Recherche avancée

Sur d’autres sites (8994)

• ffmpeg concatenate a large number of videos with different codecs and resolutions

  21 avril 2024, par Paul H

  The videos I am trying to concatenate can have different technical metadata, like codec, resolution, etc. So I wrote a filtergraph that works for my purposes. My complete command can look like this :

  



  ffmpeg -i "vid1.mp4" -i "vid2.mp4" -i "vid3.mp4" -filter_complex_script "filtergraph.txt" -map "[outv]" -map "[outa]" -c:v libx264 -r 60 -preset medium -crf 24 -c:a aac -b:a 160k "output.mp4"


  



  And "filtergraph.txt" looks like this (autogenerated by my own script before). I have the filters to change all input resolutions to Full HD. (Newlines in the following snippet are just introduced for readability, they are not contained in my filtergraph.txt)

  



  [0:v]scale=1920:1080:force_original_aspect_ratio=decrease,pad=1920:1080:(ow-iw)/2:(oh-ih)/2,setsar=1[v0];
[1:v]scale=1920:1080:force_original_aspect_ratio=decrease,pad=1920:1080:(ow-iw)/2:(oh-ih)/2,setsar=1[v1];
[2:v]scale=1920:1080:force_original_aspect_ratio=decrease,pad=1920:1080:(ow-iw)/2:(oh-ih)/2,setsar=1[v2];
[v0][0:a][v1][1:a][v2][2:a]concat=n=3:v=1:a=1[outv][outa]


  



  Now in this case, it works, I have the three source videos, they can have different resolutions, codecs or even different framerates - they are all combinded into one Full HD 60 FPS output video.

  



  The problem is, that I want to concatenate a large number of video files and get the error, that the command contains too many characters for cmd. That's why I already put the filtergraph to a seperate file and read it using -filter_complex_script which works.

  



  But it can still get too long if I have a few hundred source files that I want to concat.

  



  As far as I tested it, I can't use -f concat -i listOfFiles.txt because every source file would have to be in the same codec, resolution etc then, and I could not use the filtergraph for scaling due to the fact that ffmpeg would try to concat before demuxed output goes into filter.

  



  So, how could I concat hundreds of videos while resizing them all to Full HD and applying the other filters, but keep the cmd line short ?

  



  Is there something similar to -filter_complex_script but for input files ? Or will I have to encode each source video to my desired target format first and then concat them with stream copy ?

  


• FFMPEG Ensuring Meta Data Is Correct

  11 octobre 2020, par Devin Dixon

  I have to take any kind of video format (mp4, mov, flv,etc) and convert it h264 mp4. My command currently looks like this :

  


  ffmpeg -i input_video.mov -pix_fmt yuv420p  -crf 20 -acodec aac -strict -2 -b:a 384k -vcodec libx264 -f mp4 -movflags faststart -max_muxing_queue_size 9999 out_video.mp4


  


  Sometimes the meta data is not copied correctly, and this causes issue especially when streaming. Is there a fool proof way of copying the meta from the original file or recreating the meta ?

  


• FFMPEG : When utilizing a filter complex script on large duration and sized videos, it eats up the RAM entirely. Request a fix

  13 octobre 2023, par Sree

  I am using ffmpeg to blur certain areas of a large video file using boxblur.

  


  Using over 1000 redactions in the filter script, When I generate a ffmpeg command to apply these complex filters( create redaction areas and then overlay them) on a 1 GB video, I get ffmpeg is out of resources, ram gets used totally by the redactions and the overlays. Thus the output video is not obtained.

  


  I tried two ways :
a. First way using split to create 1000 streams, create redaction areas from them using boxblur and then overlay on top of each other in cascading effect. This generates out of memory as soon as it is run.

  


  b. In the second way, I do not split into streams at the start but generate redaction areas from the original input video and then overlay them on top of each other in cascading way. This slowly eats up the ram and breaks.

  


  Request help :

  


  a. For large video size(>1gb) and large number of redactions(>1000) is there an efficient way to generate ffmpeg filter_script so it does not use my ram entirely and outputs the result in lesser time.

  


  b. Are there any alternate ways to do this using ffmpeg.