Recherche avancée

Sur d’autres sites (5405)

• avformat/asfdec_o : Check size vs. offset in detect_unknown_subobject()

  12 janvier 2021, par Michael Niedermayer

  avformat/asfdec_o : Check size vs. offset in detect_unknown_subobject()
  Fixes : signed integer overflow : 2314885530818453566 + 7503032301549264928 cannot be represented in type 'long'
  
  Fixes : 26639/clusterfuzz-testcase-minimized-ffmpeg_dem_ASF_O_fuzzer-6024222100684800
  Alternatively this could be ignored but then the end condition of the loop
  
  would be hard to reach as avio_tell() is int64_t
  Found-by : continuous fuzzing process https://github.com/google/oss-fuzz/tree/master/projects/ffmpeg
  
  Signed-off-by : Michael Niedermayer <michael@niedermayer.cc>
  

  • [DH] libavformat/asfdec_o.c

• Reading h264 stream in Python

  24 mars 2017, par verified.human

  I am trying to read a video stream from a Parrot Bebop 2 drone.
  The video stream is written as "socket" to a file as H264 stream.

  $ ffmpeg -i [STREAM]
  
  Input #0, h264, from 'stream_h264':
  
    Duration: N/A, bitrate: N/A
  
    Stream #0:0: Video: h264 (Constrained Baseline), 1 reference frame, yuv420p(progressive, left), 1280x720, 23.98 fps, 23.98 tbr, 1200k tbn, 47.95 tbc

  Reading the video stream in MPlayer is not a problem using the parameters below. Playing it using VLC or ffmpeg should also not be too hard. For MPlayer the following works :

  mplayer -fs -demuxer h264es -benchmark stream_h264

  This plays the stream in high-res. However my goal is to perform image processing on the frames using Python (mostly OpenCV). Therefore, I would like to read the frames into NumPy arrays. I have already considered using cv2.VideoCapture but this does not seem to work for my stream. Other (somewhat easy) to use options I am not aware of, therefore my question is whether someone recommend me how to read the video frames in Python ?

  All recommendations are more than welcome !

• swscale/aarch64 : add hscale specializations

  26 mai 2022, par Swinney, Jonathan

  swscale/aarch64 : add hscale specializations
  This patch adds code to support specializations of the hscale function
  
  and adds a specialization for filterSize == 4.
  ff_hscale8to15_4_neon is a complete rewrite. Since the main bottleneck
  
  here is loading the data from src, this data is loaded a whole block
  
  ahead and stored back to the stack to be loaded again with ld4. This
  
  arranges the data for most efficient use of the vector instructions and
  
  removes the need for completion adds at the end. The number of
  
  iterations of the C per iteration of the assembly is increased from 4 to
  
  8, but because of the prefetching, there must be a special section
  
  without prefetching when dstW < 16.
  This improves speed on Graviton 2 (Neoverse N1) dramatically in the case
  
  where previously fs=8 would have been required.
  before : hscale_8_to_15__fs_8_dstW_512_neon : 1962.8
  
  after : hscale_8_to_15__fs_4_dstW_512_neon : 1220.9
  Signed-off-by : Jonathan Swinney <jswinney@amazon.com>
  
  Signed-off-by : Martin Storsjö <martin@martin.st>
  

  • [DH] libswscale/aarch64/hscale.S
  • [DH] libswscale/aarch64/swscale.c
  • [DH] libswscale/utils.c