTS-STARTPTS, scale=1920x1080, \
  
               colorchannelmixer=aa=1.0[bottom]; \
  
          [top][bottom]overlay=shortest=1" \
  
      -vcodec libx264 -qp 15 -an -shortest output.MOV

  Any idea what could be wrong ? Is there a option within FFMPEG’s available filters that I’m missing ?

  All files are ingested and rendered in 1920x1080.