Why other VR games cannot be like Alyx?

[conejeitor]

I was playing Seeking Dawn. Great game, descent story, good models, and almost good graphics, but, why it doesn´t look like Alyx...?. I´m asking developers: Why do you think most developers just can´t improve the graphics of most VR games, I mean, enough to make them look great, and not blurry, like most VR games are.
I know it is farther than simply SS. It´s something else. I thought it was a Hardware issue, until I tried Alyx. What is it then?

[Stormblessed Gamecock]

Although few seem to agree with me the best overall game next to Alyx is Star Wars Galaxies Edge IF you include the add ons. The ILMx lab team took all the lessons they learned on Vader immortal series and the guy who made the Dojos wrote the story to Galaxies edge which actually is really good if you are a fan of the parks (which I am). The IG88 chapter is incredible.

The real answer to OPs question is because VR is still a niche. And by and large the games are over priced and under developed. I still enjoy playing VR though.

[FreddieForever#1]

Megatron の投稿を引用：

Most devs dont have valve bucks.

What about kayak vr? it looks almost as good as alyx and its made by 3 guys i think.

[conejeitor]

Stormblessed Gamecock の投稿を引用：

Although few seem to agree with me the best overall game next to Alyx is Star Wars Galaxies Edge IF you include the add ons. The ILMx lab team took all the lessons they learned on Vader immortal series and the guy who made the Dojos wrote the story to Galaxies edge which actually is really good if you are a fan of the parks (which I am). The IG88 chapter is incredible.

The real answer to OPs question is because VR is still a niche. And by and large the games are over priced and under developed. I still enjoy playing VR though.

You just can´t compare a game running on PCVR vs a game running on Quest. Quest games are all blurry, and that´s exactly what Forward rendering improves.

[Rob⛧Pentakill]

conejeitor の投稿を引用：

I was playing Seeking Dawn. Great game, descent story, good models, and almost good graphics, but, why it doesn´t look like Alyx...?. I´m asking developers: Why do you think most developers just can´t improve the graphics of most VR games, I mean, enough to make them look great, and not blurry, like most VR games are.
I know it is farther than simply SS. It´s something else. I thought it was a Hardware issue, until I tried Alyx. What is it then?

Half-Life: Alyx was designed for desktop VR and it's the first and last of it's kind.

Most projects were rushed to release by major publishers in favor of dropping VR completely. This community is still coping with reality.

[Heretic]

IanL の投稿を引用：

This is indeed the reason why. Forward rendering allows developers to use MSAA antialiasing which results in very clear graphics, especially at the lower resolutions seen in VR headsets.

MSAA can be used with deferred rendering. The difference with rendering directly to the back buffer means MSAA can be handled by the driver. Using a deferred renderer requires the use of off-screen frame buffers, but there's nothing preventing a developer from using MSAA in the shaders, which will have the same result. It just takes a little bit of setting up. The problem is that it's an expensive AA technique that would have to be applied to all off-screen renders (diffuse, normals, specular layers etc etc), where it's a lot more efficient to use a post-process AA technique using the final image and possibly the depth buffer.

I would imagine a lot of games, even when not using a deferred rendering method, will still render to an off-screen framebuffer to apply post-processing or tone mapping.

Deferred rendering can be considerably faster than forward rendering when used for its intended purpose. They're used, sometimes with the stencil buffer, to prevent wasted shader processing so only what is actually on-screen will have the demanding lighting and material calculations applied.

IanL の投稿を引用：

Most flat screen game engines had moved to deferred rendering years ago which has some great advantages, especially with lighting, but cannot use the better AA techniques like MSAA, instead tending to use TAA or FXAA which at lower resolutions introduces a lot of distance blurring.

They can.

[IanL]

Technically I agree, they can since people determined methods to enable it, but as you point out it's expensive and to my knowledge no one is using deferred rendering and MSAA with VR as they need all the processing power they can utilise as it is to maintain good frame rates.

As nVidia stated in this GameWorks Documentation:

"MSAA has always been a major drawback of deferred shading, since the geometry for lighting (in this sample, a full-screen quad) is separated from the scene geometry. Therefore we don't get the benefit of hardware determining which pixels are edges as it does in traditional forward MSAA rendering. In the deferred scenario, we have to be able to determine which pixels are complex and only shade them on per-sample frequency, with the least computation and in the least divergent way possible."

For anyone interested in the detail this nVidia document highlights one solution:

Antialiased Deferred Rendering [docs.nvidia.com]

Both UE4 and Unity as far as I'm aware only support MSAA for forward rendering. There is some limited support for 3D primitives in UE4 but nothing beyond that.

I have no idea if UE5 has introduced any support for this.

If anyone knows of any deferred rendered VR games utilising MSAA I'd be only to pleased to know.

[Caldor]

As far as I know, forward rendering is all about performance. There is much to gain with deferred rendering I think, but it would be with a heavy cost to performance. I suspect its because forward rendering just makes more sense to split up into two images, but I have not tried looking into the details of it.

I do think that is also part of why we do not have several advanced features available in VR. The main reason is probably that VR just need special shaders to work, but again, at some extra cost to performance you can render pretty much any shader as far as I know, if you just render each eye individually, but you gain something like 40-60% better performance if you use a single pass method to render frames. But it can only be done with shaders that support doing so.

In Half Life Alyx its pretty clear that the Valve dev team behind it has some masters of shader programming. I mean... look at those bottles with water. That is some impressive and realistic shader work right there. Its certainly one of the things I will have to look into, how to code my own shaders, mainly to figure out how to get existing shaders to work in VR.

