Main thing for improving jaggies (a bit) is the "HMD Image Quality" setting which is kinda like supersampling in the native headset software. It's fairly GPU intensive but well worth turning it up to around 1.5 or more (maybe even sacrificing something like shadow quality a bit in order to do so while maintaining framerate). There's quite a lot of confusion about how this plays with the actual "Super Sampling" setting. My understanding is that "Super Sampling" is an instruction to the Elite: Dangerous software itself to generate a higher than required resolution image (which will then get downscaled again later, smoothing jaggies in the process) while the "HMD Image Quality" is an instruction to the headset's runtime software to do a similar thing. Most people recommend leaving "Super Sampling" at 1.0 and bumping up "HMD Image Quality". There are others however who reduce "Super Sampling" to something like 0.75, freeing more GPU for turning "HMD Image Quality" up even more (e.g. to 2.0). Personally I don't entirely get it (sounds a bit like a case of garbage in garbage out) but many swear by it. The only other thing is the "Anti Aliasing" stuff but it's notoriously ineffective in Elite, especially in VR, so personally I'd recommend turning it off and using the saved GPU cyles for more quality.
Some graphics settings questions for new VR guy
The whole game has been VR ready for several years now and it has a big VR following and the forum is here https://forum.il2sturmovik.com/forum/109-virtual-reality-and-vr-controllers/
@Old Duck; I'm pretty blown away by it to be honest, pleasantly surprised especially since the Rift S has an LCD display and I was worried the black of space wouldn't be, well 'black' enough - but it is absolutely fine.
The experience is only marred by the apparent lack if anti-aliasing (or my inability to find where to turn it on!).
@Alec Turner
Many thanks for the reply, so the HMD and SS are in the setting menu on the Oculus home screen?
That would explain why I'm not seeing that in the ED settings menu.
Thanks for clearing that up, I'll try that when I get home.
(I've not played ED for maybe 6 months, have they removed advanced GFX settings from the in game menu now? I only see LOW/MED/HIGH/ULTRA etc.?)
Thanks again
Expand the line where you choose your preset settings profile, and all the settings that are encompassed by the profile will appear. (EDIT: As well as holding the preset selector, it is header to a nested menu)
As Alec said, you'll want to use "HMD Quality".
The difference between "HMD Quality", and "Supersampling" is that while both render e.g. (x2.0) a bitmap that is twice as high and wide as originally requested, with "Supersampling" the game will then on its own scale the result back down to x1.0, before it is output, whereas with "HMD Quality" the full rendered frame is passed on to the VR runtime, which will then have that full resolution to work with, when it distorts the image, to compensate for the opposite distortion of the lens in the headset, rendering it to its native resolution.
Only reason to use the game's own "supersampling" option, is when you have already maxed out "HMD Quality" (or increased the render target size equivalently by external means), and still want to increase it further.
Keep in mind that the work load goes up by the square of the amount of supersampling, so x2.0 is four times as many pixels to render.
Last edited:
Based on my recent experience, I'm going to go against what everyone is saying here ( and my apologies in advance ). Previously, tinkered with HMD and SS - they do indeed have a very positive quality effect but can, as indicated, be murder on frame rate. So, I'm going to tout what has proven to be, for my 4690K with an old 1080Ti and a Reverb, a far more palatable approach to having much higher frame rates and very good image quality. This could be unique to my rig so use as you deem appropriate:
HMD x1.0, SS x1.0, AA SMAA. All other Low/Med/High/Settings left on High
SteamVR => Video Settings
Set the slider to 400% ( On this one I feel diminishing returns but it's a perception thing ).
So, on my rig, the above settings are every bit as good as HMD x2.0 and SS x2.0 but without the accompanying slide show. I sat outside a large ring station and flew right up to the ring and got ~143fps +/-4fps. I still swear that either Steam or MS did something in a recent update. The difference is amazing. I've just been playing SkyrimVR on these settings and it's the same as using my monitor. No matter how hard I try, I cannot find SDE either. It really has been a game changer. No idea what's been done though, maybe the magic SS fairy paid me a visit although that sounds more like a euphemism for getting a visit from some other folks back in '39.
Those SteamVR sliders used to do pretty much nothing for me but in the last month or so, they have been magic.
HMD x1.0, SS x1.0, AA SMAA. All other Low/Med/High/Settings left on High
SteamVR => Video Settings
- Check Custom Resolution
- Move the slider to 400%
Set the slider to 400% ( On this one I feel diminishing returns but it's a perception thing ).
