Recherche avancée

Sur d’autres sites (16840)

• x86 : simple_idct10_template : fix overflow in pass

  12 octobre 2015, par Christophe Gisquet

  x86 : simple_idct10_template : fix overflow in pass
  When the input of a pass has 15 or 16 bits of precision (in particular
  
  the column pass), the addition of a bias to W4 may lead to overflows
  
  in the input to pmaddwd.
  This requires postponing the adding of the bias to after the first
  
  butterfly. To do so, the fact that m15, unused although zeroed, is
  
  exploited. In case the pass is safe, an address can be directly used,
  
  and the number of xmm regs can be decreased. Otherwise, the 32bits bias
  
  is loaded into it.
  Signed-off-by : Michael Niedermayer <michael@niedermayer.cc>
  

  • [DH] libavcodec/x86/proresdsp.asm
  • [DH] libavcodec/x86/simple_idct10_template.asm

• vulkan_decode : don't call get_proc_addr on every frame's destruction

  15 septembre 2023, par Lynne

  vulkan_decode : don't call get_proc_addr on every frame's destruction
  The issue is that we cannot rely on any context existing when we free
  
  frames. The Vulkan functions are loaded in each context separately,
  
  so until now, we've just been loading them on every frame's destruction.
  Rather than do this, just save the function pointers we need in each
  
  frame. The function pointers are guaranteed to not change and exist.
  

  • [DH] libavcodec/vulkan_decode.c
  • [DH] libavcodec/vulkan_decode.h

• ImageJ / Fiji shows wrong number of frames in video (FFMPEG import)

  28 avril 2023, par locoric_polska

  I am counting the number of animals in a an area using Fiji. I import the video through the FFMPEG plug-in (videos are mp4 with mpeg-4 codec). However, I noticed that when I import the videos Fiji uploads the wrong number of frames, and I cannot understand why and how.

  


  An example. I have a video shot at 25fps which is 1582s long. If I do the calculations the video should have 39550 frames in total (1582*25). When I open it through a Computer vision package in R, I see that the video correctly contains 39550 frames. However, when loaded in Fiji, the shown number of frames is 49511. So Fiji is adding 9961 frames to the video. This is consistent across all videos that are recorded in 25fps, while it does not appear in videos shot at 24fps.

  


  Curiously, I found that the ratio between the number of frames read by Fiji and the 'real' number of frames is consistent between 0.79 and 0.80. This makes me think that Fiji is expecting the video to be 30fps and (possibly) duplicating frames to adjust the video to this assumption.

  


  Unfortunately, I discovered all this after finishing my analysis and while trying to merge this dataset with another obtained through CV. The number of frame does not match between datasets and I am not sure how to solve this.

  


  Any help would be greatly appreciated !!

  


  An idea is to multiply all the frame numbers by 0.8 to adjust them to the old assumption. This solution assumes that Fiji is duplicating frames throughout the video in a consistent way