• Hey, guest user. Hope you're enjoying NeoGAF! Have you considered registering for an account? Come join us and add your take to the daily discourse.

Tech Analysis: Unreal Engine 5 on PS5 - Epic's Next-Gen Leap Examined In-Depth



Alex from DF just put up an Inside Unreal Engine 5 article


It's a long article, some snippets

Epic's reveal of Unreal Engine 5 running in real-time on PlayStation 5 delivered one of the seismic news events of the year and our first real 'taste' of the future of gaming. A true generational leap in terms of sheer density of detail, alongside the complete elimination of LOD pop-in, UE5 adopts a radical approach to processing geometry in combination with advanced global illumination technology. The end result is quite unlike anything we've seen before, but what is the actual nature of the new renderer? How does it deliver this next-gen leap - and are there any drawbacks?

Watching the online reaction to the tech trailer has thrown up some interesting questions but some baffling responses too. The fixation on the main character squeezing through a crevice was particularly puzzling but to make things clear, this is obviously a creative decision, not a means to slow down the character to load in more data - it really is that simple. Meanwhile, the dynamic resolution with a modal 1440p pixel count has also drawn some negative reaction. We have access to 20 uncompressed grabs from the trailer: they defy traditional pixel counting techniques.

Some interesting topics have been raised, however. The 'one triangle per pixel' approach of UE5 was demonstrated with 30fps content, so there are questions about how good 60fps content may look. There have also been some interesting points raised about how the system works with dynamic geometry, as well as transparencies like hair or foliage. Memory management is a hot topic too: a big part of the UE5 story is how original, full fidelity assets can be used unaltered, unoptimised, in-game - so how is this processed? So, to what extent is the Lumen in the Land of Nanite tech demo leveraging that immense 5.5GB/s of uncompressed memory bandwidth?

Core to the innovation in Unreal Engine 5 is the system dubbed Nanite, the micro-polygon renderer that delivers the unprecedented detail seen in the tech demo.

"With Nanite, we don't have to bake normal maps from a high-resolution model to a low-resolution game asset; we can import the high-resolution model directly in the engine. Unreal Engine supports Virtual Texturing, which means we can texture our models with many 8K textures without overloading the GPU." Jerome Platteaux, Epic's special projects art director, told Digital Foundry. He says that each asset has 8K texture for base colour, another 8K texture for metalness/roughness and a final 8K texture for the normal map. So, we end up with eight sets of 8K textures, for a total of 24 8K textures for one statue alone," he adds.

ince detail is tied to pixel amount in screen size, there is no more hard cut-off - no LOD 'popping' as you see in current rendering systems. Likewise, ideally, it should not have that 'boiling' look like you can see with standard displacement as seen in with ground terrain in a game like 2015's Star Wars Battlefront (which still holds up beautifully today, it has to be said).

In lieu of triangle-based hardware-accelerated ray tracing, te UE5 demo on PlayStation 5 utilises screen-space as seen in current generation games to cover small details, which are then combined with a virtualised shadow map.

"Really, the core method here, and the reason there is such a jump in shadow fidelity, is virtual shadow maps. This is basically virtual textures but for shadow maps. Nanite enables a number of things we simply couldn't do before, such as rendering into virtualised shadow maps very efficiently. We pick the resolution of the virtual shadow map for each pixel such that the texels are pixel-sized, so roughly one texel per pixel, and thus razor sharp shadows. This effectively gives us 16K shadow maps for every light in the demo where previously we'd use maybe 2K at most. High resolution is great, but we want physically plausible soft shadows


We were also really curious about exactly how geometry is processed, whether Nanite uses a fully software-based raw compute approach (which would work well across all systems, including PC GPUs that aren't certified with the full DirectX 12 Ultimate) or whether Epic taps into the power of mesh shaders, or primitive shaders as Sony describes them for PlayStation 5. The answer is intriguing.