So, on my rig, the above settings are every bit as good as HMD x2.0 and SS x2.0 but without the accompanying slide show. I sat outside a large ring station and flew right up to the ring and got ~143fps +/-4fps. I still swear that either Steam or MS did something in a recent update. The difference is amazing. I've just been playing SkyrimVR on these settings and it's the same as using my monitor. No matter how hard I try, I cannot find SDE either. It really has been a game changer. No idea what's been done though, maybe the magic SS fairy paid me a visit although that sounds more like a euphemism for getting a visit from some other folks back in '39.
Those SteamVR sliders used to do pretty much nothing for me but in the last month or so, they have been magic.
400% SteamVR ss is essentially exactly the same thing as setting ED HMD Quality to x2.0.
Exactly what WindowsMR does, neither in general, nor in relation to SteamVR, though, I do not know: I do not have a WMR headset, but I know there has been a lot of confusion recently, many back and forths about what effect various variables the two frameworks have between them, whether any of those are relevant or deprecated, and complaints about poor image quality with settings that should have produced better.
Since you get high frame rates; Could it be that you are getting extrapolated frames, but find those acceptable to the point you no longer notice their nature? :7
Exactly what WindowsMR does, neither in general, nor in relation to SteamVR, though, I do not know: I do not have a WMR headset, but I know there has been a lot of confusion recently, many back and forths about what effect various variables the two frameworks have between them, whether any of those are relevant or deprecated, and complaints about poor image quality with settings that should have produced better.
Since you get high frame rates; Could it be that you are getting extrapolated frames, but find those acceptable to the point you no longer notice their nature? :7
This is not quite correct. HMD x2.0 causes a significant performance hit. I can set SteamVR SS to 500% and it is very little FPS impact. No it's not extrapolated frames.
While I generally agree that it is preferrable to set one's supersampling at the joint start-and-end point (...which in your case would actually be the WMR framework, rather than SteamVR), this really suggests to me that you are not actually getting the 400%, or the 500% either, but in fact a lower render target size.
As it happens, SteamVR has an upper Render target size limit, determined by the key: "maxRecommendedResolution" in your SteamVR settings files, and this defaults to 4096. If you have not increased this on your own, SteamVR's render target will be capped right there - as the width or height reaches 4096 pixels - whichever gets to that point first, even if the numbers under the supersampling slider in SteamVR settings say otherwise; they are not capped, but the resolution that SteamVR requests of the game is.
Elite's HMD Quality, and Supersampling multipliers are applied after this point, on top of your SteamVR supersampling, and are as such unrestricted by that limit, which could explain your performance differences, but that should also reflect in the quality of the output.
Last edited:
I'm not really interested in the actual numbers. Regardless of what you think the render target size is, it's working nicely.
Your description of the render order is correct - well done. And no, the output quality is no longer better via HMD SS and/or screen SS. That was the big surprise to me, hence I thought I would share.
I wonder if the improved visual fidelity is related to the higher resolution and matrix of the Reverb. While there seems to have been an update applied somewhere, is it just Reverb's that are seeing the benefit or have I just lucked into a magic config - in which case that will save me money upgrading
Like I mentioned: I have heard talk about Reverb users not getting the sorts of results they should rightly be able to expect, recently, and have seen through-the-lens macro screenshots, where the person taking them claimed to have used settings that should provide excellent imagery (...and does on their pictures from other headsets), but where you can clearly see the same pixel repeating in 2x2 pixel blocks, as if upscaled from half resolution, using a low-fi resampling method.
That high base resolution of the Reverb should make better use even of lower resolution input images, but apparently people have just not been seeing it, even when greatly supersampling.
So I could imagine maybe there has in the past been something wrong somewhere along the way, from the Reverb WMR drivers, to the WMR framework, through the WMR SteamVR driver, to SteamVR, and the way each of those speak with one another - perhaps throwing away stuff, or resampling it, at some point, so that you have had worse performance than you should before, but are up to nominal now, (EDIT: ...after an update to one of those components). :7
I'd like to know actual technical reasons, though. When people say they not only get something for nothing, but absolutely staggering results, outshining anything anybody else has experienced by orders of magnitude, I tend to assume there is a placebo effect in action...
