[REQ] Rigaya's hardware encoders support

1. offizieller Beitrag

    Hi there,

    thanks for this great software !


    It would be extremely cool to better exploit hardware (GPU) accelerated encoders too.

    Rigaya's ones are certainly the most advanced (and updated) on the open source scene:

    - QSVEnc (latest release here)

    - NVEnc (latest release here)

    - VCEEnc (latest release here)


    Thanks in advance !

  • Vouk 1. Oktober 2024 um 15:15

    Hat das Thema freigeschaltet.
    • Offizieller Beitrag

    Can you tell me what's actually better with these "better" encoders? To me it seems they just use the already present fixed functions encoders in the hardware devices.

    Help to improve this plugin and support me on patreon or paypal. Thank you.

    Do you need a custom Voukoder build really fast? Click here.

    Riguya's encoders use ffmpeg libs and have almost every option working for the various gpus. Many of which are difficult to get working with an ffmpeg command line unless you know the trick. The quality for more typical 4k settings isn't as high, however. Not sure why that is. Math libs or data width, maybe?

    I dunno where to start with this one. 🙄

    NVENC is a hardware feature, meaning it's part of your physical Nvidia graphics chip. You can't "update" NVENC unless you buy a newer graphics card. The same NVENC settings are available to all programs (FFmpeg, etc.).

    Likewise, QSV and AMF are hardware features.

    Voukoder and VoukoderPro also use FFmpeg libraries. So I don't see any advantage to your pet program.

    Maybe try actually using VoukoderPro?

    • Neu

    Well, given that the quality of hardware encoding is not (yet) as good as software, there are several advantages in using Rigaya encoders over FFMPEG ones (that's why many - if not all - encoding-suites are adopting them):

    - many specifically-optimized filters (nnedi, for example);

    - better control of hw encoder;

    - deeper documented;

    - more fine tuned.

    That said, I'm personally very interested in using (Intel) hardware acceleration for filters and/or AVC "lossless" encoding.