VR experiences IL-2 Korea

[brickcommander]

7 hours ago, Syphen said:

I can fly free missions with playable framerate on my Reverb G2 with 12700k and 3080ti, all normal settings. Once I try the B29 intercept, it becomes a slide show.   I miss the performance I got in great battles. 

To be fair, if you flew a mission with 60 plus bombers in great battles, it would also be a slideshow...

[Gunfreak]

38 minutes ago, brickcommander said:

To be fair, if you flew a mission with 60 plus bombers in great battles, it would also be a slideshow...

But won of the selling points was that you should be able to fo this now.

[brickcommander]

1 minute ago, Gunfreak said:

But won of the selling points was that you should be able to fo this now.

True. And i do thin it performs better than a similar setup would in GB, but it isn't good and that mission is unplayable in VR.

 

That said, in normal scenarios- after a lot of tweaking on my 5700x3d,5070ti system- i get similar framerates as GB with far superior visuals.

 

I am also however very annoyed with the pilot head limitation. Turning it off doesnt work and every small yead movement- particularly up/down- moves the whole cockpit around me. And the menu screens are too close and it is uncomfortable. I would imagine both issues are easy fixes that will he resolved soon.

[Blooddawn1942]

After fiddling around alot, I managed to have acceptable fps on my Pimax Christal with graphical settings in the upper range. I just turned the indirect lighting and HDR off and set shadows to medium. 

I could gain a lot more fps by lowering the settings but right now I like what I have 

 

There remains 2 significant issues in my opinion. The world scale which seems off by 10% or so and the from time to time horrible blurry ground textures. 

The floating menu is unusual, but not that big problem for me. 

After several hours of flying after massively fiddling with the settings (but this is part of the deal when a new title arrives and was to be expected) I can confidently say that there is much to like in IL2 Korea.

the attention to detail, love and thought that went into so many aspects of the Sim is stunning. This is so immersive. Be it the gentle floating of the silk of the parachute. Be it the sound of trains and water when rushing above a train or dam. Everything feels so alive. Ground pounding in general feels absolute meaningful.

Sure, there are many aspects to be improved, but anyone still remember only having the 109 and LaGG3 without any amarment, when Battle of Stalingrad went into EA? We're still at the very beginning and I have to say I'm now very positive about Korea! 

[Blooddawn1942]

One thing I have forgotten to mention. I certainly don't like the overdone brightness in VR. This has been stated by others before and I consider this a major issue. It's even worth than the overdone brightness when you look into the sun we already have in Great Battles. 

Even turning off HDR do not help. There are conditions were I absolutely can't see the reticle. 

Please give us a filter like a sun visor like in some DCS modules. In this case it would simulate sunglasses. The technology is there as we can see with the reticle filters in the MiG-15 and other aircraft. 

Edited Saturday at 08:40 AM by Blooddawn1942

[MoleUK]

13 hours ago, Jade_Monkey said:

 

Sounds like Deferred vs forward rendering has something to do with it:

 

Source: https://il2-series.com/en/faq/question_answer/

 

Except that doesn't make any sense.

DCS uses deferred rendering, not forward. The performance gains from quadview with DCS VR are absolutely immense and borderline essential. For pushing the very highest resolutions or for squeezing more performance out of older GPU's it's a lifesaver. Particularly in combination with DLAA/DLSS which can help reduce the aliasing in the lower resolution outer view sections.

Not sure how the IL-2 devs came to the conclusion that it would only be a 10% improvement. Nor why they would think it was more suited to forward rendering.

Edited Saturday at 09:55 AM by MoleUK

[AEthelraedUnraed]

1 hour ago, Blooddawn1942 said:

One thing I have forgotten to mention. I certainly don't like the overdone brightness in VR. This has been stated by others before and I consider this a major issue. It's even worth than the overdone brightness when you look into the sun we already have in Great Battles. 

Even turning off HDR do not help. There are conditions were I absolutely can't see the reticle. 

Please give us a filter like a sun visor like in some DCS modules. In this case it would simulate sunglasses. The technology is there as we can see with the reticle filters in the MiG-15 and other aircraft. 

Might be related to the headset; so far I haven't seen any overdone brightness on my Quest2.