(EDIT2: I just noticed in your first post, you said not only to supersample at 400% SteamVR general, but do it the same with the application-specific setting, which is on top of the main one. That would be 16 times more work than no supersampling at all, and would definitely be capped - especially with the Reverb, that should start out with a really high base render target (something like native resolution times 1.4, if the default is chosen by similar logic to the Vive).)
That high base resolution of the Reverb should make better use even of lower resolution input images, but apparently people have just not been seeing it, even when greatly supersampling.
So I could imagine maybe there has in the past been something wrong somewhere along the way, from the Reverb WMR drivers, to the WMR framework, through the WMR SteamVR driver, to SteamVR, and the way each of those speak with one another - perhaps throwing away stuff, or resampling it, at some point, so that you have had worse performance than you should before, but are up to nominal now, (EDIT: ...after an update to one of those components). :7
I'd like to know actual technical reasons, though. When people say they not only get something for nothing, but absolutely staggering results, outshining anything anybody else has experienced by orders of magnitude, I tend to assume there is a placebo effect in action...
(EDIT2: I just noticed in your first post, you said not only to supersample at 400% SteamVR general, but do it the same with the application-specific setting, which is on top of the main one. That would be 16 times more work than no supersampling at all, and would definitely be capped - especially with the Reverb, that should start out with a really high base render target (something like native resolution times 1.4, if the default is chosen by similar logic to the Vive).
)
Last edited:
Are you playing the Beta?I wonder if the improved visual fidelity is related to the higher resolution and matrix of the Reverb. While there seems to have been an update applied somewhere, is it just Reverb's that are seeing the benefit or have I just lucked into a magic config - in which case that will save me money upgrading
Lol - yup it's clear you don't want to believe. I guess you took the blue pillLike I mentioned: I have heard talk about Reverb users not getting the sorts of results they should rightly be able to expect, recently, and have seen through-the-lens macro screenshots, where the person taking them claimed to have used settings that should provide excellent imagery (...and does on their pictures from other headsets), but where you can clearly see the same pixel repeating in 2x2 pixel blocks, as if upscaled from half resolution, using a low-fi resampling method.
That high base resolution of the Reverb should make better use even of lower resolution input images, but apparently people have just not been seeing it, even when greatly supersampling.
So I could imagine maybe there has in the past been something wrong somewhere along the way, from the Reverb WMR drivers, to the WMR framework, through the WMR SteamVR driver, to SteamVR, and the way each of those speak with one another - perhaps throwing away stuff, or resampling it, at some point, so that you have had worse performance than you should before, but are up to nominal now, (EDIT: ...after an update to one of those components). :7
I'd like to know actual technical reasons, though. When people say they not only get something for nothing, but absolutely staggering results, outshining anything anybody else has experienced by orders of magnitude, I tend to assume there is a placebo effect in action...
(EDIT2: I just noticed in your first post, you said not only to supersample at 400% SteamVR general, but do it the same with the application-specific setting, which is on top of the main one. That would be 16 times more work than no supersampling at all, and would definitely be capped - especially with the Reverb, that should start out with a really high base render target (something like native resolution times 1.4, if the default is chosen by similar logic to the Vive).
No - unless they've added fleets a'la X or planetary cities - it's not got anything to grab me.
Lol - yup it's clear you don't want to believe. I guess you took the blue pill
As I recall, the blue pill was the one that left you in belief land, so no.
Get a Reverb and try it yourself. Easy to knock from a position of no experienceAs I recall, the blue pill was the one that left you in belief land, so no.
Easy to knock from a position of logic and mathematics.
There is reasoning, objectivity, and facts in everything I have written above (EDIT: I am also genuinely trying to be helpful). I am not going to purchase any certain HMD, just to test an unsubstantiated fantastical claim from some-guy-on-the-internet(TM). You can easily convince me, if you can explain how your results come about; I'd be deeply grateful of such knowledge, that could benefit us all - even non-Reverb owners.
At least with those who swear by their "magic" combinations of under- and supersampling with "supersampling" and "HMD Quality", I can appreciate that they may subjectively find a blurry image preferable to one with more obvious aliasing (not minding the "stair steps" that are in fact still there, only smudged into fuzzy blobs instead), even though I don't; And I can appreciate that some people really do not notice when frame synthesis kicks in to prop up their lagging frame rates. They usually do not claim a 16 times performance boost out of thin air.
Last edited:
Nope, you're still just ranting from no experience