Having shaders do MSAA is probably possible, but not sure it would be MSAA if its done in the shader? The shaders are on each object and texture and I think MSAA is done at some other part of the rendering? Also if its done with the shader, I am not sure it would be very good at figuring out what part of the object it is to render. Does it take into account whether the full object can be seen or not? Certainly is a lot of wasted ressources if it ends up fully anti-aliasing some detailed character or object in the game, where the player can only see a tiny corner of it. I am pretty sure with MSAA and most other types of anti aliasing that it is not applied to things that are obscurred from view.

[Heretic]

IanL の投稿を引用：

Technically I agree, they can since people determined methods to enable it, but as you point out it's expensive and to my knowledge no one is using deferred rendering and MSAA with VR as they need all the processing power they can utilise as it is to maintain good frame rates.

As nVidia stated in this GameWorks Documentation:

"MSAA has always been a major drawback of deferred shading, since the geometry for lighting (in this sample, a full-screen quad) is separated from the scene geometry. Therefore we don't get the benefit of hardware determining which pixels are edges as it does in traditional forward MSAA rendering. In the deferred scenario, we have to be able to determine which pixels are complex and only shade them on per-sample frequency, with the least computation and in the least divergent way possible."

For anyone interested in the detail this nVidia document highlights one solution:

Antialiased Deferred Rendering [docs.nvidia.com]

Both UE4 and Unity as far as I'm aware only support MSAA for forward rendering. There is some limited support for 3D primitives in UE4 but nothing beyond that.

I have no idea if UE5 has introduced any support for this.

If anyone knows of any deferred rendered VR games utilising MSAA I'd be only to pleased to know.

A deferred renderer will output the pixel to a number of layers when rendering the geometry, instead of the back buffer. The frame buffers (layers) are acting as data containers, much like how some people will use a texture to input data to a shader (like an array, where texture coordinates are used as an index). Diffuse will have the textured polygons, normals layer will hold a unit vector3 (normal) in the pixel and specular will probably just hold a float, or if clever, this layer can be used to store other data too, like bloom, for example. For the sake of argument, each pixel has up to four floating point values - rgba.

It is at this stage where shader MSAA can be applied, but since all layers have to line up perfectly, or risk visual issues, MSAA has to be taken into account when writing these pixels in the pixel/fragment shaders. This is for all the edges. What happens next is that these layers are used to calculate the lighting and build the final image (buffer). Usually this will be rendered to a orthogonal quad or blit directly to the back buffer.

Nvidia is merely covering for the limitations in the drivers, but the API is fully capable of MSAA at the geometry stage.

[Rafario]

I followed the entire discussion but this wasn't clear to me, Half Life alyx uses Deffered or Foward shading?


In either case, i've playing around with Lumen + nanite in VR with unreal 5.1, something thats not possible with Foward rendering, and I don't know about yall, but i'm more than happy to lower the resolution a bit to have an extremely realistic lighting in VR.

Nvidia also suports DLSS in VR now. This leads me to believe, that in a few years, most devs (not developing for f***ng quest) will prefer the use of Deferred shading.

[Caldor]

Rafario の投稿を引用：

I followed the entire discussion but this wasn't clear to me, Half Life alyx uses Deffered or Foward shading?


In either case, i've playing around with Lumen + nanite in VR with unreal 5.1, something thats not possible with Foward rendering, and I don't know about yall, but i'm more than happy to lower the resolution a bit to have an extremely realistic lighting in VR.

Nvidia also suports DLSS in VR now. This leads me to believe, that in a few years, most devs (not developing for f***ng quest) will prefer the use of Deferred shading.

https://www.reddit.com/r/oculus/comments/2y0sqy/valve_forward_renderers_win_for_antialiasing/
It uses "Forward+" rendering it seems, which is a type of hybrid rendering, but mainly forward rendering for the improved anti-aliasing.

UE5 is impressive, but it does not perform as well as the Source 2 engine with its forward rendering, because instead of relying on real time lighting and such, everything is prebased into the shaders and textures of each object.

UE5 does have some impressive additions to it. Nanite, Lumen and so on. But while it is impressive and performs very well with the quality it has, it would still require some pretty high end graphics cards to do graphics like HL Alyx, witout using similar methods, which can be done, but then you would not be using Lumen. Nanite could probably still be used though.

[CunkFeatures]

Cus you use a Rift S.

[dont_tread_82]

Lone Echo and Lone Echo 2 looks mighty good too. But yeah, Valve is in a league of it's own with some exceptions.

[Schranzi]

Why other VR games cannot be like Alyx?

Counter question : Why should anyone invest in something where more and more players lose ?

VR goes down the Drain and will always remain a niche. When I see what they have now presented at the CES at new "VR innovations", I'm really thinking about stopping with VR altogether ^^. it just doesn't go forward in VR anymore what you can see in all this garbage of the last Months

[conejeitor]

Schranzi の投稿を引用：

Why other VR games cannot be like Alyx?

Counter question : Why should anyone invest in something where more and more players lose ?

Way to take the subject to a completely diferent direction......
Anyways, if you would have played video games in the 80s, you would have heard a lot of people saying of them what you are saying about VR today. They were mostly >30s at the time, but I remember my grandmother (>50) saying that the whole "computer thing" was a niche fashion that was going to pass.
So, people that think like you are actually the majority at these stages of new tech. That´s why it goes slow. I mean, cellphones are also 80s tech.

[Rob⛧Pentakill]

Megatron の投稿を引用：

Most devs dont have valve bucks.

lol This game didn't cost Valve anything. They reused all their assets and used the same engine they've used since 2004. Most of the budget for games goes into marketing, Valve doesn't spend anything on marketing. This game was a win win even If It was a critically acclaimed failure. It failed because it didn't convince anyone but hardcore fandom to buy VR.