Testing Nvidia's Multi-Res Shading In Shadow Warrior 2 [THG]

[bystander36]

http://www.tomshardware.com/reviews/shadow-warrior-2-nvidia-multi-res-shading,4803.html

It's seems Nvidia has a new method to optimize games.

Originally intended for VR, calculating one scene from multiple viewports, Multi-Res Shading found another application: improving the gaming performance of graphics cards by reducing the rendering quality of certain portions of the image.

To me, it seems like a reasonable, optional optimization for games in 1st person view. If you need a few more FPS for enjoyment, this may be better than turning down settings for the whole image. This wouldn't work so well for isometric games, and other games where your focus wouldn't be more in the middle of the screen.

 

[Bacon1]

Its neat, but makes more sense for VR where it can track eye movement. There are a few monitors now with eye tracking as well, but having it only turn on in motion would be a bonus as well.

The IQ hit is pretty huge though:

http://media.bestofmicro.com/L/A/626734/original/Shadow-Warrior-2-11.05.2016-02.58.33.02-OFF.jpg

http://media.bestofmicro.com/L/8/626732/original/Shadow-Warrior-2-11.05.2016-02.58.57.04-HIGH.jpg

Even the hands which are "center" screen are blurry not to mention the sword itself. And all for a 5% fps gain?

Our only disappointment is that the technology as it's exposed in Shadow Warrior 2 isn't happy with just degrading the peripheral image's resolution. It also removes some components, which also affects the scene's quality right in the middle of your screen, where it's supposed to be untouched. In the end, it is far from the screen captures provided by Nvidia.

Ironically the only use case they found is to run the game @ 1440p, but then you end up downscaling most of it to under 1080p, so why not just play @ 1080p anyway?

 

[bystander36]

Ironically the only use case they found is to run the game @ 1440p, but then you end up downscaling most of it to under 1080p, so why not just play @ 1080p anyway?

While I'm sure this is a case by case feature, you might not want to play at 1080p because your monitor is 1440p. Playing at 1080p means everything is downscaled if your monitor is 1440p.

I can't say how I'd like it until I actually try it myself. An image which you scrutinize and look around the whole image, is different than when playing a game and focusing on the action.

 

[tviceman]

This feature has POTENTIAL in certain kinds of games, but it's a type of feature that would need to be well implemented and also would need to be experienced to judge if it is worthwhile.

 

[Bacon1]

While I'm sure this is a case by case feature, you might not want to play at 1080p because your monitor is 1440p. Playing at 1080p means everything is downscaled if your monitor is 1440p.

I can't say how I'd like it until I actually try it myself. An image which you scrutinize and look around the whole image, is different than when playing a game and focusing on the action.

Except using the multires means most of the screen is even lower than 1080p. I'd rather turn down a setting or two than use it when it makes your hands blurry.

Also like mentioned by Tom's, the lighting on the whole scene is effected, including the center of the screen which should otherwise be unaffected.

If it was a huge performance boost to make up for over half the screen looking worse it would make more sense. But when its 5-10% at best for the big reduction in IQ I just don't see it being worth it.

 

[tential]

Except using the multires means most of the screen is even lower than 1080p. I'd rather turn down a setting or two than use it when it makes your hands blurry.

Also like mentioned by Tom's, the lighting on the whole scene is effected, including the center of the screen which should otherwise be unaffected.

If it was a huge performance boost to make up for over half the screen looking worse it would make more sense. But when its 5-10% at best for the big reduction in IQ I just don't see it being worth it.

It could be 4k middle screen 1080p rest of the screen.

Instead of trying ways to complain about the tech, realize that it actually has potential....

Are you suggesting we should have less options from VC makers? If they were capable of doing this on all games and have it be something you could change yourself, I'd defintely use it. 4/8k resolution for the main part of the screen I view and 1080p for the edges is not bad at all. I'd definitely see how it looks.

 

[bystander36]

Except using the multires means most of the screen is even lower than 1080p. I'd rather turn down a setting or two than use it when it makes your hands blurry.

Also like mentioned by Tom's, the lighting on the whole scene is effected, including the center of the screen which should otherwise be unaffected.