Regarding the reticle, it now has adjustable brightness: https://forum.il2-series.com/topic/3828-new-gunsight-commands/. IMO it's actually a good thing that it's not always visible; it makes sense that you'd need to set it brighter on a bright day than, let's say, at night.

20 minutes ago, MoleUK said:

Not sure how the IL-2 devs came to the conclusion that it would only be a 10% improvement.

The Devs did some experiments with it, and the 10% improvement was based on those. Not sure how you came to the conclusion that it would be more, based on the results from an entirely different engine.

Edited Saturday at 10:09 AM by AEthelraedUnraed

[Nerpo]

O! My ... 

Edited Saturday at 11:26 AM by Nerpo

[MoleUK]

13 minutes ago, AEthelraedUnraed said:

Might be related to the headset; so far I haven't seen any overdone brightness on my Quest2.

Regarding the reticle, it now has adjustable brightness: https://forum.il2-series.com/topic/3828-new-gunsight-commands/. IMO it's actually a good thing that it's not always visible; it makes sense that you'd need to set it brighter on a bright day than, let's say, at night.

The Devs did some experiments with it, and the 10% improvement was based on those. Not sure how you came to the conclusion that it would be more, based on the results from an entirely different engine.

Because the effect it has it not dependent upon the game engine. It reduces GPU load and increases CPU load, that is the exchange you're making alongside the drop in visual fidelity in the lower resolution rings.

For a SP mission, it has already been stated that in Korea the AI can bottleneck performance due to the increased CPU demands. In that instance, quadview may not lead to significant performance increases.

In MP however, the performance overhead from the reduction in CPU usage (since the host machine CPU then handles AI) would lead to very significant performance improvements. And in less CPU intensive missions generally for that matter.

I suspect the devs simply tested what the result would be in SP with a not so aggressive setting. If the MP performance improvements in VR is not considered worth the amount of dev time to implement quad view, fair enough.

Edited Saturday at 10:25 AM by MoleUK

[T24_Martin]

23 hours ago, MoleUK said:

Is DLSS even working in VR? It felt like it wasn't initializing at all, as the image was still full of aliasing and I didn't really see any performance bump. This was via OpenXR, maybe I should try OpenVR.

Unfortunately the head movement restrictions feel just as over the top as they did in GB to me. I can't get very close to anything without the head movement restrictions kicking in.

And yes the game scale doesn't feel right. I think you can alter that with OpenXRtoolkit.

Currently the game both looks worse and runs worse than DCS in VR for me on a 3090. This is semi expected given that I rely very heavily on quadviews for DCS performance, but it still feels like something is iffy.

Flatscreen looks and runs great for me by comparison.

Same for me. In DCS it makes a hugh difference when I switch the DLSS settings. In Kore performance is almost the same... 

[MoleUK]

Just now, T24_Martin said:

Same for me. In DCS it makes a hugh difference when I switch the DLSS settings. In Kore performance is almost the same... 

I haven't had time to tinker with settings much so far, but I will try and poke things tomorrow to see if I can figure out what is going on. Using the VR resolution scaler/slider definitely works, as you can see the performance improvement on restart. But like you I didn't notice any real difference when switching DLSS settings. Nor was it reducing aliasing jaggies in the way it should imo. 

I know the current DLSS implementation defaults to the transformer presets which is also a problem (I have a 3090 so that's a no-no as it's only meant for 4000-5000 GPU's), I will try using DLSStweaker tomorrow or the day after and see if I can figure what's going on. Or just try FSR and see if that works off the bat.

[T24_Martin]

Yep, I am pretty disappointed about VR also. Doesn't look good and runs incredibly bad compared to IL-2 GB or DCS. 

It feels as if Monitor was still the focus. I am a VR only player and will not return to 2D Monitors. So hopefully this gets fixed till the full release. Otherwise I will stick to my old fellows GB and DCS.

Such a shame! 

1 minute ago, MoleUK said:

I haven't had time to tinker with settings much so far, but I will try and poke things tomorrow to see if I can figure out what is going on. Using the VR resolution scaler/slider definitely works, as you can see the performance improvement on restart. But like you I didn't notice any real difference when switching DLSS settings. Nor was it reducing aliasing jaggies in the way it should imo. 