"The vast majority of triangles are software rasterised using hyper-optimised compute shaders specifically designed for the advantages we can exploit," explains Brian Karis. "As a result, we've been able to leave hardware rasterisers in the dust at this specific task. Software rasterisation is a core component of Nanite that allows it to achieve what it does. We can't beat hardware rasterisers in all cases though so we'll use hardware when we've determined it's the faster path. On PlayStation 5 we use primitive shaders for that path which is considerably faster than using the old pipeline we had before with vertex shaders."

he other fundamental technology that debuts in the Unreal Engine 5 technology demo is Lumen - Epic's answer to one of the holy grails of rendering: real-time dynamic global illumination. Lumen is essentially a non-triangle ray tracing based version of bounced lighting - which basically distributes light around the scene after the first hit of lighting.

"Lumen uses ray tracing to solve indirect lighting, but not triangle ray tracing," explains Daniel Wright, technical director of graphics at Epic. "Lumen traces rays against a scene representation consisting of signed distance fields, voxels and height fields. As a result, it requires no special ray tracing hardware."

To achieve fully dynamic real-time GI, Lumen has a specific hierarchy. "Lumen uses a combination of different techniques to efficiently trace rays," continues Wright. "Screen-space traces handle tiny details, mesh signed distance field traces handle medium-scale light transfer and voxel traces handle large scale light transfer."

And finally, the smallest details in the scene are traced in screen-space, much like the screen-space global illumination we saw demoed in Gears of War 5 on Xbox Series X. By utilising varying levels of detail for object size and utilising screen-space information for the most complex smaller detail, Lumen saves on GPU time when compared to hardware triangle ray tracing.

Another crucial technique in maintaining performance is through the use of temporal accumulation, where mapping the movement of light bounces occurs over time, from frame to frame to frame.
 
Last edited:

BrentonB

Member
I wonder how much of a difference using 4K textures would make. Would that facilitate a higher frame rate if a creator wants to go that way?
 
I think the question everyone wants answered is this: How would this tech demo look and run on Xbox Series X. My bet is "exactly the same". Maybe, maybe X could push a higher res, or framerate ceiling. MS should really put out a tech demo of its own just to show the performance and get out in front of /diminish all this PS5 wins cause SSD BS.

I think it is a question of when, not if, MS will put out the XsX UE5 demo. Do they wait until July? Do they do it ASAP?
 

Evilms

Banned
  • Lumen is not ray tracing, it's using another form of tracing
  • lumen also has specular reflections
  • large objects are traced through voxels
  • medium objects are signed distance fields
  • small objects are screen space (like Gears 5 on XseX)
  • you can see screen space artifacts in the demo
  • uses temporal accumulation like RT, so there's a latency to lighting
  • micro-polygon rendering is primarily used in offline rendering like film
  • nanite uses a high resolution tiling normal map for fine details to help conserve vram through virtual texturing
  • nanite scales the model complexity by how many pixels it takes up
  • micro sized objects are shadowed with SS shadows and combined with a virtualized shadow map
  • shadow map resolution is aligned with screen resolution
  • shadows are filtered to create a penumbra
  • unknown if nanite applies to animated objects like foliage or hair, or characters
  • demo is dynamic res (mostly 1440p) at 30fps
  • resolution scaling is more expensive with this technique
 
D

Deleted member 775630

Unconfirmed Member
I think the question everyone wants answered is this: How would this tech demo look and run on Xbox Series X. My bet is "exactly the same". Maybe, maybe X could push a higher res, or framerate ceiling. MS should really put out a tech demo of its own just to show the performance and get out in front of /diminish all this PS5 wins cause SSD BS.
Depends on a couple of factors of course. How big is was this demo and how big are those assets? If they are so big that you need a PS5 speed of SSD, then the XSX can't do this. But then again, if they are so big, will developers ever create something like that? That takes huge amounts of time for a complete game (not just a demo), and how big is this game going to be? 300GB or even more? Demo's are nice because you can put a lot of resources in only a small world, and use the whole system for it. But that doesn't always realistically translate to real cases, and then you can't really use that speed to your advantage.
 

