Increase pixel density - VIVE - improve text readability

You shoukd get better results with 1.0 resolution in game and 1.4 steam vr multiplier, and turn off smaa, dont use AA in game, you are overtaxing your system, supersampling is the original and best form of AA.
Not using SMAA looks like crap. As for overtaxing my card, 63C isn't hot in the slightest and my GPU utilization is 55% so no, I'm fine.

Last night I tried SS 1.0 with 1.4 Steam multiplier, it actually looked worse and had poorer frame rate.

To my eye, best visual and framerate results are by undersampling at 0.65 with a higher Steam multiplier of 1.8 - 2.2.

It's counter intuitive, I know! Elite's SS implementation for the Vive must be really bad.
Agreed. Pre-processing the SS before it hits Elite seems to work wonders. When I get my 1180 in Q4 I'll see how high I can get the vrsettings to go. :)

The only remaining issue is the screendoor effect which I suspect will only be solved with higher resolutions unless Frontier can pull a rabbit out of the proverbial hat.
 
...
The only remaining issue is the screendoor effect which I suspect will only be solved with higher resolutions unless Frontier can pull a rabbit out of the proverbial hat.

I can share a little anecdotal information here, I believe.

Some idiot had to try breaking his very expensive toy apart, to add diffusers, and since nobody else was doing it, I decided to pick up the idiot mantle...

So... I tried with a pair of mobile phone screen protectors of "Tech Armor" brand, applied directly to the display panels, and these were my results:

As expected; The granularity was really bad -- With this type of anti-glare screen protector, even an unaided eye, when the sheet is applied to a phone screen, can see the fixed grain pattern, along with the chromatic abberrations in it (think shimmery rainbows), and its average desaturation effect -- Not surprisingly, this made for an undesired "anchoring" sensation inside the HMD - much like the "looking-through-black-silk"-y pattern, that the displays themselves exhibit in low-light scenes. Having the unmoving pattern "superimposed", makes it look like you are looking at an animated image (EDIT: ...which you are, of course :7), instead of out into open space. There was also a lot of contrast to grain edges.

As for the sought softening of the image; it turned out very well balanced, actually; Diminishing the sharp black "troughs" between subpixels, without bleeding over so much that the image becomes blurry -- a very pleasing amount; I might have liked a smidgen more, but it looked good.

Now; This did had a rather interesting negative effect: Where usually the screen door effect turns the screen into an eye-straining busy anthill of pinprick elements, when looking around, I found the diffused image could be accused of having become almost too stable. The eye has a hard time keeping track of those pesky ants, that normally scurry every which way across the view, whereas the nicely solid diffused pixels became like a superbly resolveable tiled wall - very easy to lock onto and follow with your gaze, and thus we are again back to a fixed pattern that moves along with your head, as you look around.

At the end of the day; Even though I managed to drop and physically damage one of the display panels when taking the screen protectors out again, and am now waiting for a quote on how much it is going to cost me to have it replaced; If I could come across a better diffusing material, that does not not have the grain, I would definitely try it again - I think the SDE is that bad.

:7
 
Last edited:
Hi,

I gave it a go and at first all went wrong, I forgot the comma in the sentence above...:rolleyes:

For me (i7, GTX 1080) there was noting much to complain about except the bad AA.
I had my graphics on the preset ultra, only with SS changed to 1.5

I tried all different settings but to get rid of judder I had to turn off AA first

(ED - Config file)
0,75 - 1.5 not much difference
1.0 - 2.5 is beautiful but a little over the top
0.75 - 2.0 is were I ended up, I only have judder in the galaxy map but I will try a lower setting there.

I use Virtual Desktop and that also seems to be better now.

Now waiting for someone to come with a clever idea for the AA :D

Groet Dree

PS Why do my stars look like the Hubble deep field?
 
Hi,

I gave it a go and at first all went wrong, I forgot the comma in the sentence above...:rolleyes:

For me (i7, GTX 1080) there was noting much to complain about except the bad AA.
I had my graphics on the preset ultra, only with SS changed to 1.5

I tried all different settings but to get rid of judder I had to turn off AA first

(ED - Config file)
0,75 - 1.5 not much difference
1.0 - 2.5 is beautiful but a little over the top
0.75 - 2.0 is were I ended up, I only have judder in the galaxy map but I will try a lower setting there.

I use Virtual Desktop and that also seems to be better now.

Now waiting for someone to come with a clever idea for the AA :D

Groet Dree

PS Why do my stars look like the Hubble deep field?

