Recherche avancée

Sur d’autres sites (7935)

• Does 'keyint infinite' improve overall compression with crf and mb-tree ?

  23 avril 2012, par HitomiTenshi

  After reading what intra-refresh does, I kinda got confused. I'm encoding for YouTube and YouTube will re-encode the video-file anyways, so I thought "why not removing all these heavy IDR-frames ?" I just want to compress my video as hard as possible, enabling everything that could help me getting a lower filesize, but still maintaining a high quality.

  I used to test around stuff with lossless QP encoding, but the only thing I could max out was the merange. Here is a paste of my x264 settings : Pastebin.

  I want to achieve highest compression while maintaining visually lossless quality. (using crf values around 10 - 13, and merange 32)

  Could anyone give me advice on how to compress my video super hard (without touching the crf value !). I also want to know if it's true, that keyint infinite reduces overall compression.

• parallel ffmpeg filter complex pipelines

  3 septembre 2021, par Paul Eppner

  Is there a possibility to run ffmpeg filter_complex filter in parallel pipelines ?
On the page http://trac.ffmpeg.org/wiki/Creating%20multiple%20outputs you can find an example under the heading "One filtering instance per each output". However, the filters mentioned there (boxblur,negate,yadif) are executed one after the other, although they could also all be executed simultaneously. I have quite complex filters and therefore long delays when the filters are executed one after the other.

  


  I hope there is another solution.
Cheers

  


  ---

  


  A few hours later I realised that the delay was probably not due to the pipeline, but to the analysis of the imports. However, I then only use the video and audio of the 1st stream. nevertheless, the delay time is then enormously high. Are there any ideas on how I can prevent this delay ?

  


      -i "rtmp://127.0.0.1:1935/$app/$name-de" \
    -i "rtmp://127.0.0.1:1935/$app/$name-en" \
    -i "rtmp://127.0.0.1:1935/$app/$name-fr" \
    -i "rtmp://127.0.0.1:1935/$app/$name-es" \
    -i "rtmp://127.0.0.1:1935/$app/$name-it" \
    -i "rtmp://127.0.0.1:1935/$app/$name-de" \
    -i "rtmp://127.0.0.1:1935/$app/$name-en" \
    -i "rtmp://127.0.0.1:1935/$app/$name-fr" \
    -i "rtmp://127.0.0.1:1935/$app/$name-es" \
    -i "rtmp://127.0.0.1:1935/$app/$name-it" \
    -i "rtmp://127.0.0.1:1935/$app/$name-de" \
    -i "rtmp://127.0.0.1:1935/$app/$name-en" \
    -i "rtmp://127.0.0.1:1935/$app/$name-fr" \
    -i "rtmp://127.0.0.1:1935/$app/$name-es" \
    -i "rtmp://127.0.0.1:1935/$app/$name-it" \
    -vsync -1 \
    -map 0:v -c:v:0 copy \
    -map 0:a \
    -c:a aac -ar 48000 \...```


  


• nomenclature #3324 : Déplacer les fonctions typographiques dans un plugin séparé

  29 octobre 2014, par Fil Up

  Quelques autres améliorations à faire (pas très difficile, mais pointilleux : faire des cas tests).

  1. toujours rendre les espaces insécables (même en anglais) quand l’espace est déjà présente (a ! => a !)

  2. en français-français, ajouter les espaces insécables, mais pas en français-suisse (enfin, différemment)
  http://fr.wikipedia.org/wiki/Comparatif_des_différents_codes_typographiques_francophones#Diff.C3.A9rences_entre_les_typographies_fran.C3.A7aise_et_suisse

  3. remplacer les insécables qui vont bien par des fines (en option avec réglage ou define). Le code est déjà fait, il suffit de l’intégrer (avec les tests idoines) cf. http://core.spip.org/issues/1461