If it was a huge performance boost to make up for over half the screen looking worse it would make more sense. But when its 5-10% at best for the big reduction in IQ I just don't see it being worth it.

I'd have to try it to say if it's worth it.

And best case scenario was 27%/13% (hi/low). I won't declare this a fail without trying it, or even off of a single example. The idea definitely has potential.

 

[Bacon1]

And best case scenario was 27%/13% (hi/low). I won't declare this a fail without trying it, or even off of a single example. The idea definitely has potential.

From them:

It is at 4K that the impact of Multi-Res shading is greatest, be it in frames per second, frame time stability, or perceived smoothness. Whatever the case, our three mid-range GeForce cards are not able to sustain a frame rate sufficient for the game to be smooth at this very high resolution, which is rather normal.

Our only disappointment is that the technology as it's exposed in Shadow Warrior 2 isn't happy with just degrading the peripheral image's resolution. It also removes some components, which also affects the scene's quality right in the middle of your screen, where it's supposed to be untouched. In the end, it is far from the screen captures provided by Nvidia.

This option will come in useful in specific cases, though. Take our results at 1440p, for example. A resolution like 1080p doesn't tax mid-range graphics cards enough to make the visual degradation worthwhile, and the GeForce GTX 970 and 1060 aren't fast enough to play at 4K, even with MRS. In the end, we recommend against using the more aggressive MRS mode because it impacts graphics quality too negatively for our liking.

I said I think its important tech, especially for VR where you have eye-tracking so its blurring already blurred images from the side.

 

[bystander36]

From them:



I said I think its important tech, especially for VR where you have eye-tracking so its blurring already blurred images from the side.

Your response didn't seem to be aimed at what you quoted, or what I quoted. I wasn't arguing those details.

You said at best it helped 5%-10%, the charts showed much better than that at its best. You seem to have been talking about the previous post of mine.

 

[imported_bman]

I can think of a couple of variations of this method combined with dynamic resolution scaling that would be interesting.

1)Have it so both the inner and outer windows have dynamic resolutions. Make it so the outer window drops res first until it hits a min (say 60%), after that start dropping the res of the inner window until that hits a min (same min as outer min), after that if need be drop the res of both windows uniformly from there on out.

2) Make it so the portions can be dynamically adjusted. If hitting desired FPS then 100% high res portion, low res portion grows and high res portion shrinks as FPS drops until a (low res/high res) portion limit is hit.

I can think of a few more variations that would be combination of the two above methods. Outside of setting up the governor function I would imagine the above to be trivial to implement, unless MSR has some driver limitations.

 

[bystander36]

I can think of a couple of variations of this method combined with dynamic resolution scaling that would be interesting.

1)Have it so both the inner and outer windows have dynamic resolutions. Make it so the outer window drops res first until it hits a min (say 60%), after that start dropping the res of the inner window until that hits a min (same min as outer min), after that if need be drop the res of both windows uniformly from there on out.

2) Make it so the portions can be dynamically adjusted. If hitting desired FPS then 100% high res portion, low res portion grows and high res portion shrinks as FPS drops until a (low res/high res) portion limit is hit.

I can think of a few more variations that would be combination of the two above methods. Outside of setting up the governor function I would imagine the above to be trivial to implement, unless MSR has some driver limitations.

1) sounds cool to me. Perhaps a target FPS needs to be set for this to work.
2) I think some Xbox1 games have this ability, but for the whole screen, rather than parts of it. At the least, they demoed it a while back, but I don't play consoles, so someone else would have to confirm it happened in real games.

 

[moonbogg]

Acer is coming out with eye tracking, gsync gaming monitors. I'd expect this res shading tech to work with that.

 

[antihelten]

Its neat, but makes more sense for VR where it can track eye movement. There are a few monitors now with eye tracking as well, but having it only turn on in motion would be a bonus as well.