I know the current DLSS implementation defaults to the transformer presets which is also a problem (I have a 3090 so that's a no-no as it's only meant for 4000-5000 GPU's), I will try using DLSStweaker tomorrow or the day after and see if I can figure what's going on. Or just try FSR and see if that works off the bat.

I have a 5090 and aliasing doesn't look good out of the cockpit. I will also try different models, but my assumption is, that DLSS is not working correctly on VR.

[MoleUK]

1 minute ago, T24_Martin said:

Yep, I am pretty disappointed about VR also. Doesn't look good and runs incredibly bad compared to IL-2 GB or DCS. 

It feels as if Monitor was still the focus. I am a VR only player and will not return to 2D Monitors. So hopefully this gets fixed till the full release. Otherwise I will stick to my old fellows GB and DCS.

Such a shame! 

I have a 5090 and aliasing doesn't look good out of the cockpit. I will also try different models, but my assumption is, that DLSS is not working correctly on VR.

Yeah my guess is it's just not initialising correctly, at least for some users/setups.

Looks great in flatscreen, but like you I have been spoiled by VR. Will let you know if I find anything.

Edited Saturday at 10:33 AM by MoleUK

[AEthelraedUnraed]

36 minutes ago, MoleUK said:

Because the effect it has it not dependent upon the game engine. It reduces GPU load and increases CPU load, that is the exchange you're making alongside the drop in visual fidelity in the lower resolution rings.

For a SP mission, it has already been stated that in Korea the AI can bottleneck performance due to the increased CPU demands. In that instance, quadview may not lead to significant performance increases.

In MP however, the performance overhead from the reduction in CPU usage (since the host machine CPU then handles AI) would lead to very significant performance improvements. And in less CPU intensive missions generally for that matter.

That is a very simplified view of foveated rendering. There are many more things that determine the FPS gains than just a CPU bottleneck. Foveated rendering helps a lot with speeding up pixel shading, but little with rendering the actual geometry. If Korea's rendering bottleneck is the latter, it's entirely plausible it's just a 10% speedup. Where/how these bottlenecks exist is entirely dependent on the game engine and the game content.

40 minutes ago, MoleUK said:

I suspect the devs simply tested what the result would be in SP with a not so aggressive setting.

That's an assumption. Fact remains that you don't have any data about the gains of foveated rendering in Korea's engine. The Devs apparently found it worthwhile enough to investigate, and they came to this figure of 10%. They're professional programmers, so I assume that they know what they were talking about when they did their investigation. I'm also sure they must have been aware of the gains in other games, yet this apparently did not affect their conclusion. You have less information than they did, and are basing your conclusion entirely on the performance in other game engines.

[Aapje]

3 hours ago, AEthelraedUnraed said:

The Devs apparently found it worthwhile enough to investigate, and they came to this figure of 10%. They're professional programmers, so I assume that they know what they were talking about when they did their investigation.

There are huge differences in expertise between programmers. I suggested before that they hire mbucchia (the maker of OpenXR Toolkit) as a consultant, because he is clearly world class when it comes to VR. No one at 1CGS will even have the opportunity to spend that much time on VR, nor did they ever work at a company making VR hardware, so they don't/didn't have the opportunity to get that much expertise.

I'd say that there is a pretty good chance that he would be able to point out ways to improve VR in the game beyond what 1CGS can come up with, whether that is with a better foveated rendering implementation or other improvements.

[Blooddawn1942]

4 hours ago, AEthelraedUnraed said:

Regarding the reticle, it now has adjustable brightness: https://forum.il2-series.com/topic/3828-new-gunsight-commands/. IMO it's actually a good thing that it's not always visible; it makes sense that you'd need to set it brighter on a bright day than, let's say, at night.

Yes, I'm aware the the reticle brightness is now adjustable. That's a feature I begged to have it in battle of Stalingrad back then but they never implemented it. 

Nevertheless, even with brightness at maximum the reticle is under certain conditions not visible. Especially in planes without sightfilter. 

