Sur d’autres sites (4476)

• ffmpeg : concat and encode with one command

  15 mars 2018, par Massimo Vantaggio

  I have a script that automates encode and concat a series of input video files.
  I’m trying to get encode and concat in a single command.

  At the moment it does not work causing immense empty pauses in the final video output.

  I would like to get the result with concat demux instead of concat protocol because it is much harder to automate in a script

  Do you know if it is possible to archive this result with concat demux ?
  Thanks a lot !

  Massimo

  ENCODE / CONCAT 1080p

  for f in ./*.mp4; do echo "file '$f'" >> list.txt; done
  
  
  
  ffmpeg -f concat -safe 0 -y -i list.txt -i ../logo/logo.png -c:a aac -b:a 384k -ar 48000 -ac 2 -af aresample=async=1000 -c:v libx264 -x264opts keyint=50:min-keyint=50:no-scenecut -bf 0 -r 25 -b:v 4800k -maxrate 4800k -bufsize 3000k -profile:v main -crf 22 -filter_complex "[0:v][1:v]overlay=main_w-overlay_w-10:10,scale=1920:1080,setsar=1" ../buffer/1080set.mp4

  It seems to add long pause at low frame-rate :

  [mov,mp4,m4a,3gp,3g2,mj2 @ 0x7fb03f05e600] Auto-inserting h264_mp4toannexb bitstream filter

  frame= 9745 fps= 13 q=27.0 size= 186368kB time=00:18:52.56 bitrate=1348.0kbits

  frame= 9745 fps= 13 q=27.0 size= 186368kB time=00:18:56.57 bitrate=1343.3kbits/

  frame= 9745 fps= 13 q=27.0 size= 186368kB time=00:19:00.65 bitrate=1338.5kbits/

  frame= 9745 fps= 13 q=27.0 size= 186368kB time=00:19:02.29 bitrate=1336.5kbits/

  More than 1000 frames duplicated

  Input #0, mov,mp4,m4a,3gp,3g2,mj2, from '1.mp4':
  
    Metadata:
  
      major_brand     : mp42
  
      minor_version   : 512
  
      compatible_brands: isomiso2avc1mp41
  
      creation_time   : 2016-02-27T22:25:49.000000Z
  
      encoder         : HandBrake 0.10.5 2016021100
  
    Duration: 00:06:29.87, start: 0.000000, bitrate: 3447 kb/s
  
      Stream #0:0(und): Video: h264 (High) (avc1 / 0x31637661), yuv420p(tv, bt709), 768x432, 3282 kb/s, 25 fps, 25 tbr, 90k tbn, 180k tbc (default)
  
      Metadata:
  
        creation_time   : 2016-02-27T22:25:49.000000Z
  
        handler_name    : VideoHandler
  
      Stream #0:1(eng): Audio: aac (LC) (mp4a / 0x6134706D), 48000 Hz, stereo, fltp, 159 kb/s (default)
  
      Metadata:
  
        creation_time   : 2016-02-27T22:25:49.000000Z
  
        handler_name    : Stereo
  
  
  
  Input #1, mov,mp4,m4a,3gp,3g2,mj2, from '2.mp4':
  
    Metadata:
  
      major_brand     : M4V 
  
      minor_version   : 1
  
      compatible_brands: M4V mp42isom
  
      creation_time   : 2016-02-27T18:52:37.000000Z
  
    Duration: 00:07:16.93, start: 0.000000, bitrate: 1184 kb/s
  
      Stream #1:0(eng): Video: h264 (Main) (avc1 / 0x31637661), yuv420p(tv, smpte170m), 768x432, 991 kb/s, 25 fps, 25 tbr, 25k tbn, 50 tbc (default)
  
      Metadata:
  
        creation_time   : 2016-02-27T18:52:37.000000Z
  
        handler_name    : Mainconcept MP4 Video Media Handler
  
        encoder         : AVC Coding
  
      Stream #1:1(eng): Audio: aac (LC) (mp4a / 0x6134706D), 48000 Hz, stereo, fltp, 189 kb/s (default)
  
      Metadata:
  
        creation_time   : 2016-02-27T18:52:37.000000Z
  
        handler_name    : Mainconcept MP4 Sound Media Handler
  
  
  
  Input #2, mov,mp4,m4a,3gp,3g2,mj2, from '3.mp4':
  
    Metadata:
  
      major_brand     : mp42
  
      minor_version   : 0
  
      compatible_brands: mp42mp41
  
      creation_time   : 2017-06-27T17:58:17.000000Z
  
    Duration: 00:05:05.05, start: 0.000000, bitrate: 2622 kb/s
  
      Stream #2:0(eng): Video: h264 (Main) (avc1 / 0x31637661), yuv420p(tv, smpte170m), 768x432, 2301 kb/s, 25 fps, 25 tbr, 25k tbn, 50 tbc (default)
  
      Metadata:
  
        creation_time   : 2017-06-27T17:58:17.000000Z
  
        handler_name    : Alias Data Handler
  
        encoder         : AVC Coding
  
      Stream #2:1(eng): Audio: aac (LC) (mp4a / 0x6134706D), 48000 Hz, stereo, fltp, 317 kb/s (default)
  
      Metadata:
  
        creation_time   : 2017-06-27T17:58:17.000000Z
  
        handler_name    : Alias Data Handler
  
  
  
  Input #3, mov,mp4,m4a,3gp,3g2,mj2, from '4.mp4':
  
    Metadata:
  
      major_brand     : mp42
  
      minor_version   : 0
  
      compatible_brands: mp42mp41
  
      creation_time   : 2017-06-27T17:53:32.000000Z
  
    Duration: 00:06:47.36, start: 0.000000, bitrate: 3144 kb/s
  
      Stream #3:0(eng): Video: h264 (Main) (avc1 / 0x31637661), yuv420p(tv, smpte170m), 768x432, 2819 kb/s, 25 fps, 25 tbr, 25k tbn, 50 tbc (default)
  
      Metadata:
  
        creation_time   : 2017-06-27T17:53:32.000000Z
  
        handler_name    : Alias Data Handler
  
        encoder         : AVC Coding
  
      Stream #3:1(eng): Audio: aac (LC) (mp4a / 0x6134706D), 48000 Hz, stereo, fltp, 317 kb/s (default)
  
      Metadata:
  
        creation_time   : 2017-06-27T17:53:32.000000Z
  
        handler_name    : Alias Data Handler

• Video from images in Python

  26 février 2018, par R. Patterson

  I can draw a series of images using plt.draw() and plt.pause() so it produces something similar to an animation in the python window. I have modified each of the images with various labels, drawings etc.

  import numpy as np
  
  from PIL import Image
  
  import matplotlib.pyplot as plt
  
  import math
  
  
  
  def display(Intensity):
  
      l = plt.Line2D(Intensity[0],Intensity[1],color='yellow') #draw ROI/IAL
  
      ax = plt.gca()
  
      ax.add_line(l)
  
      plt.axis('off')
  
      plt.pause(0.05)
  
      plt.draw()
  
      plt.clf()
  
  
  
  #rotate region of interest
  
  def rotate(origin,Intensity,increment):
  
      ox, oy = origin #coordinates of centre or rotation
  
      x_points=[]
  
      y_points=[]
  
      angle=math.radians(increment)#change in angle between each image
  
      for i in range(0,len(Intensity[0])):
  
          px, py = Intensity[0][i], Intensity[1][i]
  
          qx = ox+math.cos(angle)*(px-ox)-math.sin(angle)*(py-oy)
  
          x_points.append(qx)
  
          qy = oy+math.sin(angle)*(px-ox)+math.cos(angle)*(py-oy)
  
          y_points.append(qy)
  
      rotatecoordinates=[]
  
      rotatecoordinates.append(x_points)
  
      rotatecoordinates.append(y_points)
  
      return rotatecoordinates
  
  
  
  def animation(list, Intensity):
  
      inc=0
  
      for value in list:
  
          item = np.array(value)
  
          rotated=rotate([128,128],Intensity,inc)
  
          im=plt.imshow(item, interpolation='nearest')
  
          display(rotated)
  
          inc+=1
  
  
  
