Sur d’autres sites (10456)

• avfilter/vf_lagfun : increase filter precision

  11 février 2021, par Paul B Mahol

  avfilter/vf_lagfun : increase filter precision
  

  • [DH] libavfilter/vf_lagfun.c

• GoPro (MP4) video timestamp sync with precision of milliseconds

  3 février 2021, par Raphael Ottoni

  I need your help with a data sync problem... I m currently trying to sync my GoPro video with real world time (a.k.a my notebook). I manage to sync date and time of my notebook and my GoPro 3+ black perfectly. The problem is that when the GoPro save the files in disk it round up the milliseconds on the creation_time (the milliseconds is always 000000) . Thus, turning the perfect sync impossible. In attachment is a picture of the meta information (extracted by ffprobe) of the MP4 video.

  



  My question is : What I have to do, so the GoPro actually save the creation_time with precision of milliseconds ?

  



  Another small question : Looking at the attachment figure, we see the "timecode" which is a time synchronization data in the format of hours:minuts:seconds:frame. I was thinking that I could use the "frame" value to calculate the missing milliseconds value. If we take this attachment, as a example, we can see that the frame value is "36". Meaning that the millisecond that it started to record was the one associated with the 36th frame of the FPS (in this video : 60fps) value : Some thing like 1000/60 * 36 which is 600 milliseconds, thus the actual creation_time of this video would be : 2017-07-19T18:10:34.600

  



  Is this logic right ? it didn't work ! I don't know what else to do.

  



  P.S : I need this kind of time precision because I will sync the video frames with a external sensor data that is recorded at 11hz.

  



  Please Help

  



  

  



  update

  



  I forgot to mention, even if you check the original raw file information, inside the GoPro SSD card, using "stats" to read the creation time (see attachment) it still has the same timestamp without milliseconds.

  



  

  


• lavc/arm : fix lack of precision in ff_ps_stereo_interpolate_neon

  22 juin 2017, par Clément Bœsch

  lavc/arm : fix lack of precision in ff_ps_stereo_interpolate_neon
  The code originally pre-multiply by 2 the steps, causing the running sum
  
  of the h factors to drift away due to the lack of precision. It quickly
  
  causes an inaccuracy > 0.01.
  I tried diverse approaches such as multiply by 2.0 (instead of adding
  
  the value itself) without success.
  I'm unable to bench the impact of this change, feel free to compare.
  This commit fixes the incoming aacpsdsp tests.
  Following is an alternative simplified function (matching the incoming
  
  AArch64 code) that may be used :
  function ff_ps_stereo_interpolate_neon, export=1
  
          vld1.32         q0, [r2]
  
          vld1.32         q1, [r3]
  
          ldr             r12, [sp]
  
          vmov.f32        q8, q0
  
          vmov.f32        q9, q1
  
          vzip.32         q8, q0
  
          vzip.32         q9, q1
  
  1 :
  
          vld1.32         d4, [r0,:64]
  
          vld1.32         d6, [r1,:64]
  
          vadd.f32        q8, q8, q9
  
          vadd.f32        q0, q0, q1
  
          vmov.f32        d5, d4
  
          vmov.f32        d7, d6
  
          vmul.f32        q2, q2, q8
  
          vmla.f32        q2, q3, q0
  
          vst1.32         d4, [r0,:64] !
  
          vst1.32         d5, [r1,:64] !
  
          subs            r12, r12, #1
  
          bgt             1b
  
          bx              lr
  
  endfunc
  

  • [DH] libavcodec/arm/aacpsdsp_neon.S