15.04.2025, 16:45
Okay, a few things,...
Cu Selur
- Sure, the gop size limits the number of potential gain from b-pyramids, but:
- you usually don't encode with a fixed gop size (you either let the encoder decide to use gop sizes up to a limit, or you let it decide for itself without a limit)
- A b-pyramid of level 1 requires 2-bframes, level 2 requires 3-4 b-frames and level 3 7+. So by liming the number of b-frames you also limit theoretical max level of the pyramid. It rarely makes sense to use too many b-frames or too high of a b-pyramid.
- you usually don't encode with a fixed gop size (you either let the encoder decide to use gop sizes up to a limit, or you let it decide for itself without a limit)
- Your OBS parameters, seem strange:
- vbv values will be ignored by x264 unless a level and profile is also configured. (this was years ago and afaik. this has never changed)
- using opencl is a bad idea, since it will be slower and less quality than using the cpu (it basically was declared useless in the way it exists nowadays in x264 unless in some rare cases on 15+years old hardware)
- 'filler' usually makes no sense unless you use some ancient editing or streaming software/hardware.
- vbv values will be ignored by x264 unless a level and profile is also configured. (this was years ago and afaik. this has never changed)
- NVEnc and b-pyramid with ffmpeg: '-b_ref_mode <mode>' is probably the value you want to use.
iirc:
The -b_ref_mode:v option in FFmpeg's NVENC encoder supports the following values:- disabled (0): B-frames will not be used for reference
- each (1): Each B-frame will be used for reference
- middle (2): Only (number of B-frames)/2 will be used for reference
- disabled (0): B-frames will not be used for reference
Cu Selur
----
Dev versions are in the 'experimental'-folder of my GoogleDrive, which is linked on the download page.
Dev versions are in the 'experimental'-folder of my GoogleDrive, which is linked on the download page.