Recherche avancée

Sur d’autres sites (4003)

• Trouble with frame accuracy applying subcaps using .ass files to 23.976fps video [closed]

  25 août 2023, par WhatsYourFunction

  Currently testing .ass subcaps in a VFX workflow.
The goal is to drop specfic text over specific shots and the in/out points have to be frame accurate
We're working in a 23.976 project.

  


  Currently having no trouble using FFmpeg to generate frame-accurate subclips of individual shots from a full-show export by converting hh:mm:ss:ff to seconds and then handling the 24 to 23.976 offset, using the following alorithm :

  


  InPoint_Seconds = ConvertToSeconds(InPoint_Hmsf_FullShow) - ConvertToSeconds(Start_Hmsf_FullShow) // Convert from SMTPE Time Code to seconds.
InPoint_Seconds = InPoint_Seconds * (1001 / 1000) //Handle 24 to 23.976 offset
OutPoint_Seconds = [Same idea as above]
Duration_Seconds = Output_Seconds - InPoint_Seconds

> ffmpeg -ss InPoint_Seconds -t Duration_Seconds -i SourcePath -c copy DestPath


  


  So generating frame-accurate copies of portions of a larger file works with perfect accuracy

  


  BUT when applying the same logic to subcaps using .ass files, sometimes they land with frame accuracy, and sometimes they don't (They'll be 1 frame late at most, and it does not increase over the span of the source clip).

  


  Curious if anyone has any ideas.

  


• Revision af660715c0 : Make coefficient skip condition an explicit RD choice. This commit replaces zru

  28 juin 2013, par Ronald S. Bultje

  Changed Paths :
   Modify /vp9/common/vp9_rtcd_defs.sh

  
   Modify /vp9/encoder/vp9_block.h

  
   Modify /vp9/encoder/vp9_onyx_int.h

  
   Modify /vp9/encoder/vp9_quantize.c

  
   Modify /vp9/encoder/vp9_rdopt.c

  
   Modify /vp9/encoder/x86/vp9_error_sse2.asm

  
  
  Make coefficient skip condition an explicit RD choice.

  This commit replaces zrun_zbin_boost, a method of biasing non-zero
  coefficients following runs of zero-coefficients to be rounded towards
  zero, with an explicit skip-block choice in the RD loop.

  The logic is basically that if individual coefficients should be rounded
  towards zero (from a RD point of view), the trellis/optimize loop should
  take care of it. If whole blocks should be zero (from a RD point of
  view), a single RD check is much more efficient than a complete
  serialization of the quantization loop.

  Quality change : derf +0.5% psnr, +1.6% ssim ; yt +0.6% psnr, +1.1% ssim.
  SIMD for quantize will follow in a separate patch. Results for other
  test sets pending.

  Change-Id : Ife5fa641163ac5150ac428011e87188f1937c1f4

• Efficiently write a movie directly from np.array using pipes

  16 juin 2017, par Matt Billman

  I have a 4D numpy array of movie frames. I’m looking for a function to write them to a movie, at a given framerate. I have FFMPEG installed on my OS, and as I can see from these answers, the most efficient way to do so is via pipes.

  However, I have very little experience using pipes, and the explanations in the link above make little sense to me. Furthermore, very few of the answers seem actually implement pipes, and the one that does uses mencoder, not FFMPEG. I am relatively inexperienced with FFMPEG, so am not sure how to modify the command string from the mencoder answer to make it work in FFMPEG.

  WHAT I WOULD LIKE :

  A function of the following form :

  animate_np_array(4d_array, framerate) -> output.mp4 (or other video codec)

  Which implements pipes to send frames one after the other to FFMPEG, and which I can copy-paste into my existing code.

  Furthermore, it is absolutely necessary that this function never actually plots any of the frames, as calls to the matplotlib.imshow() function (as I have most typically seen used) slow things down considerably.