Dree, use the 'Chaperone Switcher' utility - it's a simple click and it's done utility for setting the supersampling to what you want without having to edit the file manually.
 
i'm indeed very interested in the 'Chaperone Switcher' utility but for switching rooms.
i don't think I will change that config file very much and if I will it's very easy now.

Thanks, Dree
 
For me (i7, GTX 1080) there was noting much to complain about except the bad AA.
I had my graphics on the preset ultra, only with SS changed to 1.5
Hi Dree,

Try high settings with 2.0 SS in the vrsettings and .65 SS in Elite settings with SMAA. That's what I am running and there is no judder for me on a 980Ti. You should have no issues with a 1080. If that works for you then kick it up to Ultra and see if there's any impact. Your card should be able to take that. Cheers.
 
Hi Dree,

Try high settings with 2.0 SS in the vrsettings and .65 SS in Elite settings with SMAA. That's what I am running and there is no judder for me on a 980Ti. You should have no issues with a 1080. If that works for you then kick it up to Ultra and see if there's any impact. Your card should be able to take that. Cheers.

Sorry, but I really struggle to believe you can do this with the 980Ti. Looking at the Frame Time graphs from steam VRs 'performace' tab, my 1070 FE, overclocked, runs 1.5 SS in vrsettings, .65 in Elite settings, with VRHigh and no AA at a consistent 90FPS, but adding SMAA to this makes the performance begin to drop frames. Running at 2.0 SS in vrsettings drops the Frame rate to 45FPS consistently regardless of the other settings. So sorry, I just don't buy that a 980Ti will run the settings you suggest at a consistent 90FPS.

I'll be posting some graphs to show my testing results with the steam vrsettings SS this evening that quantifies the performance and I'd love to see some Frame Time graphs from people with 980Ti's to see how they compare with my 1070 ones (this thread: https://forums.frontier.co.uk/showthread.php/266026-Performance-Review-GTX1070-FE-Vive). The long and short of it is that it does indeed perform better, and look better, than using the SS in the E: D settings.
 
Last edited:
Sorry, but I really struggle to believe you can do this with the 980Ti. Looking at the Frame Time graphs from steam VRs 'performace' tab, my 1070 FE, overclocked, runs 1.5 SS in vrsettings, .65 in Elite settings, with VRHigh and no AA at a consistent 90FPS, but adding SMAA to this makes the performance begin to drop frames. Running at 2.0 SS in vrsettings drops the Frame rate to 45FPS consistently regardless of the other settings. So sorry, I just don't buy that a 980Ti will run the settings you suggest at a consistent 90FPS.

I'll be posting some graphs to show my testing results with the steam vrsettings SS this evening that quantifies the performance and I'd love to see some Frame Time graphs from people with 980Ti's to see how they compare with my 1070 ones (this thread: https://forums.frontier.co.uk/showthread.php/266026-Performance-Review-GTX1070-FE-Vive). The long and short of it is that it does indeed perform better, and look better, than using the SS in the E: D settings.

I am running ED on a MSI 980Ti Gaming card with an i7 4770K overclocked to 4.1Ghz, and yes I agree with you totally that the reported success stories seem to take the 45fps with reprojection as a smooth running experience.

If I run with 2.0 and 0.65 I will most of the time run into the 45fps with reprojection. Currently I am using 1.6 and 0.65, which in total is 1.04 with high settings and this gives me almost always 90fps. Although the net resolution is only slightly higher, the visual clearity is much better than with running 1.0 and 1.0.

I will post my graphs as well this evening if time permits.
 
Hi Dree,

Try high settings with 2.0 SS in the vrsettings and .65 SS in Elite settings with SMAA. That's what I am running and there is no judder for me on a 980Ti. You should have no issues with a 1080. If that works for you then kick it up to Ultra and see if there's any impact. Your card should be able to take that. Cheers.

I'm on 0.65 - 2.0 and preset ultra already, only AA is off but in comparison the AA got better with this config mod, I'm happy.

Dree

Edit, I'm playing this for 1.5 years or so and I never saw any difference in any AA setting or Off, if you ask me it's a dummy .
 
Last edited:
I've added some Frame Time graphs to https://forums.frontier.co.uk/showthread.php/266026-Performance-Review-GTX1070-FE-Vive/page2 quantifying the performance hit for setting vrsettings to 2.0 or 1.5 with ingame SSx0.65 with the 1070 FE - these can be directly compared with the original graphs I posted looking at the performance behaviour of other in-game settings. Sadly improvements in apparent graphical quality are subjective, so while the performance can be quantified, the impact ont he image quality is harder to show...
 
