Part 1 :
Test without modifying the .txt file value in steamvr folder.
( Pitool at 1.25 + SteamVR at 80% and 110% clearly show that both benchmark use the GPU the exact same way. So no supersampling result difference between both test. If you think it look better that way at 100% instead of 70% or 80% it is just in your brain.
Part 2 :
Test with the .txt file value changed from 4096 to 8192 for the max vr resolution setting.
( Pitool at 1.25 + SteamVR at 80% clearly show that the GPU usage is now 10 to 15% higher than before as a sign that the superampling is now taking effect. Outside camera I did the same test but with the supersampling in steam vr at 100% and my GPU usage was another 10-15% higher fluctuating between 72% and 80% instead of 60% to 70%.
I’m only a few real-days of use into my Pimax 5K + and noticing similar issues, not just with AC. For example, in PCars 2 with the FOV set to Large none of my SS settings seem to have any effect whatsoever. Everything works the moment I turn the FOV back to Normal. (AC works fine with the Large FOV, however, and is gloriously crisp!). What we need is a specific thread where people can list all currently known issues, sorted by game. I spent about an hour trying to figure out what the heck with going on with PCars 2 looking terrible; would have been great to be able to look that up someplace.
The default.vrsettings file in SteamVR is the issue. If you have your headset FOV set to Large and are using PiTool 1 or above and haven’t changed the
“maxRecommendedResolution”: 4096, then any render settings with PiTools and SteamVR that produces over 4096 render resolution be it vertical or horizontal per eye, SteamVR will just stay with the 4096 value.
Unfortunately in my experimentation, the size limit you can tell SteamVR to use is 8192. Beyond that and you get out of sync per eye scaling issues.
For me personally, I notice performance differences also with setting SteamVR at 80% compared to 100% with the later being more effective performance wise even though you are rendering more pixels. I have experimented with SteamVR auto setting, 40%, 50%, 60%, 80% 100%, 150% and 200% render settings. 50%, 100% and 200% seem to get me the best performance at a predetermined PiTool Render setting.
Maybe it is the math and image process conversion that might have to be done when you use non binary alterations of the original PiTool render setting info that takes its toll on ones system. Just a thought.
With your 8192 setting and PiTool 1.25 with SteamVR at 100%, did you try FFR? I find it does a good job with performance without much visual degradation of the in game image.
