PD / Supersampling - VR vs DCS

[Adam]

Hi,

 

I was just wondering what the final concensus is when it comes to Supersampling (or as DCS calls it Pixel Density)?

 

I've read some posts that say leave DCS as 1.0 and do it all in steam because steam uses less resources. I've read other posts that have Steam reduced from the default (150%) down to 100% and PD increased in DCS.

 

As a result - I'm left not knowing which is the better method, or the pro's and con's of either and was just wondering if someone is able to expand on the single issue of PD/SS as to what the benefits or cons are of each?

 

Thanks

 

Adam

[speed-of-heat]

Ss scale is not the same as as PD, functionally they both largely do the same thing, pd is a blunter instrument than SS, and a pd of 1.4 is the equivalent of 196% ss, both in terms of performance impact and display quality.

 

I currently use neither, as they have a crushing impact on performance at that level on my hardware.

 

PD is perceived as having a higher impact on resources becuase it is a lower number, and people try and equate pd of 1.4 with ss 140%, which is wrong.

[dburne]

Ss scale is not the same as as PD, functionally they both largely do the same thing, pd is a blunter instrument than SS, and a pd of 1.4 is the equivalent of 196% ss, both in terms of performance impact and display quality.

 

I currently use neither, as they have a crushing impact on performance at that level on my hardware.

 

PD is perceived as having a higher impact on resources becuase it is a lower number, and people try and equate pd of 1.4 with ss 140%, which is wrong.

 

Yeah with the resolution of the Reverb I would not think any SS would be needed.

[Fab]

Just wait and see, after 2 weeks, they want more

[Adam]

Ss scale is not the same as as PD, functionally they both largely do the same thing, pd is a blunter instrument than SS, and a pd of 1.4 is the equivalent of 196% ss, both in terms of performance impact and display quality.

 

I currently use neither, as they have a crushing impact on performance at that level on my hardware.

 

PD is perceived as having a higher impact on resources becuase it is a lower number, and people try and equate pd of 1.4 with ss 140%, which is wrong.

 

Thanks SoS...

 

Are you saying that they're not different at all (as in pro's or con's between the two) - just that the variable is on a different scale and people are confused to what equals what between the two, or does one have benefits over the other in certain areas?

[Adam]

Yeah with the resolution of the Reverb I would not think any SS would be needed.

 

FWIW - I have the Reverb and I find there's a huge difference between 1.0 and 1.5 - especially when it comes to being able to read labels and the screens in the cockpit.

 

At native resolution of 1.0 - there's a level of guesswork and fuzziness - especially (but not limited to) the DDI's. At 1.5 I find things much much clearer.

 

The problem with my 2080 is that I get ghosting (if that's the right term) when I do rolls or turn hard through the sky with the terrain. Even though my FPS show 30 or more (through steamVR) - it seems as though the frames are closer to 10-15 - I can see when they jump quite easily - and the jump is quite a distance as well between images as the two overlap/ghost over each other.

 

I'm new to VR and have no idea of what I should be expecting, and whether other DCS users just put up with the poor text in the cockpit, or put up with the jumpy frames when doing sudden movements so they can read DDI's clearer, or if I'm doing something wrong and it's possible to be able to read the DDI's as well as have smooth performance in fast turns.

 

MSAA and SSAA are all off. Shaddows are off (or set to flat for terrain). Steam is set to 100% when I bump DCS up to 1.5.

 

I appreciate the 3 replies and the interest in my thread - however I still as clueless as I was when I first posted the question as to what the differences are between using SteamVR's SS as opposed to DCS's PD and which one is better to use if using them... and why.

[FoxTwo]

As people have said, they're the same, just at different scales. 1.5 PD is not the same as 150% in steam.

 

TBH I find 150% in steam to be a nice balance of performance and clarity with my reverb.

[Steel Jaw]

I currently use neither, as they have a crushing impact on performance at that level on my hardware.

 

 

Dude this is not encouraging...your hardware is stellar.:(

[Adam]

As people have said, they're the same, just at different scales. 1.5 PD is not the same as 150% in steam.

 

TBH I find 150% in steam to be a nice balance of performance and clarity with my reverb.

 

Aah - I see. So no benefit at all using one over the other - just be aware of the different in measurements to not mistakenly counter one for the other. Thanks!

[Adam]

Dude this is not encouraging...your hardware is stellar.:(

 

64GB Ram, 2080 Super, i7 10700K @ 3.8GHz. I haven't overclocked anything yet but I was expecting to be able to read what was on the screen without having to stretch forward and squint at the blur.

 

That's why I was wondering if I'm doing something wrong with my settings, whether there's difference between SS and PD (which there isn't apparently) - or whether it's just to be expected. I'm a complete VR noob - have no idea on expectations, what's reasonable, etc - so it's not a complaint - just looking for clarification.

 

So far - I'm guessing what I have is to be expected, and I need to put up with the fuzzy cockpit and DDI screen to have a smooth experience.

