Recherche avancée

Sur d’autres sites (8600)

• FFMpeg Command work in command line, but in python script the image is blueish ? (Semi Solved)

  17 juin 2015, par Fooldj

  Okay, kind of a weird problem. But I’m not sure whether it’s python, ffmpeg, or some stupid thing I’m doing wrong.

  I’m trying to take a video, and take 1 frame a second, and output that frame to an image. Right now, if i use the command line with ffmpeg :

  ffmpeg -i test.avi -r 1 -f image2 image-%3d.jpeg -pix_fmt rgb24 -vcodec rawrvideo

  It outputs about 10 images, the images look fine, awesome. Now I have this code (right now some code from some github, as I wanted stuff that i was relatively sure would work, and mine is allll convoluted)

  import subprocess as sp
  
  import numpy as np
  
  import re
  
  import cv2
  
  import time
  
  
  
  FFMPEG_BIN = r'ffmpeg.exe'
  
  INPUT_VID = 'test.avi'
  
  
  
  def getInfo():
  
      command = [FFMPEG_BIN,'-i', INPUT_VID, '-']
  
      pipe = sp.Popen(command, stdout=sp.PIPE, stderr=sp.PIPE)
  
      pipe.stdout.readline()
  
      pipe.terminate()
  
      infos = pipe.stderr.read()
  
      infos_list = infos.split('\r\n')
  
      res = re.search(' \d+x\d+ ',infos)
  
      res = [int(x) for x in res.group(0).split('x')]
  
      return res
  
  res = getInfo()
  
  command = [ FFMPEG_BIN,
  
          '-i', INPUT_VID,
  
          '-f', 'image2pipe',
  
          '-pix_fmt', 'rgb24',
  
          '-vcodec', 'rawvideo', '-']
  
  pipe = sp.Popen(command, stdout = sp.PIPE, bufsize=10**8)
  
  n = 0
  
  im2 = []
  
  try:
  
      mog = cv2.BackgroundSubtractorMOG2(120,2,True)
  
      while True:
  
          raw_image = pipe.stdout.read(res[0]*res[1]*3)
  
          # transform the byte read into a numpy array
  
          image =  np.fromstring(raw_image, dtype='uint8')
  
          image = image.reshape((res[1],res[0],3))
  
          rgbImg = image.copy()
  
  
  
          fname = ('_tmp%03d.png'%time.time())
  
          cv2.imwrite(fname, rgbImg)
  
          # throw away the data in the pipe's buffer.
  
          #pipe.stdout.flush()
  
          n += 1
  
          print n
  
  except:
  
      print 'done',n
  
      pipe.kill()
  
      cv2.destroyAllWindows()

  When I run this, I get 10 images, but they all have a Blue Tint ! I cannot for the life of me figure out why. I’ve done tons of searches, I’ve tried quite a few different codecs (usually just messes things up worse). The media info for the video file is here :

   General
  
   Complete name                            : test.avi
  
  Format                                   : AVI
  
  Format/Info                              : Audio Video Interleave
  
  File size                                : 85.0 KiB
  
  Duration                                 : 133ms
  
  Overall bit rate                         : 5 235 Kbps
  
  
  
  Video
  
  ID                                       : 0
  
  Format                                   : JPEG
  
  Codec ID                                 : MJPG
  
  Duration                                 : 133ms
  
  Bit rate                                 : 1 240 Kbps
  
  Width                                    : 640 pixels
  
  Height                                   : 480 pixels
  
  Display aspect ratio                     : 4:3
  
  Frame rate                               : 30.000 fps
  
  Color space                              : YUV
  
  Chroma subsampling                       : 4:2:2
  
  Bit depth                                : 8 bits
  
  Compression mode                         : Lossy
  
  Bits/(Pixel*Frame)                       : 0.135
  
  Stream size                              : 20.1 KiB (24%)

  Any suggestions ? It seems like it should be an RGB mixup...just not sure where at...

  EDIT : So I fixed the problem by switching the blue and red channels with this code :
  bChannel = rgbImg[ :, :,0]
  rChannel = rgbImg[ :, :,2]
  gChannel = rgbImg[ :, :,1]

              rgbArray = np.zeros((res[1],res[0],3), 'uint8')
  
              rgbArray[...,0] = rChannel
  
              rgbArray[...,1] = gChannel
  
              rgbArray[...,2] = bChannel

  So I guess this is now a question of, why is python mixing up these channels ? Is it a problem with python, or ffmpeg, the codec ?