INC

Member
Lighting is great, and is basically faux ray tracing

If it uses far less resources and looks like that, I couldnt give a fuck about ray tracing
As long as it speeds up dev time to worry about other stuff, like fixing bugs, I'm all for this lighting solution

On pc ray tracing basically means half your frame rate
 
Last edited:
How does one spin this into less ssd = same visuals? Lets find out soon in this thread
12.28 TF is orders of magnitude better than 10.28! and that SSD adds no TF at all, so whatever.

I still don't get how they pull all these models into memory, and where they store them... that demo should be above 1TB by itself :-/
 

INC

Member
12.28 TF is orders of magnitude better than 10.28! and that SSD adds no TF at all, so whatever.

I still don't get how they pull all these models into memory, and where they store them... that demo should be above 1TB by itself :-/


Bit like American muscle cars and euro/jap cars

American use a massive v8 power plant to get 500bhp
Euro/jap use a 3 litre and turbo chargers (ssd) to get 600bhp

Disclaimer: I have no idea if that works, I know fuck all about cars, or even consoles

Just sounded nice lol
 
Last edited:

Dural

Member
I think the question everyone wants answered is this: How would this tech demo look and run on Xbox Series X. My bet is "exactly the same". Maybe, maybe X could push a higher res, or framerate ceiling. MS should really put out a tech demo of its own just to show the performance and get out in front of /diminish all this PS5 wins cause SSD BS.

Maybe they'll have a demo showing Gears5 using the tech? They already had the screen space GI demo with it, so I could see a lumen demo coming.
 

longdi

Banned
Lighting is great, and is basically faux ray tracing

If it uses far less resources and looks like that, I couldnt give a fuck about ray tracing
As long as it speeds up dev time to worry about other stuff, like fixing bugs, I'm all for this lighting solution

On pc ray tracing basically means half your frame rate

Yeah faux RT. The lighting still looks current gen.
Nvidia newest RT demo, the marble game, shows off hardware RT. Very realistic feels and sight!
 

INC

Member
Yeah faux RT. The lighting still looks current gen.
Nvidia newest RT demo, the marble game, shows off hardware RT. Very realistic feels and sight!

No doubt, but this is good enough, and looks perfectly natural to my eyes
But current gen is still for th most part, bake lighting, this from what they're saying isnt, so its rel time, that will speed up dev time, which is what I care about more

Same with ray tracing, in the long run itll speed up dev time, but it's far to resource heavy, for even monster PCs. (At the moment, with more time and optimisation itll become less of an issue)

This is a great middle ground for next gen games, so cant really complain
 
Last edited:
I think the question everyone wants answered is this: How would this tech demo look and run on Xbox Series X. My bet is "exactly the same". Maybe, maybe X could push a higher res, or framerate ceiling. MS should really put out a tech demo of its own just to show the performance and get out in front of /diminish all this PS5 wins cause SSD BS.
If it pushes an higer resolution (let's say 1600p instead of 1440p) it won't push an higer frame rate, if it pushes an higher frame rate (60fps) it will have to cut the resolution, likely close to 1080p... unless it cuts on the details, then it would have to render less details, so obviously it would run better... and it's very likely that it would look so close that nobody would know better and they enjoy the higher resolution and/of frame rate more than they would notice the compromise, but that would not be because of the GPU.
 
That demo was impressive, take a way what "platform" it was "showcased on" and just watch it for what it is. Take off whatever platform googles you have on. The tech they demo'd in that thing sounds so cool. I keep hearing the argument about "where's the ray tracing" Ray tracing never even crossed my mind. The lighting looked so good I don't care what tech is being used, I'm not a program WTF I care about if it has ray tracing or not?!? Oh it's 1440P... oh I'll be honest, I have a 4k Monitor in front of my face and I couldn't tell you the difference between 1440p, 4k, 2k up scaled, blah blah blah.

