Answering to the original thread is unfortunately not possible.
So, I’ve tried this thing. I have an overclocked RTX2080, a pimax 8K+. In pitool use the normal FoV, the render resolution is 1. In SteamVR the render resolution is on automatic, no special settings for HL:A.
For the command line I used these commands:
-console -vconsole + sc_no_cull 1 + vr_multiview_instancing 0 + volume_fog_enable_stereo 0
When I start HL:A with it, I have an average of TWICE AS HIGH frame time than when I start it without it.
The discoverer of this thing (NextGenVR) now writes:
"For those wondering what Multiview instancing does. It basically saves on some rendering costs by rendering the scene once for the left eye, and then shifting and adjusting the image for the right eye. This works as intended on smaller FOV headsets like the index. But for wide FOV headsets, the rendering gets messed up. So disabling it, means rendering the same image again for the right eye. "
So it’s actually clear why. But then the question is: what does the whole thing actually bring? If I run HL:A with PP it will cost me about 25% GPU power.