Recherche avancée

Sur d’autres sites (6089)

• Split mixed content into sections by scan type using FFmpeg ?

  7 août 2023, par Vivian

  Trying to deinterlace and detelecine "Star Trek : Voyager" and "Star Trek : Deep Space Nine" NTSC DVD rips made using MakeMKV without changing the total number of frames. The videos are a variable frame rate combination of interlaced, telecined, and progressive frames.

  


  I'd like to be able to convert each episode to a sequence of video clips, split by scan type, so I can then feed those individually into Selur's Hybrid for processing, then stitch them back together and feed the reconstructed video into Topaz Video AI.

  


  Is there a way to use FFmpeg to split a video into sections by scan type, and if so, how ?

  


  I've tried automated handling in Hybrid and HandBrake, but both created artifacts and left unfixed issues as a result of applying filters to frames that didn't need them as well as not applying filters to frames that did.

  


• Merge .mkv files with already included subtitles using ffmpeg-python

  9 mars 2023, par Eeshe

  I'm trying to merge two (and more in the future) .mkv files that already have subtitles included in them, this means I DO NOT have .srt files that need to be encoded. Let's call the .mkv files intro.mkv and episode.mkv.

  


  However, the way I want to merge them is so a portion of the episode.mkv file is played until a specified timestamp, then intro.mkv plays completely, and after that the rest of episode.mkv.

  


  I've managed to get close with the following code.

  


  import ffmpeg

initial_episode_video = ffmpeg.input('episode.mkv', to='00:00:34.033', hwaccel='cuvid')
intro_video = ffmpeg.input('intro.mkv', hwaccel='cuvid')
remaining_episode_video = ffmpeg.input('episode.mkv', ss='00:00:34.033', hwaccel='cuvid')

full_episode_video = ffmpeg.concat(initial_episode_video, intro_video, remaining_episode_video)

output_video = ffmpeg.output(
    full_episode_video,
    filter_complex='[0:v][0:a][1:v][1:a][2:v][2:a]concat=n=3:v=1:a=1',
    vprofile='main', 
    vcodec='h264_nvenc',
    pix_fmt='yuv420p',
    crf=22,
    map="0:s",
    filename=f'Output/{episode_path}')
ffmpeg.run(output_video)


  


  With this code, the videos are correctly trimmed. The problem is that once intro.mkv starts playing, it doesn't show the subtitles corresponding to it, instead it shows the episode.mkv subtitles corresponding to the '00:00:34.033' timestamp.

  


  My objective is to display the subtitles of intro.mkv, and then, once the full intro is played, continue with the subtitles from episode.mkv.

  


  NOTE : I am aware that I am re-encoding both the episode.mkv and intro.mkv files. If possible, I'd prefer if it only re-encodes episode.mkv, since intro.mkv only needs to be re-encoded once (I can do that individually). I am re-encoding to go from H.264 Hi 10 to H.264 main profile.

  


• ffmpeg adds silence when used with input text file

  1er octobre 2022, par AncientToaster

  I'm using ffmpeg's concat option to combine a series of still images with an mp3 file. The total duration of my input.txt file is 12532 seconds, which is exactly the same as the total duration of my mp3 file (3:28:52, or 12532 seconds).

  


  But ffmpeg is adding 200 seconds of audio-free content to the end of the video. I've tried totaling and re-totaling the total duration of input.txt and it all seems to add up to the same amount. Including or excluding the -shortest flag also doesn't seem to matter.

  


  


  • ffmpeg command :

    


    ffmpeg -f concat -i input.txt -i input.mp3 -shortest -c:v libx264 -r 1 -pix_fmt yuv420p output.mp4


    


  


  • input.txt source

    


  


  • input.mp3

    


  


  • output file

    


  


  


  I'm having the same issue (differing amounts of time) with other videos.