It works for me. So i can tell you it works, though i cannot tell you why it does not for you.
Okay, I can revise my statement as well, as I got it to run. I updated DLSS swapper as well as updated to the newst Varjo Base 4.10.0.7 and now it works. Most likely something was messed up with my VB version I guess. As pointed out above, Open XR Motion Compensation currently messes up the eye-tracking and the focus area is somehow reversed in movement. But it works! Great.
I was reading some posts here and in another thread about quad views. I’m currently running a AMD Ryzen 7900 x3d with a Nvidia 4090 on a Varjo Aero. I was just experimenting with the settings and discovered something quite surprising. I’m getting better frames and lower time on the render thread. I’ll post some screenshots and my settings later when I get home.
There are people here with objectively better hardware getting the opposite effect. Someone with a 9800x3d and another with a 5090 were getting worse performance. I’m not sure why but I have a a couple of hypotheses. It could be my graphic settings. I have TLOD set at 70 and OLOD at 100. Everything else is set on high with the exception of ambient occlusion (off) and glass cockpit refresh rate (medium). I’m using the latest beta as of this post. I ran my test in SFO on 28R in the MAX8. I was getting 37FPS and render thread in the low 20s. with TAA. OXRTK was on but every feature was off. FPS would drop into the 20’s with foveated rendering off and render thread would jump into the 40s. I get better performance, low to mid 40s FPS, with DLSS but I need to bump up the resolution with OXRTK to make it look good.
It could be the processor. This is just a guess but a 9800x3d only has 8 cores and all are 3d v-cached. The 7900x3d has only 6 3d v-cached cores but has an addition 6 non 3d v-cached cores. Maybe the extra cores help increase the overhead? Or maybe it’s the fact that the non x3d cores operate at a higher frequency or whatever controls foveated quad view runs better on the non x3d cores?
Ok, took some screen shots of several different settings. Things aren’t as I originally thought. In fact, things are quite a bit simpler than I thought.
TAA, stereo rendering
TAA, FR on
TAA, FR on, NIS 120% (OXRTK)
DLSS Perf, stereo
DLSS Perf, FR on
DLSS DLAA, FR on
If you’re using TAA, there is quite a large benefit in performance using FR while DLSS takes a performance hit when FR is on. This all seems to confirm what’s been said. If you’re CPU limited, FR will make things worse. If you’re GPU limited, FR has a benefit.
I really do wish I had more control over the resolutions in FR. I feel that if you could increase the resolution only in the focus area, DLSS would have a benefit in with FR. Not only that, DLSS looks dramatically better the higher the pixel density is. The tradeoff in image quality in TAA vs performance in DLSS is worth it.
4 Likes
I’ve been playing with DLSS settings and Fixed Foveated Rendering (with Quest 3) a lot during this week. My experience is, that FFR does not work, if you are are using DSLL quality, balanced or performance. Image quality is awful. However, if you are using DLSS DLAA, FFR helps really much. With the newest beta (1.4.11.0) I’m finally able to use DLAA @36/72hZ on my Quest 3 and everything is really smooth, even I do not have the fastest possible HW 
I tried even TAA, but it didn’t look nearly as good as DLAA…
2 Likes
Hello all.
I have been searching for a comprehensive VR guide to MSFS 24, but it really seems like there is no such thing.
I am pretty familiar with DCS and its settings, in regards to VR, openXR, quadviews, etc. but cant seem to find any info about MSFS 24.
I am running a OG Crystal with 9800x3d and 4090, and performance is less than stellar in Sim Update 2
- from what I read here, neither pimax play integrated PimaxXR or PimaxXR Quadviews (mbucchias) have anything to do with MSFS 24 foveated option, so what is the best Pimax Play setup for a smooth experience?
- Is it still necessary to run the DLSS swapper program?
- Is there a comprehensive settings analysis somewhere to show what setting takes the biggest VR performance hit when turned up?
- Is openxr toolkit a performance gain or the opposite? I have read both
Thanks all
I did get Meta Quest pro. So it has eye tracking but when enavle foveated rendering in MSFS2024, there is still visible rendering square. Is eye tracking still disabled in MSFS2024. I didnt have time to check the latest, just had opportunity to buy Quest pro with 500e (4 times used).
Eye tracking worked in the sim last I checked.
If you’re using Quest Link, you need to go through the nonsense of enabling developer mode.
If you’re using Virtual Desktop or Steam Link, it’s a matter of enabling the right settings in the corresponding app.
That said, I think I have heard that the current Quest firmware (on the headset) broke eye tracking entirely, regardless of what streaming solution you use.
1 Like
I have v74 with quest pro and using vd. But i have also dev mode enabled because i use vd with reversed tethering method.
But eye tracking isnt just working. Rendering box in the middle. What was the test, where i can see is it working?
Edit. Got it working, but performance in awefull. Cant keep 90fps with SSW enabled and MSFS foveated rendering enabled and GPU load is 50%. With Q3 i can have constant 72fps without SSW.
I get much worse performance with DFR as well. I can’t get the eye tracking to work on the quest pro with MSFS but it does work with DCS world. Is there anything else required to get it working?
1 Like
I had this working in earlier betas, but it seems to have stopped again.
I know eye tracking data is being reported by the headset (QPro) as I can test it in other applications. I’m using the Oculus OXR runtime via AirLink.
(Running v77 PTC on the headset and v76 on PC - due to a Meta bug in v77 PC software)
I’m on SU3 beta, version of today. I didn’t see the different color of the rendered square any more while flying through clouds today. Either I got lucky in terms of the lighting conditions, or they fixed that problem. Has anybody else noticed that?