No. The 2 VR photos were taken in 2 completely different FS sessions to enable me to triple-check I had everything, especially the render scale), set properly before I entered the cockpit. And while it’s pretty much impossible to ensure that my head was in the exact same spot for both pics, I made all reasonable efforts to make sure it was very, very close. So it is possible and pic was a little closer than the other, it was a matter of fractions of an inch, at least I think and hope so.
ditto, I’m glad you shared these photos which are showing the difference I can see!
Sorry for looking at the grass instead of the compass though, but honestly the compass is not as huge a difference for me than the grass. I find the compass looking too similar and just different by size, not much by detail. However the grass shows more details at first glance.
As much as I feel it is important to tell what’s wrong when I find it is, I’m not shy telling what its good when I find it is too… and if you search my post history you’ll find that I’m always trying to offer possible solutions or explanations to the problems I’m reporting, in a spirit of constructive criticism (well my spirit is always constructive in good will, my words might be more criticizing sometimes).
I don’t know whether it will break this trick, and I’m unfortunately not at liberty to tell what it will be like, but it is hard for me not to be enthusiastic. All this to say I’m quite certain you shouldn’t be disappointed even if the trick is no longer working…
Just a small advice: When I need to do such screenshot comparison, I always have my headset on my desk, not on my head, with a small towel (for cleaning glasses) on it to trigger the proximity sensor (G2 have one I’m guessing?). Like this, after resetting the VR view, all screenshot are exactly take at the exact same position, so comparison is far more easy switching between them
It’s actually worse. ASW 1.0 create even more artifact. Honestly, even on scene where I can fly 45fps locked ASW ON like at UT25 (monument valley) and using a light plane (like the Cri-cri) the game is pretty fluid but artifact of ASW 1.0 when I turn my head or look on the side kill my immersion.
After seeing what ASW 2.0 can do to fix that, I prefer to fly ASW OFF with MSFS. I can understand the 18 lock or 30 lock don’t allow to play ASW OFF (it will be jittery festival). Only 45 and 90 (obviously) accept ASW OFF. And the RenderScale trick don’t work also with this 18fps, I also tried
When changing 2D settings it affects the Mirror (it shouldn’t but does), so it’s pointless taking screenshots of the Mirror image as they do not represent what is being seen in the HMD.
Do you mean it is only changing rendering in the mirror view, but this doesn’t reflect in the HMD itself? Not just UI elements like gamer tags or POI labels, but actual rendering of the scene?
If that it the case, could this mean the mirror view is a separate rendering from the one sent to the panels, as if the game is rendering things twice (once for the mirror and once for the HMD)? I hope it is not otherwise it can only be affecting overall fps in VR.
To answer your first question yes that is correct, changing the 2D settings only affects the Mirror not the HMD. This is actually an issue as no other VR game/sim that I know of acts this way as the Mirror is supposed to be a direct copy of the feed sent to the GPU for the HMD (there’s a flow chart somewhere on the web that explains this better).
That is exactly why I have classed it as a bug as per the link above, rendering twice is pointless and performance eating. I hope that Asobo, in their rush to release VR, know about this and have fixed it for SU5 so that it becomes part of the performance update, if not then they’re really missing a trick somewhere and need to be informed.
As I monitor with a precise test scene not involving random or external traffic etc. I can say for sure than when I tested this 200% render scale trick I didn’t have any difference in VR regarding CPU and GPU usage between having the monitor view windowed 1280x720, or 1920x1080, and both with PC RenderScale from 30% to 200% step 10%, and vsync OFF or ON (yeah, I spent time seeking this trick).
My VR FPS was as usual 45 locked and my GPU usage was 83% all the time.
So I don’t think 2D settings affect performance of the VR render, at least on my configuration. Or, possibly, they affect VR perfs the same amount all the time (large or small). We will know if one day they give us a way to fully disable it.
Saying that, I didn’t studied the monitor image when playing with the PC Render Scale, only the result in headset, so if that is true (I believe you @MarcG888, I voted) this is disturbing…
Devs have really unwanted interactions between both PC and VR settings they need to fix, seriously
Edit : possibly I was CPU limited during my test…Thinking out loud…
I still couldn’t resolve myself to thinking they’d render twice, instead of just doing a blit of the render buffer pixels into a window for the mirror view. This must be something else which is causing the differences in the mirror view?!
Easy to test for yourself, put everything to the lowest in 2D but keep VR settings as per normal and you should see the difference on the Mirror output. I’m interested to know whether this happens across the board or just certain setups (I’m using WMR through SteamVR for example).
