Oculus Rift and DCS World Discussion

[dot]

Do you think it's safe to assume CV1 will get a bit more usable pixels as well due to the separate screens? I feel like the lenses were limited due to the single screen with DK1&2. I hope this also solves IPD issues.

 

It will definitely be an improvement, but I fear it won't be a big one.

 

If you have a chance, have a look at Gear VR, it has about 42% higher resolution than CV1

and about 77% increase in resolution compared to DK2

 

Here is comparison between 1080p and 1440p mobile screen:

http://www.androidpit.com/full-hd-vs-quad-hd-vs-4k-phone-screen-resolutions

 

Even 1440p screen is not that great up close, so going for a much lower resolution is (imho) sad.

[SkateZilla]

The size dimensions of the screens been stated yet?

 

DK1:

215.6 PPI, 7 Inch LCD, 1280x800 @ 60Hz, 640x800 Per Eye.

 

DK2:

400.5 PPI, 5.5 Inch OLED, 1920x1080 @75Hz, 960x1080 Per Eye.

 

CV1:

2160x1200 Combined, Dual Screens @ 90Hz, 1080x1200 Per Eye

 

So, assuming they dont go above 5.5 total (ie no screen surface on the area of the orbital/nose bridge)

 

At 3 Inches Per it would be 538.1 PPI

At 2.75 Per it would be 587 PPI

At 2.5 Per it would be 645.7 PPI

 

One 5.5 Screen would be 449 PPI.

 

Then you have to look at added.clarity by dropping the pentile SubPX Arrangement.

[dot]

I think that screen dimensions are not that important.

Smaller screens go closer and larger further away from the lenses.

What matters is the FOV you can see in the headset and the screen resolution.

[AurelTristen]

I understand the fixation on resolution, but the only thing we REALLY need fine detail for is spotting enemy aircraft. As discussed in other threads, due to DCS's lack of scaling, it is already almost impossible to spot aircraft at realistic distances, even on very large monitors.

 

Whatever scaling solution that ends up fixing this will probably help out with VR. Besides, as it stands you could put as many additional pixels as you want in the Rift, but only the guy with triple SLI Titan X's is going to be able to push those pixels at the requisite refresh rate.

 

Luckily, most of our time in DCS is spent using the instruments to start up, take off, navigate and land. Other than gun-runs (which work out just fine on my DK2), most of the fighting in the A-10 involves looking at MFCDs. All of these things are almost usable in the DK2, so CV1 will probably be reasonably well equipped for simming.

[dot]

I think I'll buy either Vive or Oculus (Vive seems better atm), but they could have done much, much better in terms of screen specs. (apparently Oculus sold themselves to FB so they can get a custom screen)

[Chivas]

Oculus suggests that the CV1 display has a high fill rate, meaning there is a much smaller space between pixels. That should help reduce screen door effect, but I'm not sure how it effects the long distance view, and the ability to spot dots. I wonder since the pixels appear larger when our eyes are so close to the display, a single pixel dot, representing a distant aircraft could be more visible while in VR???

[dot]

This might work if ED guys tweak their engine to render distant aircraft much bigger (so they appear within 1 pixel on the screen) as most game engines go, if something is too far (occupies less than 1 pixel) then most likely it won't be visible on the screen.

 

The way to reduce this problem is to supersample, as explained in this article: http://www.geforce.co.uk/whats-new/articles/dynamic-super-resolution-instantly-improves-your-games-with-4k-quality-graphics

 

However, the higher the screen resolution, the less of a problem this is (atm it's a huge problem).

[Chivas]

I think I'll buy either Vive or Oculus (Vive seems better atm), but they could have done much, much better in terms of screen specs. (apparently Oculus sold themselves to FB so they can get a custom screen)

 

There are definitely displays with higher resolutions, but I'm not so sure there are higher resolution displays with the necessary qualities for VR like, high fill rate, 90Hz, and the Global something other that seems very important. Not to mention the fact a higher resolution screen might require a much higher recommended computer system. One of the reasons that Vive and Oculus appear to have settled on very similar displays atleast as far as resolution goes.

