Recherche avancée

Sur d’autres sites (8402)

• mov : Support mdcv and clli boxes for mastering display an color light level

  4 octobre 2017, par Vittorio Giovara

  mov : Support mdcv and clli boxes for mastering display an color light level
  Signed-off-by : Vittorio Giovara <vittorio.giovara@gmail.com>
  

  • [DH] libavformat/mov.c

• Haskell - Turning multiple image-files into one video-file using the ffmpeg-light package

  25 avril 2021, par oRole

  Background
  
I wrote an application for image-processing which uses the ffmpeg-light package to fetch all the frames of a given video-file so that the program afterwards is able to apply grayscaling, as well as edge detection alogrithms to each of the frames.
  

  


  Now I'm trying to put all of the frames back into a single video-file.

  


  Used Libs
  
ffmpeg-light-0.12.0
  
JuicyPixels-3.2.8.3
  
...

  


  What have I tried ?
  
I have to be honest, I didn't really try anything because I'm kinda clueless where and how to start. I saw that there is a package called Command which allows running processes/commands using the command line. With that I could use ffmpeg (not ffmpeg-light) to create a video out of image-files which I would have to save to the hard drive first but that would be kinda hacky.

  
Within the documentation of ffmpeg-light on hackage (ffmpeg-light docu) I found the frameWriter function which sounds promising.

  


  frameWriter :: EncodingParams -> FilePath -> IO (Maybe (AVPixelFormat, V2 CInt, Vector CUChar) -> IO ()) 


  


  I guess FilePath would be the location where the video file gets stored but I can't really imagine how to apply the frames as EncodingParams to this function.

  


  Others
  
I can access :
  

  


  

  • r, g, b, a as well asy. a values
    

  


  • image width / height / format

  


  


  Question
  
Is there a way to achieve this using the ffmpeg-light package ?

  


  As the ffmpeg-light package lacks of documentation when it comes to conversion from images to video, I really would appreciate your help. (I do not expect a fully working solution.)

  


  Code
  
The code that reads the frames :

  


  -- Gets and returns all frames that a given video contains
getAllFrames :: String -> IO [(Double, DynamicImage)]
getAllFrames vidPath = do 
  result <- try (imageReaderTime $ File vidPath) :: IO (Either SomeException (IO (Maybe (Image PixelRGB8, Double)), IO()))
  case result of 
    Left ex -> do 
                 printStatus "Invalid video-path or invalid video-format detected." "Video" 
                 return []
    Right (getFrame, _) -> addNextFrame getFrame [] 

-- Adds up all available frames to a video.
addNextFrame :: IO (Maybe (Image PixelRGB8, Double)) -> [(Double, DynamicImage)] -> IO [(Double, DynamicImage)]
addNextFrame getFrame frames = do
  frame <- getFrame
  case frame of 
    Nothing -> do 
                 printStatus "No more frames found." "Video"
                 return frames
    _       -> do                             
                 newFrameData <- fmap ImageRGB8 . swap . fromJust <$> getFrame 
                 printStatus ("Frame: " ++ (show $ length frames) ++ " added.") "Video"
                 addNextFrame getFrame (frames ++ [newFrameData]) 


  


  Where I am stuck / The code that should convert images to video :

  


  -- Converts from several images to video
juicyToFFmpeg :: [Image PixelYA8] -> ?
juicyToFFmpeg imgs = undefined


  


• tf.contrib.signal.stft returns an empty matrix

  9 décembre 2017, par matt-pielat

  This is the piece of code I run :

  import tensorflow as tf
  
  
  
  sess = tf.InteractiveSession()
  
  
  
  filename = 'song.mp3' # 30 second mp3 file
  
  SAMPLES_PER_SEC = 44100
  
  
  
  audio_binary = tf.read_file(filename)
  
  
  
  pcm = tf.contrib.ffmpeg.decode_audio(audio_binary, file_format='mp3', samples_per_second=SAMPLES_PER_SEC, channel_count = 1)
  
  stft = tf.contrib.signal.stft(pcm, frame_length=1024, frame_step=512, fft_length=1024)
  
  
  
  sess.close()

  The mp3 file is properly decoded because print(pcm.eval().shape) returns :

  (1323119, 1)

  And there are even some actual non-zero values when I print them with print(pcm.eval()[1000:1010]) :

  [[ 0.18793298]
  
   [ 0.16214484]
  
   [ 0.16022217]
  
   [ 0.15918455]
  
   [ 0.16428113]
  
   [ 0.19858395]
  
   [ 0.22861415]
  
   [ 0.2347789 ]
  
   [ 0.22684409]
  
   [ 0.20728172]]

  But for some reason print(stft.eval().shape) evaluates to :

  (1323119, 0, 513) # why the zero dimension?

  And therefore print(stft.eval()) is :

  []

  According to this the second dimension of the tf.contrib.signal.stft output is equal to the number of frames. Why are there no frames though ?