Recherche avancée

Sur d’autres sites (8668)

• Re-sampling H264 video to reduce frame rate while maintaining high image quality

  4 mars 2019, par BrianTheLion

  Here’s the mplayer output for a video of interest :

  br@carina:/tmp$ mplayer foo.mov 
  
  mplayer: Symbol `ff_codec_bmp_tags' has different size in shared object, consider re-linking
  
  MPlayer 1.0rc4-4.5.2 (C) 2000-2010 MPlayer Team
  
  mplayer: could not connect to socket
  
  mplayer: No such file or directory
  
  Failed to open LIRC support. You will not be able to use your remote control.
  
  
  
  Playing foo.mov.
  
  libavformat file format detected.
  
  [lavf] stream 0: video (h264), -vid 0
  
  [lavf] stream 1: audio (aac), -aid 0, -alang eng
  
  VIDEO:  [H264]  1280x720  24bpp  59.940 fps  2494.2 kbps (304.5 kbyte/s)
  
  ==========================================================================
  
  Opening video decoder: [ffmpeg] FFmpeg's libavcodec codec family
  
  Selected video codec: [ffh264] vfm: ffmpeg (FFmpeg H.264)
  
  ==========================================================================
  
  ==========================================================================
  
  Opening audio decoder: [faad] AAC (MPEG2/4 Advanced Audio Coding)
  
  AUDIO: 44100 Hz, 2 ch, s16le, 128.0 kbit/9.07% (ratio: 15999->176400)
  
  Selected audio codec: [faad] afm: faad (FAAD AAC (MPEG-2/MPEG-4 Audio))
  
  ==========================================================================
  
  AO: [pulse] 44100Hz 2ch s16le (2 bytes per sample)
  
  Starting playback...
  
  Movie-Aspect is 1.78:1 - prescaling to correct movie aspect.
  
  VO: [vdpau] 1280x720 => 1280x720 Planar YV12

  I’d like to use ffmpeg, mencoder, or some other command-line video transcoder to re-sample this video to a lower framerate without loss of image quality. That is, each frame should remain as crisp as possible.

  Attempts

  ffmpeg -i foo.mov -r 25 -vcodec copy bar.mov

  • The target frame rate — 25fps — is achieved but individual frames are "blocky."

  mencoder -nosound -ovc copy foo.mov -ofps 25 -o bar.mov

  • Videos are effectively un-viewable.

  Help !

  This seems like a simple enough use case. I’m very surprised that obvious things are not working. Is there something wrong with my approach ?

• ffmpeg png to png quality loss

  13 avril 2018, par kilo

  I did a python script that managed to unshuffle a shuffled (png) image according to a specific pattern, that python script uses ffmpeg and does 12 encodes to unshuffling it (by cropping a specific part and pasting it over the existing picture).
  As such the same file is re-encoded into a new file each time, which shouldn’t be a problem since i am doing png conversion (lossless, right ?), but i still lose quality on it.

  Here are the pictures :

  • https://i.imgur.com/TMzBUGq.png (input)
  • https://i.imgur.com/FCwfViJ.png (output)

  Notice the loss of quality on the "ONE PUNCH MAN" text. The rest of the picture is, seemingly, literally identical. So the problem seems to be with the colors.

  Here are the ffmpeg commands i ran to get to the output :

  ffmpeg -loglevel panic -y -i "output/001.png" -i "001.png" -qscale:v 2 -filter_complex "[0:v]crop=200:280:200:0[t];[0:v][t]overlay=0:280" "output/001.png"
  
  ffmpeg -loglevel panic -y -i "output/001.png" -i "001.png" -qscale:v 2 -filter_complex "[0:v]crop=200:280:400:0[t];[0:v][t]overlay=0:560" "output/001.png"
  
  ffmpeg -loglevel panic -y -i "output/001.png" -i "001.png" -qscale:v 2 -filter_complex "[0:v]crop=200:280:600:0[t];[0:v][t]overlay=0:840" "output/001.png"
  
