Recherche avancée

Sur d’autres sites (10988)

• sws/aarch64/yuv2rgb : honor iOS calling convention

  8 avril 2016, par Clément Bœsch

  sws/aarch64/yuv2rgb : honor iOS calling convention
  y_offset and y_coeff being successive 32-bit integers, they are packed
  
  into 8 bytes instead of 2x8 bytes.
  See https://developer.apple.com/library/ios/documentation/Xcode/Conceptual/iPhoneOSABIReference/Articles/ARM64FunctionCallingConventions.html
  > iOS diverges from Procedure Call Standard for the ARM 64-bit
  
  > Architecture in several ways
  
  [...]
  
  > In the generic procedure call standard, all function arguments passed
  
  > on the stack consume slots in multiples of 8 bytes. In iOS, this
  
  > requirement is dropped, and values consume only the space required.
  
  [...]
  
  > Padding is still inserted on the stack to satisfy arguments’ alignment
  
  > requirements.
  

  • [DH] libswscale/aarch64/yuv2rgb_neon.S

• how to convert PCM file to MLP file using FFmpeg in windows ?

  21 octobre 2014, par user2877921

  I have tried converting .mp3 file to .pcm file.now i have added a encoder file which converts it into MLP file.i need the command in command prompt to convert pcm file to mlp file.

• Finding corrupted videos with python

  16 septembre 2022, par tusker

  I have huge video dataset (about 500000 16 frame length clips). Some of the videos produce messages like

  


  [mpeg4 @ 0x561b46a44b80] marker does not match f_code
[mpeg4 @ 0x561b480c0740] Error at MB: 4718
[mpeg4 @ 0x561b4811b640] header damaged


  


  I want to find which videos cause the problem.
  
First I tried with this https://superuser.com/questions/100288/how-can-i-check-the-integrity-of-a-video-file-avi-mpeg-mp4 solution, but I got nothing in the error files.

  


  My second idea was to read the first image of a video and if it throws an error, save the filename to a file. However, it seems this is not an error message : although the messages appeared on console, read returned with success=True.

  


  Then I tried to print the filenames to standard error (where the warning messages also go), saving it to a file and do some postprocessing to extract the videos which has problem.

  


  import os
from tqdm import tqdm
import sys

folder = '/path/to/videos'
for f in tqdm(os.listdir(folder)):
  sys.stderr.write('BEFORE VIDCAP ' + f + '\n')
  vidcap = cv2.VideoCapture(os.path.join(folder,f))
  sys.stderr.write('BETWEEN VIDCAP AND SUCCESS' + f + '\n')

  success, image = vidcap.read()
  sys.stderr.write('AFTER SUCCESS ' + f + '\n')


  


  However, the standard error outputs in the following order :

  


  AFTER SUCCESS
[mpeg ...] 
[mpeg ...]
...
BEFORE VIDCAP


  


  It doesn't make sense how can it happen in this order, some broken videos should be seen somewhere between BEFORE and AFTER. (And the ones following or preceding are not broken neither if I try to open them individually.)

  


  My last idea was to create for every video a file, making it as the stderr and check the non-empty files. (I know it's insanely inefficient, but I didn't have any other idea.)

  


  for f in os.listdir(folder):
  sys.stderr = open(f + '.txt', 'w')
  vidcap = cv2.VideoCapture(os.path.join(folder,f))
  success, image = vidcap.read()


  


  This doesn't work also, all the error files are empty, although the messages still appear on standard error.

  


  ---

  


  So my questions are the following :

  


  

  1. What would be an efficient solution for this problem ?

  


  2. How is this order of outputs (first code example) possible ? There is some parallelism under the hood ?

  


  3. Where are this messages going if not to the standard error (second code example) ? If yes, why are they printed to the console instead of the specified error file ?