I’d do some benchmarks of my own to see if there is a difference (I may of done this before) by changing 2D settings, but I’m awaiting an HMD replacement…!
Asobo were never going to add VR Support in the first place until the outcry from the community persuaded them to do so.
Thusly it was released “late” as a Patch/Update and clearly wasn’t “finished” in any way shape or form, it’s initial (and current) state is still one that is clearly rushed “just to get it out there”
They have said in a recent Q&A that they are revisiting VR later in the year and have some good things planned for it, so at least that’s something.
So to go by Point #2 it shows with the way the Mirror is handled that it’s clearly not working as VR is supposed to work with regards to the Mirror image itself, as above this may well be dealt with in SU5 or if not hopefully later in the year (ideally the Mirror should be just the output of One Eye not both).
Ultimately changing 2D settings should not in any way shape or form affect the Mirror image, that’s the key issue here so it’s not surprising that this threads “trick” is working for some and not others.
Ok so just tried and I can confirm PC Render Scale at 30% or at 100% or at 200% don’t affect my mirror view, and don’t affect any of my performance or cpu or gpu usage (oscillating between 81% and 83%). For reminder I use MSFS bought from Steam, an Oculus CV1 directly (no steamVR). So it should depends hardware.
Saying that when testing few minutes ago I found a new bug… When PC Render Scale was at 30%, my PC render view was affected obviously before switching in VR. But switching back and my 2D view was using the 70% Render Scale of my VR Render Scale settings. Going in PC settings, 30% and render to 576x324 still displayed. I had to move it back and forth and save to retrieve the pixelated view. Pffiiuuuuu…
Screenshot I just made:
First with PC Render Scale at 30%, and monitoring:
Did I just quoted myself in the same message? Wooo first time I do that
That could explain why I don’t have the benefit of PC Render Scale at 200%. In fact, I never have the PC Render Scale at 200% even if PC 2D settings display 200%, as my PC Render Scaling is always in fact equal to my VR Render Scaling 70% when I’m inside a VR session. That’s the bug I posted few days ago in Zendesk, how to corrupt PC Render Scale. That’s with my Oculus configuration, so the way mirror is managed is headset and software dependant.
I’m not sure which scenario I prefer - that Asobo has already patched this and is watching with amusement 600 posts and hundreds of hours of us trying to figure it out for ourselves - or That Asobo is reading this thread like the rest of us and desperately willing Cpt Lucky and co to make a breakthrough
Sadly I think they are far away from fixing anything in VR right now. They have other priorities before July 27, and I can understand that. Just hoping they will fix all this mess regarding interactions between 2D and VR settings affecting differently each users. I’m guessing people can’t reproduce my own problem of PC Render Scaling falling at the same value of my VR Render Scaling setting like I do.
A mess for sure…Patience we need…
No matter what settings I change, including using full screen 1K or 4K, windowed 1K or 4K, it doesn’t change anything in the HMD. However what is certain is that when setting the screen in 1K, there are less pixels on the screen than each view in the HMD (960x1080 vs ~2200x2500).
When using 200% render scale you’re most likely enabling the super sampling Shaders in FS2020, which would be the additional pass post render to “shrink” the render back from 4K (200% render scale) to 1K (monitor res). It is at a minimum implemented as a bitblit with bilinear filtering. In 2D this is revealing more details because it is combining and merging the information of 4 pixels into 1. This additional pass is expected to be deactivated if your 2D render scale is 100% because there is no need (1 pixel = 1 pixel).
My take on this “trick” is the following:
It is possible that when in 100% VR mode and 200% 2D mode, the VR Shader setup code has a bug and keeps this post-render shrinking pass enabled in the Shader code. In this case, it is possible the mirror view “somehow” is displaying a much better image because it would be using the same “averaging” shrinking method as the one used with super sampling, instead of using a much simpler nearest neighbor resize blit.
Since on a 1K monitor, each half of the mirror view is 960x1080 pixels, most likely less than your HMD panel res, or your OXR render res, averaging your HMD mirror view pixels into this lower res mirror view must be more legible in this case.
This is just speculation though because I really have no clue at all, I’m just trying to make sense of all this. But this doesn’t explain why some of you are also seeing more details in the HMD itself and not just in the mirror view!
For those of you where this trick is working in the HMD (not just mirror view), you might want to compare the visuals you get in the HMD between:
Maybe. But it’s our only choice, since it’s unlikely that everyone on the forum is going to come to my house to put on my HMD and see it for themselves. And it does clearly show the increased clarity. Not nearly as strongly as if people did come to my home and see for themselves, but it’s clear pictorial evidence that even Stevie Wonder can see, especially the zoomed in pics of the compass.