  ffmpeg -loglevel panic -y -i "output/001.png" -i "001.png" -qscale:v 2 -filter_complex "[1:v]crop=200:280:0:280[t];[0:v][t]overlay=200:0" "output/001.png"
  
  ffmpeg -loglevel panic -y -i "output/001.png" -i "001.png" -qscale:v 2 -filter_complex "[0:v]crop=200:280:400:280[t];[0:v][t]overlay=200:560" "output/001.png"
  
  ffmpeg -loglevel panic -y -i "output/001.png" -i "001.png" -qscale:v 2 -filter_complex "[0:v]crop=200:280:600:280[t];[0:v][t]overlay=200:840" "output/001.png"
  
  ffmpeg -loglevel panic -y -i "output/001.png" -i "001.png" -qscale:v 2 -filter_complex "[1:v]crop=200:280:0:560[t];[0:v][t]overlay=400:0" "output/001.png"
  
  ffmpeg -loglevel panic -y -i "output/001.png" -i "001.png" -qscale:v 2 -filter_complex "[1:v]crop=200:280:200:560[t];[0:v][t]overlay=400:280" "output/001.png"
  
  ffmpeg -loglevel panic -y -i "output/001.png" -i "001.png" -qscale:v 2 -filter_complex "[0:v]crop=200:280:600:560[t];[0:v][t]overlay=400:840" "output/001.png"
  
  ffmpeg -loglevel panic -y -i "output/001.png" -i "001.png" -qscale:v 2 -filter_complex "[1:v]crop=200:280:0:840[t];[0:v][t]overlay=600:0" "output/001.png"
  
  ffmpeg -loglevel panic -y -i "output/001.png" -i "001.png" -qscale:v 2 -filter_complex "[1:v]crop=200:280:200:840[t];[0:v][t]overlay=600:280" "output/001.png"
  
  ffmpeg -loglevel panic -y -i "output/001.png" -i "001.png" -qscale:v 2 -filter_complex "[1:v]crop=200:280:400:840[t];[0:v][t]overlay=600:560" "output/001.png"

  Anyone got any idea why is there this quality loss ?
  Strangely enough, there is no quality loss when i do it in an entirely different way (crop each square into an individual file, then each of them are put into a 1x2 vstack with the next one, then each of the resulting 1x2 files are vstacked with a second file to make a 1x4 file, then each of those are hstacked to make a 2x4 file, and finally we hstack the two resulting file for the resulting 4x4 output), even though there is more than double the amount of encodes.

• Produce waveform video from audio using FFMPEG

  30 novembre 2020, par RhythmicDevil

  I am trying to create a waveform video from audio. My goal is to produce a video that looks something like this

  


  

  


  For my test I have an mp3 that plays a short clipped sound. There are 4 bars of 1/4 notes and 4 bars of 1/8 notes played at 120bpm. I am having some trouble coming up with the right combination of preprocessing and filtering to produce a video that looks like the image. The colors dont have to be exact, I am more concerned with the shape of the beats. I tried a couple of different approaches using showwaves and showspectrum. I cant quite wrap my head around why when using showwaves the beats go past so quickly, but using showspectrum produces a video where I can see each individual beat.

  


  ShowWaves

  


  ffmpeg -i beat_test.mp3 -filter_complex "[0:a]showwaves=s=1280x100:mode=cline:rate=25:scale=sqrt,format=yuv420p[v]" -map "[v]" -map 0:a output_wav.mp4


  


  This link will download the output of that command.

  


  ShowSpectrum

  


  ffmpeg -i beat_test.mp3 -filter_complex "[0:a]showspectrum=s=1280x100:mode=combined:color=intensity:saturation=5:slide=1:scale=cbrt,format=yuv420p[v]" -map "[v]" -an -map 0:a output_spec.mp4


  


  This link will download the output of that command.

  


  I posted the simple examples because I didn't want to confuse the issue by adding all the variations I have tried.

  


  In practice I suppose I can get away with the output from showspectrum but I'd like to understand where/how I am thinking about this incorrectly. Thanks for any advice.

  


  Here is a link to the source audio file.