  Thanks !

• FFMpeg Command work in command line, but not in python script. (Semi Solved)

  20 février 2015, par Fooldj

  Okay, kind of a weird problem. But I’m not sure whether it’s python, ffmpeg, or some stupid thing I’m doing wrong.

  I’m trying to take a video, and take 1 frame a second, and output that frame to an image. Right now, if i use the command line with ffmpeg :

  ffmpeg -i test.avi -r 1 -f image2 image-%3d.jpeg -pix_fmt rgb24 -vcodec rawrvideo

  It outputs about 10 images, the images look fine, awesome. Now I have this code (right now some code from some github, as I wanted stuff that i was relatively sure would work, and mine is allll convoluted)

  import subprocess as sp
  
  import numpy as np
  
  import re
  
  import cv2
  
  import time
  
  
  
  FFMPEG_BIN = r'ffmpeg.exe'
  
  INPUT_VID = 'test.avi'
  
  
  
  def getInfo():
  
      command = [FFMPEG_BIN,'-i', INPUT_VID, '-']
  
      pipe = sp.Popen(command, stdout=sp.PIPE, stderr=sp.PIPE)
  
      pipe.stdout.readline()
  
      pipe.terminate()
  
      infos = pipe.stderr.read()
  
      infos_list = infos.split('\r\n')
  
      res = re.search(' \d+x\d+ ',infos)
  
      res = [int(x) for x in res.group(0).split('x')]
  
      return res
  
  res = getInfo()
  
  command = [ FFMPEG_BIN,
  
          '-i', INPUT_VID,
  
          '-f', 'image2pipe',
  
          '-pix_fmt', 'rgb24',
  
          '-vcodec', 'rawvideo', '-']
  
  pipe = sp.Popen(command, stdout = sp.PIPE, bufsize=10**8)
  
  n = 0
  
  im2 = []
  
  try:
  
      mog = cv2.BackgroundSubtractorMOG2(120,2,True)
  
      while True:
  
          raw_image = pipe.stdout.read(res[0]*res[1]*3)
  
          # transform the byte read into a numpy array
  
          image =  np.fromstring(raw_image, dtype='uint8')
  
          image = image.reshape((res[1],res[0],3))
  
          rgbImg = image.copy()
  
  
  
          fname = ('_tmp%03d.png'%time.time())
  
          cv2.imwrite(fname, rgbImg)
  
          # throw away the data in the pipe's buffer.
  
          #pipe.stdout.flush()
  
          n += 1
  
          print n
  
  except:
  
      print 'done',n
  
      pipe.kill()
  
      cv2.destroyAllWindows()

  When I run this, I get 10 images, but they all have a Blue Tint ! I cannot for the life of me figure out why. I’ve done tons of searches, I’ve tried quite a few different codecs (usually just messes things up worse). The media info for the video file is here :

   General
  
   Complete name                            : test.avi
  
  Format                                   : AVI
  
  Format/Info                              : Audio Video Interleave
  
  File size                                : 85.0 KiB
  
  Duration                                 : 133ms
  
  Overall bit rate                         : 5 235 Kbps
  
  
  
  Video
  
  ID                                       : 0
  
  Format                                   : JPEG
  
  Codec ID                                 : MJPG
  
  Duration                                 : 133ms
  
  Bit rate                                 : 1 240 Kbps
  
  Width                                    : 640 pixels
  
  Height                                   : 480 pixels
  
  Display aspect ratio                     : 4:3
  
  Frame rate                               : 30.000 fps
  
  Color space                              : YUV
  
  Chroma subsampling                       : 4:2:2
  
  Bit depth                                : 8 bits
  
  Compression mode                         : Lossy
  
  Bits/(Pixel*Frame)                       : 0.135
  
  Stream size                              : 20.1 KiB (24%)

  Any suggestions ? It seems like it should be an RGB mixup...just not sure where at...

