Recherche avancée

Sur d’autres sites (13052)

• Recapping WebM’s First Week

  25 mai 2010, par noreply@blogger.com (John Luther) — webm, vp8, vorbis

  The WebM project launched last Wednesday with broad industry backing (watch video of the announcement). The list of supporters keeps growing with new additions such as the popular VLC media player, Miro Video Converter, HeyWatch cloud encoding platform, and videantis programmable processor platform. We’re also happy to see that future versions of IE will support playback of VP8 when the user has installed the codec.

  Our announcement sparked discussions in the community around the design and quality of our developer release. We’ve done extensive testing of VP8 and know that the codec can match or exceed the quality of other leading codecs. Starting this week, the engineers behind WebM will post frequently to this blog with details on how to make optimal use of its VP8 video codec and Vorbis audio codec. We are confident that the open development model will bring additional improvements that will further optimize WebM. In fact, the power of open development is already visible, with developers submitting patches and the folks at Flumotion enabling live streaming support in their product just three days after the project was launched.

  Keep an eye on this blog for regular updates on the adoption and development of WebM. To participate in the conversation or to ask questions of the WebM team, please join our discussion group.

  John Luther
  Product Manager, Google

  

• FFMPEG Stream via UDpxy and HDMI Grabber

  4 mars 2020, par Fabian Schäfer

  We use a HDMI Grabber and the manual from here : https://blog.danman.eu/reverse-engineering-lenkeng-hdmi-over-ip-extender/
  You can find the Script here : https://gist.github.com/danielkucera/0a2f36bc53959e4879cb567149aafb78

  Now we want to grab the Stream via FFMPEG and UDPxy to TVHeadEnd. But FFMPEG produces some Error as you can see below.

  FFMPEG Configuration :

  sudo ./recvlkv373.py 226.2.2.2 2068 /dev/stdout | ffmpeg -re -i pipe: -c:v h264 -c:a copy -f mpegts udp://239.0.0.1:1234

  [mjpeg @ 0x5578d5a232e0] error count: 73e1c90ea5aa3699
  
  [mjpeg @ 0x5578d5a232e0] error y=80 x=15
  
  frame=  746 fps= 25 q=28.0 size=     229kB time=00:00:27.48 bitrate=  68.2kbits/s speed=0.922frame=  758 fps= 25 q=28.0 size=     231kB time=00:00:27.96 bitrate=  67.7kbits/s speed=0.923frame=  771 fps= 25 q=28.0 size=     234kB time=00:00:28.48 bitrate=  67.2kbits/s speed=0.924frame=  783 fps= 25 q=28.0 size=     236kB time=00:00:28.96 bitrate=  66.7kbits/s speed=0.925frame=  796 fps= 25 q=25.0 size=     242kB time=00:00:29.48 bitrate=  67.2kbits/s speed=0.926frame=  809 fps= 25 q=28.0 size=     291kB time=00:00:30.00 bitrate=  79.5kbits/s speed=0.928frame=  821 fps= 25 q=28.0 size=     294kB time=00:00:30.48 bitrate=  79.1kbits/s speed=0.928frame=  834 fps= 25 q=28.0 size=     298kB time=00:00:31.00 bitrate=  78.8kbits/s speed=0.93xframe=  847 fps= 25 q=28.0 size=     301kB time=00:00:31.52 bitrate=  78.1kbits/s speed=0.931[mjpeg @ 0x5578d5a232e0] error count: 65b697ff00611d13
  
  [mjpeg @ 0x5578d5a232e0] error y=0 x=45
  
  frame=  859 fps= 25 q=28.0 size=     303kB time=00:00:32.00 bitrate=  77.5kbits/s speed=0.932frame=  872 fps= 25 q=28.0 size=     309kB time=00:00:32.52 bitrate=  77.8kbits/s speed=0.933[mjpeg @ 0x5578d5a232e0] error count: 6464e73ce47d9ba1e
  
  [mjpeg @ 0x5578d5a232e0] error y=73 x=10
  
  [mjpeg @ 0x5578d5a232e0] mjpeg_decode_dc: bad vlc: 0:0 (0x5578d5a2a328)
  
  [mjpeg @ 0x5578d5a232e0] error dc
  
  [mjpeg @ 0x5578d5a232e0] error y=74 x=38
  
  [mjpeg @ 0x5578d5a232e0] error count: 64cd4eab52d7b5de8

  When I do it Step by Step (Save the File via the Script and then transcode with FFMPEG) everything works fine.
  When I use a Pipe the Video Quality is bad with flicker.
  It only runs rudimentary good when I use a buffer, a crf higher than 35 and Youtube with Speed 0.25.

  First, it runs on a Raspberry Pi 3, now it is on a dedicated server. Is it possible that the Server still has not enough Power for it or is FFmpeg just limited ?
  OS : Ubuntu
  CPU : i7-3770k
  RAM : 8GB DDR3
  Graphic : Quadro M4000
  

  Have you some suggestions about what to change, that the stream would run better ?

• lavu/tx : support in-place FFT transforms

  10 février 2021, par Lynne

  lavu/tx : support in-place FFT transforms
  This commit adds support for in-place FFT transforms. Since our
  
  internal transforms were all in-place anyway, this only changes
  
  the permutation on the input.
  Unfortunately, research papers were of no help here. All focused
  
  on dry hardware implementations, where permutes are free, or on
  
  software implementations where binary bloat is of no concern so
  
  storing dozen times the transforms for each permutation and version
  
  is not considered bad practice.
  
  Still, for a pure C implementation, it's only around 28% slower
  
  than the multi-megabyte FFTW3 in unaligned mode.
  Unlike a closed permutation like with PFA, split-radix FFT bit-reversals
  
  contain multiple NOPs, multiple simple swaps, and a few chained swaps,
  
  so regular single-loop single-state permute loops were not possible.
  
  Instead, we filter out parts of the input indices which are redundant.
  
  This allows for a single branch, and with some clever AVX512 asm,
  
  could possibly be SIMD'd without refactoring.
  The inplace_idx array is guaranteed to never be larger than the
  
  revtab array, and in practice only requires around log2(len) entries.
  The power-of-two MDCTs can be done in-place as well. And it's
  
  possible to eliminate a copy in the compound MDCTs too, however
  
  it'll be slower than doing them out of place, and we'd need to dirty
  
  the input array.
  

  • [DH] doc/APIchanges
  • [DH] libavutil/tx.c
  • [DH] libavutil/tx.h
  • [DH] libavutil/tx_priv.h
  • [DH] libavutil/tx_template.c
  • [DH] libavutil/version.h