 

Facebook monies could allow Oculus have a custom resolution display made, and equally important is their partnership with Samsung, that can actually get it done, at a reasonable price point. No word yet if the display Oculus is using has been specially made by Samsung with a higher fill rate, etc etc. Not so sure that HTC makes displays, but they probably have a contract with a display manufacturer that can deliver a display at a decent pricepoint.

[Chivas]

This might work if ED guys tweak their engine to render distant aircraft much bigger (so they appear within 1 pixel on the screen) as most game engines go, if something is too far (occupies less than 1 pixel) then most likely it won't be visible on the screen.

 

The way to reduce this problem is to supersample, as explained in this article: http://www.geforce.co.uk/whats-new/articles/dynamic-super-resolution-instantly-improves-your-games-with-4k-quality-graphics

 

However, the higher the screen resolution, the less of a problem this is (atm it's a huge problem).

 

I have no idea, but can something be rendered smaller than a pixel? Aren't distant aircraft rendered as small as one pixel or more as it gets closer. If that's true, it stands to reason that the distant aircraft would appear bigger than on a monitor at the same resolution since a pixel appears larger in VR.

[dot]

I have no idea, but can something be rendered smaller than a pixel? Aren't distant aircraft rendered as small as one pixel or more as it gets closer. If that's true, it stands to reason that the distant aircraft would appear bigger than on a monitor at the same resolution since a pixel appears larger in VR.

 

No, have a look at the nvidia article, if there is nothing in the sampled pixel, usually nothing is drawn.

 

This is the same situation supersampled:

 

You can imagine that the lower the resolution the bigger problem this is.

[hansangb]

F19...loved that game. And Gunship! Thank god for Belsimtek's Huey. what a great sim that is.

 

hsb

[SkateZilla]

These headsets were developed to be generally affordable.

 

Add A 4K screen and required chips you're looking at a substantial increase in cost.

 

Then another increase in system requirements, 4K rendered one two separate.viewports at 90FPS would take an insane ammount of GPU Power and Ram, likely 2 years away (HBM2 GPUs with upwards of 16GB each).

 

Not to mention there currently is no mobile/sub 10'' 4K screens capable of a 90Hz refresh rate in production. Nor a MIPI chip capable of 90Hz at 4K,

 

Even.the SG Note 4 @ 1440P, only runs a MAX of 60FPS, any higher would require.a more expensive MIPI chip, which isnt in mass production at all.

[blksolo]

I have a performance question for you guys. I have a DK2 with a GTX 760. I'm thinking about maybe snagging a used 970 or 980 on Ebay. For those of you with a 970 or 980 what is the performance like with DK2 and DCS,do you get an acceptable frame rate with max settings? My system specs are i7-2600, GTX 760, 16 gig RAM, SSD drive, Win 7.

Edited May 22, 2015 by blksolo

[Markeebo]

I have a performance question for you guys. I have a DK2 with a GTX 760. I'm thinking about maybe snagging a used 970 or 980 on Ebay. For those of you with a 970 or 980 what is the performance like with DK2 and DCS,do you get an acceptable frame rate with max settings? My system specs are i7-2600, GTX 760, 16 gig RAM, SSD drive, Win 7.

 

 

I recently moved from a 690 to a 980. Really didn't notice any improvement. CPU is the bottleneck. My experiences are not scientific. Simply subjectively observed.

 

From the few limited other apps that I have seen, ED really didn't put a ton of work into implementing the DK2 into its current engine. If I recall, at first, they simply refused to consider Rift support. I'm glad they changed their minds.

 

However, I think that they did the minimum required to get the code to accept rift support, opting rather to roll resources into EDGE and improve the experience there.

 

Better performance is just around the corner. 2-3 weeks.

[dot]

Are there any words on recommended PC specs required to get 90FPS on Oculus CV1 and HTC Vive?

 

Will Oculus recommended specs do?

