NVEnc multipass in NVidia SDK ver 10

  • Vouk

    Set the Label from Feature Request to Investigating
  • Just a heads up and I know we're dependent on ffmpeg but this just got released in as part of a standalone encoder:

    https://github.com/rigaya/NVEnc/issues/248

    Let's not forget, that Rigaya's NVENC is a state of art HW encoder since its very beginnings. It is de-facto the only encoder which uses all known encoding features/options of an Nvidia card. FFMPEG have much fewer options. Rigaya also incorporated most of the Avisynth features too, He also transformed most of the Avisynth options to be Cuda accelerated. It is also much more stable software than the buggy FFMPEG version of NVENC

    Edited once, last by No limit ().

  • It would be ideal for Voukoder to hook into (and support) the Rigaya package for all things NVENC and then use FFMpeg for everything else. This would definitely complicate/duplicate a lot of Voukoder code. There is another thread somewhere way back where the issues were addressed.

  • The modular Voukoder successor is planned to not rely on FFmpeg encoders only.

    Do you plan to implement VFW or directshow version of Voukoder? Again, they work well not only in avi output container, but VFW and directshow codecs can work nice in mp4 mkv ts/mts flv etc.. other container formats too.


    The FFmpeg is so buggy (and dumb) for HW encoding, that's why it was cancelled by Staxrip and Hybrid transcoders among their HW encoding options, and they replaced it with Rigaya's encoder engines. FFMPEG did not know the internal transcoding option, when both the decoding and encoding data were processed internally within the videocard. It makes transcoding much faster in a comparison with FFMPEG traditional transcoding when the process goes trough the CPU & system memory etc... .

    Edited 5 times, last by No limit ().