There sort of already exists a piece of software that does this, namely HiAlgo's Boost software (HiAlgo was recently bought by AMD coincidently)

HiAlgo Boost is a bit more primitive than MRS, in that it doesn't divide the frame into multiple areas and then render them at different resolution, but instead it simply switches between different resolutions (100% and 50%) based upon motion. Of course this also leads to a much higher performance boost than MRS, with HiAlgo Boost potentially doubling the framerate (although obviously only when the camera is moving).

The impact is not nearly as noticeable as one might suspect, since the camera moving around helps to hide the lower resolution:

The downside to HiAlgo Boost is the general lack of compatibility (although still vastly better than MRS), although that might improve with the acquisition by AMD.

Personally I just really hope that either AMD or Nvidia will bring something similar to Oculus' ASW to non-VR games.

 

[Phynaz]

Not a fan of this idea.

 

[antihelten]

Not a fan of this idea.

The way I personally look at it, is that this is more of a method to improve IQ, rather than a method to improve performance.

For instance if your setup can comfortably run the game at 1080p, but not at the next step up (1440p for instance), then MRS is a way to get at least some of the benefit of the higher resolution, but at a smaller performance cost.

 

[Headfoot]

Multi Res Shading could be immensely useful in Eyefinity/Surround and 21:9 sizes. The peripheral vision monitors could be shaded at much lower resolution - they already look pretty bad because of the fish eye effect. This would mean you can play surround with a small hit to FPS but a big uplift in immersion. Much like that Xbox wall projector idea Microsoft had a while back.

I think this is the most useful use case outside of VR with eye tracking.

 

[bystander36]

I do like that idea as well. I can see a lot of potential in these ideas. Sacrificing IQ around the edge of a game can allow for higher IQ in the middle. The same for surround/eyefinity, and ultra wide screens. The big questions are how difficult is it to implement, is it cost effective, and do you gain overall IQ at the same performance.

 

[antihelten]

Multi Res Shading could be immensely useful in Eyefinity/Surround and 21:9 sizes. The peripheral vision monitors could be shaded at much lower resolution - they already look pretty bad because of the fish eye effect. This would mean you can play surround with a small hit to FPS but a big uplift in immersion. Much like that Xbox wall projector idea Microsoft had a while back.

I think this is the most useful use case outside of VR with eye tracking.

That would certainly be an interesting option, although I think it would primarily be useful with FPS games and the like where you tend to look around by moving the mouse, and less so with cockpit based games (e.g. flight sims and racing games) where you look around by turning your head.

 

[Rifter]

Sounds good in theory but its either broken, or Nvidia is lieing about it. Because "Our only disappointment is that the technology as it's exposed in Shadow Warrior 2 isn't happy with just degrading the peripheral image's resolution. It also removes some components, which also affects the scene's quality right in the middle of your screen, where it's supposed to be untouched. In the end, it is far from the screen captures provided by Nvidia."

If this also degrades IQ in the middle of the screen this totally defeats its purpose.

Perhaps after they work the bugs out this will be a option but i dont see it being a option as is.

 

[bystander36]

Sounds good in theory but its either broken, or Nvidia is lieing about it. Because "Our only disappointment is that the technology as it's exposed in Shadow Warrior 2 isn't happy with just degrading the peripheral image's resolution. It also removes some components, which also affects the scene's quality right in the middle of your screen, where it's supposed to be untouched. In the end, it is far from the screen captures provided by Nvidia."

If this also degrades IQ in the middle of the screen this totally defeats its purpose.

Perhaps after they work the bugs out this will be a option but i dont see it being a option as is.

With every GPU tech, implementation is key. I doubt it's a lie, although they might gloss over some undesired side effects, or details they are still working on. You know, the typical marketing slides from everyone.

I doubt this tech is simply a checkbox by the dev's. There are likely a lot of things that need to be adjusted to make it look proper.

 

[Headfoot]

Tech like this sets the stage for actually doing something like the 360 degree room projector with a monitor in the center as the full-res focal point. Im sad MS didn't go further with that idea, I thought it was insanely cool

 

[nathanddrews]

Tech like this sets the stage for actually doing something like the 360 degree room projector with a monitor in the center as the full-res focal point. Im sad MS didn't go further with that idea, I thought it was insanely cool

The new Razr projector system took it over.