Well, I'm not going to argue the point. All I know is that at my quoted settings, it plays smoothly, I have no headaches or motion sickness after almost three hours of E: D and it looks frikkin' fantastic. I can see nicks, dings and scratches in my DBS canopy window supports. Planets are immense and wondrous to behold. Outposts, stations and other ships are amazingly detailed. So if this is only 45 FPS then frankly, I don't care. It's working just fine for me and I don't need a graph to validate it.

Have a great night all.
 
I agree..
I sometimes have some hiccups in the mission board and if I quickly move my head in the galaxy map I see the names of the stars double, I don't care.
For the rest everything is smooth.

Groet Dree
0.65 - 2.0
 
I am running ED on a MSI 980Ti Gaming card with an i7 4770K overclocked to 4.1Ghz, and yes I agree with you totally that the reported success stories seem to take the 45fps with reprojection as a smooth running experience.

If I run with 2.0 and 0.65 I will most of the time run into the 45fps with reprojection. Currently I am using 1.6 and 0.65, which in total is 1.04 with high settings and this gives me almost always 90fps. Although the net resolution is only slightly higher, the visual clearity is much better than with running 1.0 and 1.0.

I will post my graphs as well this evening if time permits.

This exactly mirrors my experience, which makes sense becasue we know the 1070 should have similar (or marginally better) performance to a 980Ti.
 
I can verify that 2.0-0.65 works in 90 fps (excluding hyperjumps and menu fiddling). I have been running the steamvr debugger to report missing frames so I know that I get constant 90 fps. But I do not run ultra settings. I "just" have vr high with 0.65 but that still gives awesome gfx. I have not really had the time to optimise the settings, I'm just very satisfied with what I get.
 
Well, I'm not going to argue the point. All I know is that at my quoted settings, it plays smoothly, I have no headaches or motion sickness after almost three hours of E: D and it looks frikkin' fantastic. I can see nicks, dings and scratches in my DBS canopy window supports. Planets are immense and wondrous to behold. Outposts, stations and other ships are amazingly detailed. So if this is only 45 FPS then frankly, I don't care. It's working just fine for me and I don't need a graph to validate it.

Have a great night all.

I get exactly the same Shadragon, OC 980ti with the same settings and it looks great.
 
I can verify that 2.0-0.65 works in 90 fps (excluding hyperjumps and menu fiddling). I have been running the steamvr debugger to report missing frames so I know that I get constant 90 fps.

If you have no missing frames reported it does not mean that you are running at 90fps. It just means that the system was able to display all frames in the current situation, that means also 45fps reprojection.

For a lot of people, like shadragon, the 45fps is a smooth experience and that is fine. Some people like me who are more prone to vr sickness do notice it though and for those it is better to stick to the 90fps.

And on a purely technically note, the 45fps reprojection is intended as a last resort in temporariliy demanding scenes and (quote from the Valve developers) must never be used as tool to cover bad application performance. meaning if the application developers rely on this behaviour, there is no room left for a fallback, which will lead to dropped frames.
 
Last edited:
I have tried to max settings for a short while when in an engineering base and lost every other frame (45 fps) which showed up clearly in the developer graphics. And that was not a pleasant experience. But please enlighten me of how to detect frame loss if the dev tools is not showing the correct information. Also, is there anyone with a 980ti to who is not able to run 2.0-0.65 smoothly (experience wise)?
 
I may have expressed myself a bit unclear. Whenever the dev tools report frame drops they are real frame drops. This information is always valid. You will have to use the timing graphs to determine if the system has switched to 45fps with reprojection. The tools do not consider a switch to 45fps a lost frame because it was able to catch it by doing the reprojection.
My point was, that from the strict point of view of having 90fps, running in 45fps with reprojection is essentially dropping every other frame.
You can easily disable reprojection in the settings and then you will see for real how many frames you are really dropping.
 
This is not how I interpret what valve developer Aaron Leiby describes regarding the developer tool frame timing graph:

"If you bring up SteamVR's frame timing graph, you will see a red line on the stacked gpu view when interleaved reprojection is active and can get a feel for how it gets turned on/off. There is also a checkbox to display the frame timing in-headset (attached to the back of your right controller). Displaying the frame timing in headset incurs a little bit of overhead (you'll see the green gpu band get a bit thicker) so keep that in mind when using it to track down issues."

So I should be able to see when this happens even if I do not turn reprojection off.
 
Back
Top Bottom