[speed-of-heat]

Dude this is not encouraging...your hardware is stellar.:(

 

It's my GPU that holds me up at the moment, the 1080ti is a good card, but the frame times don't lie, I know that a 2080ti can handle a about 120-150 SS depending on your settings... I will let you know where the bottle neck is on the 3090 ;)

[Taz1004]

Aah - I see. So no benefit at all using one over the other - just be aware of the different in measurements to not mistakenly counter one for the other. Thanks!

 

Yea, they multiply each other. Advice of leaving in-game PD to 1 is not necessarily because it's inefficient. I personally have not noticed performance difference between in game PD and Steam SS. But rather, to keep it simple by adjusting at one place. And I prefer adjusting in SteamVR settings because it tells you the exact resolution. And you can make finer incremental adjustments.

 

And make sure you check the Per application setting in SteamVR too. That'll be the final resolution.

Edited September 2, 2020 by Taz1004

[speed-of-heat]

64GB Ram, 2080 Super, i7 10700K @ 3.8GHz. I haven't overclocked anything yet but I was expecting to be able to read what was on the screen without having to stretch forward and squint at the blur.

 

That's why I was wondering if I'm doing something wrong with my settings, whether there's difference between SS and PD (which there isn't apparently) - or whether it's just to be expected. I'm a complete VR noob - have no idea on expectations, what's reasonable, etc - so it's not a complaint - just looking for clarification.

 

So far - I'm guessing what I have is to be expected, and I need to put up with the fuzzy cockpit and DDI screen to have a smooth experience.

 

I would definitely OC your cpu tbh, that's likely what is holding you back atm, depending on your cooling solution/mb/Psu you should get 4.8 easy out of a 10700k

 

Invest in fps r, to see where your bottle neck is,...

[Steel Jaw]

What are the odds that a HMD is bottlenecked by the CPU in his case? AFAIK HMD are GPU intensive.

[speed-of-heat]

Honestly it's difficult to judge, unless you know the full hardware spec, and the settings and what background apps and services are active... fpsVR will absolutely show you if it is gpu or cpu bound, for me it's pretty obvious I have gpu frame times in excess of 12ms usually 14 or 15 and cpu times of 7 or 8ms. Cpu is in the green, which is good, gpu is orange which means not so good, and intensive scenarios the cpu goes to orange, and the gpu red...

[imacken]

Yeah with the resolution of the Reverb I would not think any SS would be needed.

 

Not true. Varies from individual to individual, but I use 200% on my Reverb in SteamVR SS and the difference is huge in clarity terms.

 

Massive hit on resources obviously, but like everything else in DCS VR, it's all about finding the compromise.

[Adam]

Honestly it's difficult to judge, unless you know the full hardware spec, and the settings and what background apps and services are active... fpsVR will absolutely show you if it is gpu or cpu bound, for me it's pretty obvious I have gpu frame times in excess of 12ms usually 14 or 15 and cpu times of 7 or 8ms. Cpu is in the green, which is good, gpu is orange which means not so good, and intensive scenarios the cpu goes to orange, and the gpu red...

 

 

 

For me, after making a few changes and reducing the PD/SS - I've got GPU mostly green, and yellow on occasion. Seems that my processor is doing fine though - it hasn't been maxed out at all from what I can tell (except when initially loading)

 

I'm just starting to work through what all the settings mean (I'm a noob with all this stuff, and the phrase 'all the gear and no idea' is highly applicable - I'm on a steep learning curve).

 

The main question was just if there was a difference between PD and SS which there obviously isn't - so I'll use steam. (Gives more control over it in the end - finer adjustments). From what I can tell there's a significant difference between 100% and 140% - but after that, the improvements are difficult to see while the performance hit continues to take it's toll...

 

Now that I've figured out that, I'll move onto trying to understand the next lot of settings.

 

I'm also wondering if I might be expecting too much from DCS and VR at this time. (As mentioned, zero previous experience with VR and straight into this) - I was probably thinking that pictures would be a little sharper than what they are. I guess I should have tried a rift or something with less resolution first to appreciate what I've got. (Either that, or I'm doing something very dumb and a single setting somewhere I need to change which is probably a good possibility too).

 

Specs as requested:

 

2080 Super

i7-10700K @3.80Ghz

64GB RAM

2 x 2TB PCIe SSD

HP Reverb Pro

 

At this stage I'm managing to get 60fps for most of the time, and it can get as low as 30fps at some times (using Steam FPS VR). When it gets low, it's the GPU that's in the orange, not the CPU.

 

I've managed to get rid of the stepping / ghosting and most of the playing I've done doesn't seem to get me any clearer cockpit but seems to chop into my FPS's a lot.