[Dash,Polder]

8 hours ago, brickcommander said:

I am also however very annoyed with the pilot head limitation. Turning it off doesnt work and every small yead movement- particularly up/down- moves the whole cockpit around me. And the menu screens are too close and it is uncomfortable. I would imagine both issues are easy fixes that will he resolved soon.

 

[TCW_Brzi_Joe]

Why I do not have options for upscaling in Korea? i5-14600kf, ram 32gb, Nvidia 4090: 
PS: I disabled in Pimax software sharpening and upscaling there, to not interfere..
Any hints?

[ROSS_Papa_JA2]

4 минуты назад, TCW_Brzi_Joe сказал:

PS: I disabled in Pimax software sharpening and upscaling there, to not interfere..
Any hints?

Поставьте разрешение экрана не 1280Х960, а выше, например, 1920х1080 и кнопка  Upscaling будет активной, не спрашивайте почему так сделано, я не знаю))

[Dash,Polder]

Yeah what he said, only certain resolutions for your flat screen will allow you to access those settings, just click through starting from the res he told you on up to find the lowest one that'll work for you.  It's weird, but what we currently have to deal with.  

[Aapje]

Seems like a basic bug where they disable upscaling for very low resolutions, which makes sense, but forgot about how that works with VR, where you have separate resolutions for the screen and the headset.

[TCW_Brzi_Joe]

Thx, it works only with native resolution of my screen (full hd); that means with VR I need to have mirror display over my whole monitor, can not peak at discord if I want DLSS. 🫢

[NeilBP]

Yeah, I haven’t flown over any major cities yet. 

[Antonovomatic]

7 hours ago, AEthelraedUnraed said:

Might be related to the headset; so far I haven't seen any overdone brightness on my Quest2.

Regarding the reticle, it now has adjustable brightness: https://forum.il2-series.com/topic/3828-new-gunsight-commands/. IMO it's actually a good thing that it's not always visible; it makes sense that you'd need to set it brighter on a bright day than, let's say, at night.

The Devs did some experiments with it, and the 10% improvement was based on those. Not sure how you came to the conclusion that it would be more, based on the results from an entirely different engine.

You're making the same mistake the author of that answer to the question seems to have, conflating variable rate shading and quad-views

mbuchcia's (dev of the Quadviews implementation for DCS) succinct explanation of how radically different the techniques are, and why their impact of perforamnce is vastly different despite the perceptual similarities in the foveated image projection:
 

Quote

In a simplified way, yes foveated rendering will "render" less pixels in the regions of the screen you are not looking at.

But the key difference here is the definition of "rendering".

With the traditional foveated rendering method, called Variable Rate Shading, your GPU only skips one stage of "rendering" those pixels, aka the pixel shading stage. Granted, this stage is typically one of the most expensive, especially with complex lighting in the scene. But every other stage of "rendering" is still here and it consumes memory bandwidth on the GPU (because even tough the pixel colors are not computed for each pixels, they are still written back to memory).

With the quad views method, your GPU is now drawing each view twice, but one of them (where you are not looking at) is at a much much lower resolution. So the applications ends up "rendering" less pixels, with here the definition of "rendering" is skipping **all** stages, including depth testing (heavy on memory bandwidth), rasterizing (determining triangles to fill and the transformed coordinates of each pixels), the pixel shading stage discussed above as well as writing back pixel colors into memory.

So the end result is the same: yes your "render" less pixels, but the quad views way of "rendering less" is much more effective than the other way.

 

Edited Saturday at 05:22 PM by Antonovomatic

[YoYo]

3 hours ago, Aapje said:

There are huge differences in expertise between programmers. I suggested before that they hire mbucchia (the maker of OpenXR Toolkit) as a consultant, because he is clearly world class when it comes to VR. No one at 1CGS will even have the opportunity to spend that much time on VR, nor did they ever work at a company making VR hardware, so they don't/didn't have the opportunity to get that much expertise.

I'd say that there is a pretty good chance that he would be able to point out ways to improve VR in the game beyond what 1CGS can come up with, whether that is with a better foveated rendering implementation or other improvements.

Hiring mbucchio would be a great idea (just for the duration of the optimization work and possible support later). He'd definitely work his magic and everyone would be happy. But that's just a user's wish...