Implement Quad Views for VR rendering to improve performance

No Absolutely!

Good new but just “spectacular” is lacking in some details. :slight_smile:

But I’ll be blown away if the menu position bug and end of flight bugs are fixed. :smiley:

Good morning everyone. Any news for the implementation of Quad Views for VR in Flight simulator 2020? This is an amazing feature. In DCS it allows an incredible performance boost. It definitely needs to be implemented. But there are few requests for it, probably because very few people have a VR device with eye tracking. The implementation is not difficult and involves little work.

Yes, anything to make VR better will be much appreciated. :slightly_smiling_face:

1 Like

I’m finding:

image

See Microsoft! It’s not just a gimmick!

2 Likes

Untill Meta puts it in their non-pro model I doubt we would see it in the sim. I do hope I’m wrong! :grin:

Here’s another video demonstrating the value of these techniques:

With future higher resolution headsets, this will become increasingly important. It’s absolutely transformative in DCS as Steve says. I know I’m playing a lot more DCS these days because of it!

1 Like

Great video by VR FlightSim Guy.
Quad views has to be a must for VR in MSFS2024.
This is the sim that will benefit most from quad views with it rich graphical environments.
Build it in now to future proof MSFS2024 VR.

There are several barriers that you’re not seeing and that’s making this very unlikely to happen:

  1. no platform other than Varjo and Somnium (aka both extra niche) support quad views out-of-the-box.
  2. it’s not trivial to implement quad views standalone in-game (meaning without my mod), especially for Oculus since they don’t support fovMutable.
  3. it’s not trivial to implement quad views with eye tracking standalone in game on many platforms due to the lack of support for the vendor-agnostic eye tracking API (for example on Oculus, this API is not implemented at all).
  4. assuming MSFS 2024 may have a stronger focus on DX12, the Quad-Views-Foveated mod will not work. So back to 1) 2) and 3).
  5. without the game engine implementing stereo instanced rendering, quad views will be a waste of time as it will raise your CPU usage significantly.

So I wouldn’t get my hopes up.

    1. and 3) are all driven by PC VR vendors having deserted the gaming industry. Most notably Meta not caring at all about PC VR gaming and not implementing any of the OpenXR support that makes sense for cross-vendor support. They only care about their Unity/Unreal plugin (which is basically a test platform for their Android AR/VR product), which they have made sure will not work on anything but their limited OpenXR runtime.

The Quad-Views-Foveated mod was driven by 1) 2) and 3) to attempt to fix the damage created by Meta.

  1. is the result of me deserting this dead industry. No vendors investing. Most content is now only designed for Oculus and their weak feature set. Nobody except Pimax focuses on PC VR gaming, and they are too niche to make a real difference unfortunately. No matter how much Meta is killing this industry, you are still giving them your money to do so.

Bottom line: unless Meta gets their thing together and addresses 1) 2) and 3), there is no future for quad views. In fact, we can even say without them stopping their carnage, there is no future for PC VR.

Sorry for the rough wake-up call. This is the reality today from the mouth of an industry expert who has now deserted this industry (my last day as a Mixed Reality/OpenXR employee was last month).

2 Likes

Yeah, i was just going to post similar. You forgot about mentioning eye tracking as well. And mainstream doesnt even include that.

Thanks for the reality check! All those seem like real barriers, but how was it so comparatively easy in DCS? Very high end PC VR is a niche within a niche, but you were able to do it as a mod. One would think that MS/Asobo could do something similar.

Who said it was a mod? Or easy?

That’s a solid question.

The key thing to understand for DCS is that this feature was implemented as the result of some sort of contract between Eagle D and Varjo. I know nothing of the specifics, but I assume that “somebody at Eagle D was paid by Varjo to do this work”.

So it’s about incentives.

For DCS the incentive was “Eagle D wins a contract with Varjo for doing this work, with some sort of business/financial advantage”.

What is the incentive today for Asobo to spend time to deliver a feature that will be only available out-of-the-box (unmodded) on:

  • the Varjo Aero, an EOL device that’s been abandoned by their vendor
  • the Varjo XR-4, a $4000 device out of reach for the large majority of people
  • the Somnium VR-1, another $2000 device that hasn’t really made a name for itself yet AFAICT and still mostly out of reach for the large majority of people

Bottom line: you are talking about them building a feature for maybe 1 or 1.5% of the current VR population. Asobo dedicating development efforts to make the experience of such infinitesimal portion of the gaming population does not present any business/financial advantages.

You’ll say “but we can also use QVFR on Pimax, Reverb Omnicept and more importantly Quest Pro!? And even as an FFR option on any headset”. The keyword in the previous paragraph was “out-of-the-box / unmodded”.

Who is accountable for supporting the modded solution?

The quad views support in OpenXR, whether 1.0 or 1.1 is an intentional feature to be provided at the platform level, not only game level.
(btw the only difference between the two is naming, even in OpenXR 1.1 the concept of quad views is a totally optional feature for the platforms to implement).

  1. Platform provides support for quad views via OpenXR API
  2. Game consumes API

Here we have 1) missing from all major plarform implementation.
The API layer (mod) takes the place of 1), by delivering the OpenXR API on top of the lower-level OpenXR stereo API.

It is not the responsibility of the game developer to provide 1). It is possible for the game developer to provide 1), however the #1 vendor in the industry has (purposedly?) made that significantly more complicated than it should be, specifically point 2) and 3) from my original message. The majority of my mod is dealing with these 2 points, and this is not trivial to understand/implement for a game developer. Writing platform code is a different discipline and state of mind than writing game code. I would not be able to write MSFS game engine. Conversely, they are not able to write QVFR.

FWIW on something like WMR which has proper fovMutable support, doing in-app quad views FFR would be very easy. It’s in fact a test app that my former team wrote for WMR 4 years ago. But no, Meta won’t support this type of stuff.

So you end up with the sole accountable party being me, on my free time.

QVFR was implemented specifically for DCS, and therefore implements the minimum needed for DCS, that is DX11. Implementing QVFR wasn’t too hard (because again, platform dev, specialized in OpenXR - not game devs skills). Supporting QVFR was way harder. For many reasons, but in general people running many mods, different game configs, different QVFR settings… think of it that for every one of the tools I’ve developed since 2021, the balance is something like <20% feature development (fun part!!) and then >80% user support, UX and often dealing with rude and annoying people. As you can see, this isn’t a desirable position to be in, spending 4x more time on chores than on things you enjoy doing.

All of this with no real business or strategic benefit (if we speak from a company perspective).

TL;DR

There is no real incentive to implement quad views in any application today, not until all or at least the dominant platforms support it out-of-the-box.

1 Like

See my reply above, especially about the responsibility of platform to deliver their side of things.

Thanks for that analysis–that’s really interesting. As you say, hopefully Meta will eventually get their act together, but you never know. We can still hope for improved PC VR!