Recherche avancée

Sur d’autres sites (6726)

• FFmpeg does not always fully concat video ?

  25 juin 2014, par user3550401

  I’m trying to concat two video clips, sometimes i end up with only three fourths of the video, sometimes I only get one half, and sometimes i manage to get the whole thing. Anyone ever deal with this ? Here is my ffmpeg command :

  ffmpeg -i "concat:videos/clip1.ts|videos/clip2.ts"  -ar 44100  -b:a 128k -b:v 800k -vcodeclibx264 -bsf:v h264_mp4toannexb -f mpegts -t 120 videos/output.ts

  I just dont understand why sometimes it works perfectly and other times i only get part of the video.

• FFmpeg only reading one channel

  29 août 2017, par AbstractDissonance

  Trying to read a stereo wav file and ffmpeg is only reading one channel. ffprobe returns 2 channels.

  I have used -ac 2 and then added -channel_layout stereo but all return 1 channel(or basically filling half the buffer I created).

  Basically the output size of ffmpeg is about half the wav file size.

  What I would like is for it to return every channel in

  Channel1_sample1, channel2_sample1, ..., ChannelN_sample1, channel1_sample2, etc...

  But, in reality, I’d rather just have it work with stereo ;) I’m giving it plenty large enough buffer to read to, so that isn’t the problem either.

  Here is the output

  ffprobe.exe -hide_banner -v quiet -print_format flat -show_streams -i temp.wav
  
  streams.stream.0.index=0
  
  streams.stream.0.codec_name="pcm_s16le"
  
  streams.stream.0.codec_long_name="PCM signed 16-bit little-endian"
  
  streams.stream.0.profile="unknown"
  
  streams.stream.0.codec_type="audio"
  
  streams.stream.0.codec_time_base="1/44100"
  
  streams.stream.0.codec_tag_string="[1][0][0][0]"
  
  streams.stream.0.codec_tag="0x0001"
  
  streams.stream.0.sample_fmt="s16"
  
  streams.stream.0.sample_rate="44100"
  
  streams.stream.0.channels=2
  
  streams.stream.0.channel_layout="unknown"
  
  streams.stream.0.bits_per_sample=16
  
  streams.stream.0.id="N/A"
  
  streams.stream.0.r_frame_rate="0/0"
  
  streams.stream.0.avg_frame_rate="0/0"
  
  streams.stream.0.time_base="1/44100"
  
  streams.stream.0.start_pts="N/A"
  
  streams.stream.0.start_time="N/A"
  
  streams.stream.0.duration_ts=14200200
  
  streams.stream.0.duration="322.000000"
  
  streams.stream.0.bit_rate="1411200"
  
  streams.stream.0.max_bit_rate="N/A"
  
  streams.stream.0.bits_per_raw_sample="N/A"
  
  streams.stream.0.nb_frames="N/A"
  
  streams.stream.0.nb_read_frames="N/A"
  
  streams.stream.0.nb_read_packets="N/A"
  
  streams.stream.0.disposition.default=0
  
  streams.stream.0.disposition.dub=0
  
  streams.stream.0.disposition.original=0
  
  streams.stream.0.disposition.comment=0
  
  streams.stream.0.disposition.lyrics=0
  
  streams.stream.0.disposition.karaoke=0
  
  streams.stream.0.disposition.forced=0
  
  streams.stream.0.disposition.hearing_impaired=0
  
  streams.stream.0.disposition.visual_impaired=0
  
  streams.stream.0.disposition.clean_effects=0
  
  streams.stream.0.disposition.attached_pic=0
  
  streams.stream.0.disposition.timed_thumbnails=0
  
  ffmpeg.exe -i temp.wav -loglevel quiet -f s16le -ac 2 -channel_layout stereo -

  temp.wav is just a standard stereo wav file.

• FFmpeg python : different length after fps changed and interpolated

  8 mai 2022, par Pablo Dawson

  For my application I'm adapting the code from the ffmpeg-python github for tensorflow streaming.

  


  It basically decodes each frame of your input, lets you process it with python, and encodes it again.

  


  To optimize things I'm adding a fps filter at the input to get half the fps, so I can process only half the frames, then interpolate frames in the encoder with minterpolate to get the original fps.

  


  My decoder looks like this :

  


  def decoder(in_filename):
args = (
    ffmpeg
    .input(in_filename)
    .filter('fps', fps=30/2)
     (... more filters in between ...)
    .output('pipe:', format='rawvideo', pix_fmt='rgb24')
    .compile()
)
return subprocess.Popen(args, stdout=subprocess.PIPE)


  


  And my encoder after processing the frames with python :

  


  def encoder(out_filename, width, height):

args = (
    ffmpeg
    .input('pipe:', format='rawvideo', pix_fmt='rgb24', s='{}x{}'.format(width, height))
    .filter('minterpolate', fps=30)
    .filter('fps',fps=30)
    .output(out_filename, pix_fmt='rgb24')
    .overwrite_output()
    .compile()
)
return subprocess.Popen(args, stdin=subprocess.PIPE)


  


  After that I horizontally stack the original input with the processed video

  


  subprocess.run("ffmpeg -i {} -i {} -filter_complex hstack=inputs=2 -pix_fmt yuv420p -c:v libx264 {}".format(in_filename,out_filename,"out.mp4"))


  


  Here's the problem : The "processed" video is faster and ends before the original. It's like they follow the same timestamp but it never actually interpolated frames. What am I doing wrong ?