This tools is a must have for MSFS and I can’t thank the Devs enough for getting it out for the community to use. It’s very disappointing that MS didn’t bother with these VR features in either MSFS or MRP. I’m just very glad we got them now.
I’m using a G2 driven by an EVGA 3080 ti FTW3 and both OCed 10700k and 32GB DDR4. Using FSR 90% at 80% scaling, 85% world scaling. My MSFS and OpenXR scale settings are at 100%. I get a nearly consistent 40 fps everywhere, with excellent visuals and smoothness (at least considering the limitations 40 fps). AFAIK there is no need, nor advantage, to use non 100% scale settings outside of this tool. I imagine that doing so is placing additional execution overhead on either MSFS or OpenXR or both. Any thoughts on this?
This is intentional. If something causes a crash shortly after changing settings, we adopt a principle of precaution as we don’t want to store settings that might be causing the crash (and you end in a CTD loop).
I really want to thank the devs for this nice tool. I am using it to reduce the slight shake I have in my G2, even when my heart beats. I also like the FPS counter.
Unfortunately on My G2, RTX 3080, my previous posts from the earlier NIS scaler versions still hold true. Having an in game TAA scale set to say 80% and NIS/FSR off, I get 37fps on one particular area I use to test frame rates (the start of the NYC activity). With TAA set to 100% and NIS or FSR on and set to around 80, I also get 37/38fps, but the quality is most definitely not as good as TAA. I have tried varying sharpening levels for both NIS and FSR. I can see the in cockpit text looks much better without NIS/FSR. I think if it was the other way around and we all had NIS/FSR for many months, I think many people would be amazed if TAA was then introduced.
I really urge people to really compare the two, to see at the same quality level, is NIS/FSR really better for you. I would really bet that it isn’t. I would also bet if you are happy to have slightly less quality and higher performance, that reducing TAA would also give you that same result, without the shimmering NIS/FSR undoubtedly brings due to the nature of how it works and how its limited in a way that TAA isn’t. My key take away is for my own self anyways is that TAA set to the desired level to reach the performance you want, with NIS/FSR off, will give a better image, due to how TAA works.
I’m very excited by some of the upcoming features of the tool though, so keep up the good work.
I’m using -20% to start with, I might go further. I’m going by some of the figures quoted in the toolkit docs that -20 to -40 are a good starting point.
Hi, I only changed the prediction dampening to -40 after some experimentation, which solved the problem of shaking in my case; I left everything else at the default value as I already had a sharp immage.
When you say “excellent visuals”, are you really saying you don’t see really distracting shimmering/bad aliasing on autogen buildings, trees, and some other objects when they are at mid-distance?
Is there any possibility to implement a feature which would allow us to quickly force the reprojection to stay at a given rate during the flight? By default it is automatic which causes stutters each time the app decides to switch between 1/3 and 1/2 rate. I’m on 60Hz mode so it is 20 and 30fps respectively.
Locking reprojection is already ready for the next version of the toolkit, and this option will be available for WMR headsets only. So I hope you have a WMR headset.
I’m almost certain that this is very subjective and fully depends on the user. I can definitely see the “shimmer” which is essentially antialiasing on all the hard surface edges, so buildings and edges of any water bodies, do shimmer.
However from MY PERSONAL enjoyment of the sim, I can still say that I’m getting “good visuals” because I PERSONALLY can ignore the shimmer and it doesn’t bother me that much.
Some people cannot ignore it and hence are surprised that some are finding the visuals to be “good” or “excellent”.
Granted, that the “shimmer” for me is visible ONLY on the distant objects and is most prominent on the buildings.
So it also GREATLY depends on how the user is playing in their sim. If you fly in urban areas a lot, you will experience shimmer a lot more and might perceive graphics to be “worse”. If you are flying in more rural areas, then the shimmer is less noticeable on the trees and they will perceive graphics to be “good”.
I’m assuming if you are flying an airliner and are tens of thousands of feet above the world and mostly likely more concerned with the instruments rather than the outside view, except for take of and landing, you might not even care about this shimmer.
So everyone’s perception will vary, even though the underlying experience is relatively similar.
I’m not seeing any shimmering/bad aliasing on what you describe. Everything outside of the cockpit looks great in that regard, but 40 fps does reveal itself on sharp turns (perhaps that’s what you’re referring to). I notice slight AA artifacts on the edges of the cockpit glass panels and occasionally on the edges of the windscreen/canopy. Using TAA and have reprojection disabled. I also can see a hardware limitation of my G2, which occasionally shows as a gridded graininess on terrain and sky features, i.e. its version of the dreaded SDE, but calling it that is really unfair compared to what SDE can be like. BTW, I only fly GA planes and don’t often go above 15k ft.
I’m on Quest 2 HMD, so when bringing the up-scaling down to 80%, my final resolution (according to the OpenXR Toolkit) is around 1500x1500 I believe. That also needs to be taken into consideration that all HMDs have very different final resolution. At 1500 there is definitely noticeable aliasing/shimmer on the hard edges, however if I bump the resolution back to full 100% or even super-sample it, I obviously get a much cleaner picture.
I’ve got a nagging question… does the Ultraleap hand-tracking “stuff” actually work? Like, by work I mean (not to be critical) “fingers grabbing knobs and spinning them” “index pointer finger pushing buttons” etc. Anyone got it working to those standards?
Leap Motion with OpenXR kit is emulating VR controller, as there is no native Leap Motion support in MSFS. So “fingers grabbing knobs” is not exactly possible. However, you can see your skeleton hands and I configured the “ghost controllers” to align with my index finger, with a “finger gun” gesture for activating trigger, so yes, you can point your finger and touch a knob and press the button, flip a switch etc. Rotation is a bit finicky, but so is the hardware VR controller rotation - it’s a MSFS bug, exacerbated by Leap Motion rotation + gesture which is difficult to track. I use encoder box for precise rotations.
There are a couple of bugs that need to be solved so it works reliably. Right now it works for a while and then interaction can get lost. But when it works - it really does: pressing buttons, triggering switches, even grabbing levers and rotating knobs (limited by MSFS).
Next version should hopefully include translating haptic “tick” to the trigger press. That way you would just touch a button with your index finger to press it, same for switches.
You can use a pinch gesture to grab knobs and turn them, but your hands must remain fairly steady.
You can use that same pinch gesture to pull levers, and “actuate” buttons (not really a push motion). As @RomanDesign said in his post, with the addition of “haptics response”, we’re going to try to make it that index finger pushing buttons will work the way you expect, but it’s not done yet.
The main issue at this time is the loss of interaction bug, which seems to be related to the use of Motion Reprojection and frame limiting.
There’s an old video of me showing it back in December, where you can see me turning some knobs (altimeter and console dimmer) and interacting with throttle/brakes).