 

NVIDIA GTX 970 / AMD 290 equivalent or greater

Intel i5-4590 equivalent or greater

8GB+ RAM

[PiedDroit]

Some info about the minimum specs (I think this has been posted already, along with other links):

 

• NVIDIA GTX 970 / AMD 290 equivalent or greater
  
• Intel i5-4590 equivalent or greater
  
• 8GB+ RAM
  
• Compatible HDMI 1.3 video output
  
• 2x USB 3.0 ports
  
• Windows 7 SP1 or newer
  

https://www.oculus.com/blog/the-rifts-recommended-spec-pc-sdk-0-6-released-and-mobile-vr-jam-voting/

http://www.gizmag.com/oculus-rift-specs/37536/

 

If you plan to do your upgrade now you should consider waiting as the price of hardware will drop.

Edited May 22, 2015 by PiedDroit
added link to www.oculus.com

[dot]

Sorry, I should have said; what are the recommended specs to get DCS/ DCS2 to run at 90fps in CV1 or Vive?

 

I agree, there is no point buying new PC for VR before Vive or CV1 come out (unless you currently have macbook air and want to play with DK2).

 

Also next get gpus seem to have much better memory bandwidth (virtual triangles will love this :)

 

http://blogs.nvidia.com/blog/2015/03/17/pascal/

http://www.legitreviews.com/amd-radeon-r9-390x-video-card-specifications-performance-numbers-leak_160191

[SkateZilla]

Sorry, I should have said; what are the recommended specs to get DCS/ DCS2 to run at 90fps in CV1 or Vive?

 

I agree, there is no point buying new PC for VR before Vive or CV1 come out (unless you currently have macbook air and want to play with DK2).

 

Also next get gpus seem to have much better memory bandwidth (virtual triangles will love this :)

 

http://blogs.nvidia.com/blog/2015/03/17/pascal/

http://www.legitreviews.com/amd-radeon-r9-390x-video-card-specifications-performance-numbers-leak_160191

 

There are no recommended specs for DCS/DCS2 at 90 FPS.

[PiedDroit]

Is there an extra CPU load for the second point of view?

I mean, since the positions of the objects are already calculated, my guess is that the only thing left would be the GPU (and bandwdth) necessary to render the extra point of view.

Maybe an Oculus DKx has some feedback about this?

 

I tried search for info about it but I wasn't very lucky...

Someone posted a similar request a while ago on Oculus forum but without a lot of answers:

https://forums.oculus.com/viewtopic.php?t=5568

[SkateZilla]

The Viewports are rendered separately.

 

One for Left and One for Right.

 

So everything is rendered twice.

 

Everything, Draw Calls, textures, is doubled. (essentially).

 

You can run DCS with side by side left/right viewports without OR

[cichlidfan]

The Viewports are rendered separately.

 

One for Left and One for Right.

 

So everything is rendered twice.

 

Everything, Draw Calls, textures, is doubled. (essentially).

 

You can run DCS with side by side left/right viewports without OR

 

There is a nice three camera lua in the default file set, if someone wants to see what three viewports will do to there frame rates. :D

[PiedDroit]

There is a nice three camera lua in the default file set, if someone wants to see what three viewports will do to there frame rates. :D

This mode renders three different scenes, it's a bit different from rendering the same scene twice.

[Rangi]

This mode renders three different scenes, it's a bit different from rendering the same scene twice.

 

 

It's not the same scene twice it's a slightly different scene twice.....

Edited May 23, 2015 by Rangi
spelling

[PiedDroit]

It's not the same scene twice it's a slightly different scene twice.....

If you like, what I mean is that the CPU don't have to calculate extra objects compared to rendering completely different views with new objects to display.

 

Also, I tested (using a replay) with the "stereo.lua" that renders the two viewpoints, I got a slight FPS hit (90 to 60) and CPU did not seem to change, but the scene I used wasn't too object-heavy either.

Edited May 23, 2015 by PiedDroit

[cichlidfan]

Also, I tested (using a replay) with the "stereo.lua" that renders the two viewpoints, I got a slight FPS hit (90 to 60) and CPU did not seem to change, but the scene I used wasn't too object-heavy either.

 

You consider a 33% loss to be a 'slight' FPS hit? :helpsmilie: