Recherche avancée

Sur d’autres sites (6230)

• Pipe raw OpenCV images to FFmpeg

  15 février 2017, par BrianTheLion

  Here’s a fairly straightforward example of reading off a web cam using OpenCV’s python bindings :

  '''capture.py'''
  
  import cv, sys
  
  cap = cv.CaptureFromCAM(0)                    # 0 is for /dev/video0
  
  while True :
  
      if not cv.GrabFrame(cap) : break
  
      frame = cv.RetrieveFrame(cap)
  
      sys.stdout.write( frame.tostring() )

  Now I want to pipe the output to ffmpeg as in :

  $ python capture.py | ffmpeg -f image2pipe -pix_fmt bgr8 -i - -s 640x480 foo.avi

  Sadly, I can’t get the ffmpeg magic incantation quite right and it fails with

    libavutil     50.15. 1 / 50.15. 1
    libavcodec    52.72. 2 / 52.72. 2
    libavformat   52.64. 2 / 52.64. 2
    libavdevice   52. 2. 0 / 52. 2. 0
    libavfilter    1.19. 0 /  1.19. 0
    libswscale     0.11. 0 /  0.11. 0
    libpostproc   51. 2. 0 / 51. 2. 0
  Output #0, avi, to ’out.avi’ :
      Stream #0.0 : Video : flv, yuv420p, 640x480, q=2-31, 19660 kb/s, 90k tbn, 30 tbc
  [image2pipe @ 0x1508640]max_analyze_duration reached
  [image2pipe @ 0x1508640]Estimating duration from bitrate, this may be inaccurate
  Input #0, image2pipe, from ’pipe :’ :
    Duration : N/A, bitrate : N/A
      Stream #0.0 : Video : 0x0000, bgr8, 25 fps, 25 tbr, 25 tbn, 25 tbc
  swScaler : 0x0 -> 640x480 is invalid scaling dimension
  

  • The captured frames are definitely 640x480.
  • I’m pretty sure the pixel order for the OpenCV image type (IplImage) is GBR, one byte per channel. At least, that’s what seems to be coming off the camera.

  I’m no ffmpeg guru. Has anyone done this successfully ?

• Pipe raw OpenCV images to FFmpeg

  6 octobre 2022, par BrianTheLion

  Here's a fairly straightforward example of reading off a web cam using OpenCV's python bindings :

  



  '''capture.py'''
import cv, sys
cap = cv.CaptureFromCAM(0)                    # 0 is for /dev/video0
while True :
    if not cv.GrabFrame(cap) : break
    frame = cv.RetrieveFrame(cap)
    sys.stdout.write( frame.tostring() )


  



  Now I want to pipe the output to ffmpeg as in :

  



  $ python capture.py | ffmpeg -f image2pipe -pix_fmt bgr8 -i - -s 640x480 foo.avi

  



  Sadly, I can't get the ffmpeg magic incantation quite right and it fails with

  



  
  libavutil     50.15. 1 / 50.15. 1
  libavcodec    52.72. 2 / 52.72. 2
  libavformat   52.64. 2 / 52.64. 2
  libavdevice   52. 2. 0 / 52. 2. 0
  libavfilter    1.19. 0 /  1.19. 0
  libswscale     0.11. 0 /  0.11. 0
  libpostproc   51. 2. 0 / 51. 2. 0
Output #0, avi, to 'out.avi' :
    Stream #0.0 : Video : flv, yuv420p, 640x480, q=2-31, 19660 kb/s, 90k tbn, 30 tbc
[image2pipe @ 0x1508640]max_analyze_duration reached
[image2pipe @ 0x1508640]Estimating duration from bitrate, this may be inaccurate
Input #0, image2pipe, from 'pipe :' :
  Duration : N/A, bitrate : N/A
    Stream #0.0 : Video : 0x0000, bgr8, 25 fps, 25 tbr, 25 tbn, 25 tbc
swScaler : 0x0 -> 640x480 is invalid scaling dimension


  



  

  • The captured frames are definitely 640x480.

  


  • I'm pretty sure the pixel order for the OpenCV image type (IplImage) is GBR, one byte per channel. At least, that's what seems to be coming off the camera.

  


  



  I'm no ffmpeg guru. Has anyone done this successfully ?

  


• Store image specific metadata in video

  5 mai 2021, par VocoJax

  I'd like to store XMP data in my PNG images and then compress my images using a lossless compression codec in a video for long-term storage.

  


  Is this possible ? Is per-image metadata discarded when compressing into a video ?

  


  For context, I need a pythonic solution utilizing FFMPEG to do the video compression. I'm aware that OPENCV discards VIDEO metadata. Further, I care not for VIDEO metadata, only per image metadata. Additionally, the type of metadata I'm storing is in reference to the timestamp of the frame, the camera it was taken on, and maybe even some information regarding on-prem AI detecting like bounding box of an object in frame or something like that.

  


  My current approach is just to use OPENCV to get the frames and then record the time into a fat array. Once my video file gets to a specific size or frame count, I'll just prepend the json array of timestamps to the video file along with a 32-bit integer regarding how long the array is. This solution will definitely work, but, it's a hassle to record the timestamps in memory, plus write custom scripts to load my video again (parsing the timestamps off the front of it beforehand).

  


  Who's got the magic solution ? Could I use the approach I'm currently using except attach the frame timestamps to the VIDEO metadata instead ?