 

This is currently with the following settings

 

NVidia Control Panel:

 

Image Sharpening - Off

Anisotropic Filtering:16x

Antialasing FXAA - Use Global Setting (Off)

Antialasing Gamma Correction - Use Global Setting (Off)

Antialasing Mode - Use Global Setting (App Controlled)

Antialasing Setting - Use Global Setting (App Controlled)

CUDA - GPU's - Use Global Setting (All)

Low Latency Mode - Use Global Setting (Off)

Max Frame Rate - Use Global Setting (Off)

MFAA - Use Global Setting (Off)

OpenGL Rendering GPU - Use Global Setting (Auto-Select)

Power Management - Perfer Max Performance

Shader Cache - Use Global Setting (On)

Texture Filtering - Anisotropic Sampling Optimization - On

Texture Filtering - Negative LOD Bias - Clamp

Texture Filtering - Quality - (I've played with both high performance and High Quality)

Texture Filtering - Trilinear Optimization - Use Global Setting (On)

Threaded Optimisation - Use Global Setting (Auto)

Tripple Buffering - Use Global Setting (Off)

Vertical Sync - - Use Global Setting (Use 3D Application Setting)

Virtual Reality Pre-Rendered Frames - Use global setting (1)

Virtual Reality - Variable Rate Super Sampling - Adaptive

 

 

Mixed Reality Portal

 

Experience Option - Optimize for Performance

Resolution - Automatic Upscaling (best performance)

Frame Rate - 60Hz

 

 

SteamVR

 

Render Resolution - Custom

Resolution Per Eye - 2564x2512 (136%)

Advanced Supersampling Filtering - Off

Overlay Render Quality - Low

Motion Smoothing - Enabled for DCS

Use Legacy Projection Mode - Off

 

 

DCS

 

Textures - High

Terrain Textures - High

Civ. Traffic - Off

Water - Medium

Visib Range - High

Heat Blur - Off

Shaddows - Off

Resolution - 1280x768 (lowest it will go)

Aspect Ratio - 1.66666

Monitors - 1

Res of Cockpit Display: 1024

MSAA - Off

Dept of Field - Off

Lens Effect - None

Motion Blur - OFF

SSAA - OFF

SSLR - OFF

Clutter/Grass - 750

Trees Visibilty - 90%

Preload Radius - 100,000

Chimney Smoke Density - 0

Gamma - 2.2

Anisotropic Filtering - 16x

Terrian Object Shaddows - Default

Cockpit Global Illumination - OFF

Messages font scale - 1

Scale GUI - 1

Rain - Off

VSync - Off

Full Screen - Checked

 

VR Pixel Density - 1.0

Bloom Effect - Off

MSAA Mask Size - 0.53

 

 

There is probably stuff there that could be turned on that would make nicer images with little performance cost - I'm still working my way through what it all means and what is and isn't good change for little performance hit. PD / SS was my first thing to get my head around on what appears to be a long journey.

 

Thanks to all who have replied and helped!

[Adam]

Yeah with the resolution of the Reverb I would not think any SS would be needed.

 

Not true. Varies from individual to individual, but I use 200% on my Reverb in SteamVR SS and the difference is huge in clarity terms.

 

Massive hit on resources obviously, but like everything else in DCS VR, it's all about finding the compromise.

 

Just thought I better add what I've just discovered in incase it's helpful elsewhere. I've been boosting PD / SS to try and bring more clarity - especially to the cockpit.

 

I've recently been playing around with settings outside of DCS (in particular in WMR) and choosing quality over performance in some areas. (The main things I changed that I recall were changing performance to quality and 60hz to 90hz). To the point where things are much clearer now for me at 1.0PD / 100% SteamVR SS in the cockpit than they were with PD at 1.2!!!!

 

I've got a lot more learning to do and a bit more verification - last minute success late at night when tired - but it seems that my problems with clarity were external to DCS, and that supersampling was a heavy handed approach to correct which was not required.

 

It may be that increasing PD increases clarity even more - but at present the improvement I'm seeing is so much better than what I was experiencing that I'd be willing to say no supersampling is quite playable and very readable - if the other settings are done correctly! (I'm just wondering how much more improvement I can get by tweaking the other settings now).

 

When I have time I'll try and relist my specs and settings compared to what I've posted above to give an indication.

[dburne]

Regarding setting PD in the game versus SS through another method like Steam VR or Oculus.

 

I myself go by one rule. If the game has built in support for it, I use the in game setting. If not, I use the SS setting for my Rift S. I never use both.

 

Not saying that is right or wrong either way, just the way I do it and it works fine for me that way.

[Steel Jaw]

What is the visual difference in cockpit of 1024 versus 1024 every frame?

[imacken]

Try it and see!

[Adam]

What is the visual difference in cockpit of 1024 versus 1024 every frame?

 

I have no idea - I've tried it and couldn't tell if it was a placebo effect or if there really was a difference.

 

I'd love some clarification as well, but figure it'd get a greater chance of being answered in a dedicated thread. So... https://forums.eagle.ru/showthread.php?p=4475670#post4475670 ;)

Edited September 3, 2020 by Adam