Recherche avancée

Sur d’autres sites (4935)

• OpenCV : issues with playing videos created using ffmpeg and concat

  25 mars 2022, par tobiasBora

  I concatenated a few videos using :

  


  $ ffmpeg -f concat -safe 0 -i list_files.txt -c copy total.mp4


  


  where list_files.txt is like :

  


  file 'file1.mp4'
file 'file2.mp4'


  


  Unfortunately, if I play it with opencv :

  


  import cv2
import numpy as np

# Create a VideoCapture object and read from input file
# If the input is the camera, pass 0 instead of the video file name
cap = cv2.VideoCapture('../Video_outputs/010_introduction.mp4')

# Check if camera opened successfully
if (cap.isOpened()== False): 
  print("Error opening video stream or file")

# Read until video is completed
while(cap.isOpened()):
  # Capture frame-by-frame
  ret, frame = cap.read()
  if ret == True:

    # Display the resulting frame
    cv2.imshow('Frame',frame)

    # Press Q on keyboard to  exit
    if cv2.waitKey(24) & 0xFF == ord('q'):
      break

  # Break the loop
  else: 
    break

# When everything done, release the video capture object
cap.release()

# Closes all the frames
cv2.destroyAllWindows()


  


  the first video plays fine, but the next concatenated videos are ugly looking :

  


  

  


  Am I missing something, or is it a bug to report upstream ?

  


  PS : I'm using this code :

  


  import cv2
import numpy as np

# Create a VideoCapture object and read from input file
# If the input is the camera, pass 0 instead of the video file name
cap = cv2.VideoCapture('../Video_outputs/010_introduction.mp4')

# Check if camera opened successfully
if (cap.isOpened()== False): 
  print("Error opening video stream or file")

# Read until video is completed
while(cap.isOpened()):
  # Capture frame-by-frame
  ret, frame = cap.read()
  if ret == True:

    # Display the resulting frame
    cv2.imshow('Frame',frame)

    # Press Q on keyboard to  exit
    if cv2.waitKey(24) & 0xFF == ord('q'):
      break

  # Break the loop
  else: 
    break

# When everything done, release the video capture object
cap.release()

# Closes all the frames
cv2.destroyAllWindows()


  


• Does AVFrame store AV_PIX_FMT_YUV420P data as YVU ?

  28 juin 2017, par Peter

  I am decoding raw H.265 data using avcodec_decode_video2 api. When I examine the resulting instance pictYUV of type AVFrame, I see that pictYUV->format is AV_PIX_FMT_YUV420P and pictYUV->data[0] points to Y-plane. Both of these are expected. However, it appears pictYUV->data[1] seem to contain V-plane data and pictYUV->data[2]seem to contain U-plane data. My intuition was that pictYUV->data would store YUV planes in that order and not YVU planes. Wondering if the data is always ordered as YVU or is there some flag I failed to look at. Regards.

• Android and JNI, pipe data to FFmpeg

  2 décembre 2014, par William Seemann

  I’m trying to create a metadata retriever based on FFmpeg. Since raw Android application resources are often only accessible using a file descriptor I need a way to pipe this data to FFmpeg via JNI. I know FFmpeg supports a "pipe" protocol :

  UNIX pipe access protocol.
  
  
  
  Allow to read and write from UNIX pipes.
  
  
  
  The accepted syntax is: 
  
  
  
      pipe:[number]
  
  
  
  number is the number corresponding to the file descriptor of the pipe (e.g. 0 for stdin, 1 for stdout, 2 for stderr). If number is not specified, by default the stdout file descriptor will be used for writing, stdin for reading.
  
  
  
  For example: cat test.wav | ffmpeg -i pipe:0

  My question is, how do I programmatically emulate cat test.wav | ffmpeg -i pipe:0 using JNI and avformat_open_input using a FileDescriptor ? Is this possible ?