  Image_list=[]
  
  for i in range(0,50):
  
      array=np.linspace(0,1,256*256)
  
      mat=np.reshape(array,(256,256))
  
      img=Image.fromarray(np.uint8(mat*255),'L') #create images
  
      Image_list.append(img)
  
  
  
  myROI=([100,150,150,100,100],[100,100,150,150,100]) #region of interest on image
  
  animation(Image_list,myROI)

  I would like to produce a video file using the images produced. I can’t use the module imageio, imagemagick, opencv, cv2 etc. I think ffmpeg would work, I have the following code.

  def save():
  
      os.system("ffmpeg -r 1 -i img%01d.png -vcodec mpeg4 -y movie.mp4") 

  I don’t understand how to use it in relation to the code I already have. It doesn’t take any arguments, how would I relate it to the images I have ? I know how to use imagej/fiji to produce videos from images but I would like to do this in python and also it runs out of memory (I have a lot of images, over 2000). Any help would be appreciated, thank you.

• Convert mp4 video to avi video using imageio package

  3 février 2018, par Li.Chenyang

  I use the python imageio package to convert a .mp4 video to a .avi video, keeping the fps and size same. The following is my code :

  import imageio
  
  
  
  src_dir = "my/source/video.mp4"
  
  dst_dir = "my/dst/video.avi"
  
  
  
  reader = imageio.get_reader(src_dir)
  
  fps = reader.get_meta_data()['fps']
  
  writer = imageio.get_writer(dst_dir, fps=fps)
  
  
  
  for im in reader:
  
      writer.append_data(im[:, :, :])
  
  writer.close()

  I successfully make it.
  However, I find that the video.mp4 is 27.1 MB, while the video.avi is only 3.70 MB.
  Then I use cv2 to do the same thing :

  import cv2
  
  
  
  src_dir = "my/source/video.mp4"
  
  dst_dir = "my/dst/video_1.avi"
  
  
  
  video_cap = cv2.VideoCapture(src_dir)
  
  fps = video_cap.get(cv2.cv.CV_CAP_PROP_FPS)
  
  size = (int(video_cap.get(cv2.cv.CV_CAP_PROP_FRAME_WIDTH)),   
  
          int(video_cap.get(cv2.cv.CV_CAP_PROP_FRAME_HEIGHT)))  
  
  video_writer = cv2.VideoWriter(dst_dir, cv2.cv.CV_FOURCC('M', 'J', 'P', 'G'), fps, size) 
  
  
  
  success, frame = video_cap.read()
  
  while success:
  
      video_writer.write(frame)
  
      success, frame = video_cap.read()

  This time I get a video_1.avi, which is 65.6 MB.
  My questions :

  1. What is the different between these two method, why the video.avi (using imageio method) is so small ;
  2. Is there any problem to use the video.avi (using imageio method) to train a 3D-CNN instead of using video_1.avi (using cv2 method) ?

  ---

  supplement
  Here is the information of my video file :

  lichenyang@lichenyang-All-Series:~/chalearn2014/script$ file video.mp4
  
  video.mp4: ISO Media, MPEG v4 system, version 2
  
  lichenyang@lichenyang-All-Series:~/chalearn2014/script$ file video.avi
  
  video.avi: RIFF (little-endian) data, AVI, 640 x 480, 20.00 fps, video: H.264 X.264 or H.264
  
  lichenyang@lichenyang-All-Series:~/chalearn2014/script$ file video_1.avi
  
  video_1.avi: RIFF (little-endian) data, AVI, 640 x 480, 20.00 fps, video: Motion JPEG