  EDIT : So I fixed the problem by switching the blue and red channels with this code :
  bChannel = rgbImg[ :, :,0]
  rChannel = rgbImg[ :, :,2]
  gChannel = rgbImg[ :, :,1]

              rgbArray = np.zeros((res[1],res[0],3), 'uint8')
  
              rgbArray[...,0] = rChannel
  
              rgbArray[...,1] = gChannel
  
              rgbArray[...,2] = bChannel

  So I guess this is now a question of, why is python mixing up these channels ? Is it a problem with python, or ffmpeg, the codec ?

  Thanks !

• Compile FFMPEG for command line usage

  3 septembre 2014, par Lior Iluz

  I’ve been trying to compile FFMPEG so I can use it with my Android application with commands. The result should be 1 static file, "ffmpeg", that is not package dependent. No .so files.

  I managed to compile it with guardianProject and everything is working but the source was too old and lacks options I need.
  I’m using the latest Ubuntu on VirtualBox, all essentials are installed and updated (gawk, yasm, aptitude, etc...).

  There are multiple examples around the web. Here are the issues I’ve experienced with each of the options I tried. I’d appreciate help with either one of the following errors :

  1. Guardian project - I managed to compile it and get the ffmpeg file but it uses an old version of ffmpeg that doesn’t include the "-movFlags faststart" option. I tried throwing the new ffmpeg (2.3.3) in there but it just throws error. (I ran git submodule init and update)

  2. JayH5 - A more generic build file that basically should work with every ffmpeg. I updated the build script to work with the latest NDK and it seems to be working but than it fails on /home/dor/Desktop/ndk/android-ndk-r10/toolchains/arm-linux-androideabi-4.8/prebuilt/linux-x86/bin/arm-linux-androideabi-pkg-config not found, library detection may fail.
     I read about this and it seems to require a hack, which I don’t understand how to implement.

  3. Trovao - this is a very nice project and the build script even downloads the latest ffmpeg and x264 sources, and I even succeeded in compiling it to FFMPEG and X264 files but when I use it, I get this error : could not load library "libx264.so.142 which is weird, as x264 is there and the whole idea of this project is to cancel the need of .so files.

  I’d appreciate help with figuring out how to solve any of the issues. I prefer using the JayH5 build script as it seems the most straight forward out there...

  JayH5 build file :

  #!/bin/bash
  
  
  
  # set the base path to your Android NDK (or export NDK to environment)
  
  
  
  if [[ "x$NDK_BASE" == "x" ]]; then
  
      NDK_BASE=/opt/android-ndk
  
      echo "No NDK_BASE set, using $NDK_BASE"
  
  fi
  
  
  
  NDK_PLATFORM_VERSION=14
  
  NDK_ABI=arm
  
  NDK_COMPILER_VERSION=4.8
  
  NDK_SYSROOT=$NDK_BASE/platforms/android-$NDK_PLATFORM_VERSION/arch-$NDK_ABI
  
  NDK_UNAME=`uname -s | tr '[A-Z]' '[a-z]'` # Convert Linux -> linux
  
  HOST=$NDK_ABI-linux-androideabi
  
  NDK_TOOLCHAIN_BASE=$NDK_BASE/toolchains/$HOST-$NDK_COMPILER_VERSION/prebuilt/$NDK_UNAME-x86
  
  CC="$NDK_TOOLCHAIN_BASE/bin/$HOST-gcc --sysroot=$NDK_SYSROOT"
  
  LD=$NDK_TOOLCHAIN_BASE/bin/$HOST-ld
  
  
  
  BUILD_PATH=build/ffmpeg
  
  
  
  ./configure \
  
  $DEBUG_FLAG \
  
  --arch=arm \
  
  --target-os=linux \
  
  --enable-runtime-cpudetect \
  
  --enable-pic \
  
  --disable-shared \
  
  --enable-static \
  
  --cross-prefix=$NDK_TOOLCHAIN_BASE/bin/$NDK_ABI-linux-androideabi- \
  
  --sysroot="$NDK_SYSROOT" \
  
  --extra-cflags="-march=armv7-a -mfloat-abi=softfp -fPIC -DANDROID" \
  
  --extra-ldflags="" \
  
  --enable-ffmpeg \
  
  --disable-ffplay \
  
  --disable-ffprobe \
  
  --disable-ffserver \
  
  --disable-network \