I don't care if this engine is made to reduce cost to GPU and CPU, I don't care, If it helps make games that look this good why the hell would I care? Oh it's not ray tracing but why do you care if it looks that good? Oh that's right, it's a bullet point you can use as part of your console war argument. 12 v 10, oh that's right it's another bullet point. If that's what games could potentially look like in the next gen, I'm in! My bank account will hate me, but I'm in.
 
If it pushes an higer resolution (let's say 1600p instead of 1440p) it won't push an higer frame rate, if it pushes an higher frame rate (60fps) it will have to cut the resolution, likely close to 1080p... unless it cuts on the details, then it would have to render less details, so obviously it would run better... and it's very likely that it would look so close that nobody would know better and they enjoy the higher resolution and/of frame rate more than they would notice the compromise, but that would not be because of the GPU.
I don't think people think like you do. There is a cost to everything and there will be trade offs, like you mentioned. I'm not making the argument that the PS5's SSD is making up for the lower TF numbers, but if this tech is relying so much on streaming of assets, streaming of "billions" of triangles ( :messenger_tears_of_joy:) wouldn't the PS5's much higher speed make a difference? I'm asking cause I don't know.
 
I don't think people think like you do. There is a cost to everything and there will be trade offs, like you mentioned. I'm not making the argument that the PS5's SSD is making up for the lower TF numbers, but if this tech is relying so much on streaming of assets, streaming of "billions" of triangles ( :messenger_tears_of_joy:) wouldn't the PS5's much higher speed make a difference? I'm asking cause I don't know.
What I suspect is that like the TF it will make a difference in edge cases... only maybe more because it's twice as fast as the competition in therms of bandwidth as opposed to 15-20% difference in GPU computing power.

That being said, what we don't really know is how Sony's solution relates to Radeon SSG, where the GPU is directly connected to the SSD, as opposed to what seems to be a decent NVmE 3.0 drive that MS got for their machine... the difference being how close it is to the memory banks, and how often it can get information from it in a given time slice... and does it matter at all? sometimes or all the times?

I suspect that for existing games both solutions work pretty similarly, there is not much return to having twice the bandwidth to play the latest Assassin's Creed, at least I suspect, but games designed around that tech should show benefits.
 
Last edited:

eot

Banned
Damn, Battlefront still looks pretty sick five years later.

Anyway, I didn't get his point about the scaling at the end. Clearly this technique would see a much bigger difference in performance when changing resolution, since polycounts and shadows are directly tied to the internal resolution. While that means it's harder to render at native 4K, you'd also gain FPS much quicker when dropping the resolution, so why was he speculating that 1080p/60 might be unrealistic?
 

INC

Member

Does that mean lower detail on a slower ssd/hdd then, like a rival console or pc then? If they're not using the same speed ssd as the demo (ps5 ssd)

So indirectly the ssd is adding to the detail of the game world, and does have a big impact?
 

Andodalf

Banned
Damn, Battlefront still looks pretty sick five years later.

Photogrammetry is a hell of a drug. The rocks shown in the demo look similar, but what’s awesome is the dynamic scaling of the assets to give even higher than the highest quality at extreme close range, while having perfectly appropriate quality at range, automaticity giving ideal performance and quality
 

onQ123

Member
Does that mean lower detail on a slower ssd/hdd then, like a rival console or pc then? If they're not using the same speed ssd as the demo (ps5 ssd)

So indirectly the ssd is adding to the detail of the game world, and does have a big impact?


Yes the speed of the HDD/SSD will have a big impact on the level of detail because fast streaming allows devs to use the pool of physical RAM for what is being rendered on the screen instead of the RAM being used to store stuff that you might need in the next 20 - 30 seconds.



This is the reason why tech demos of small scenes look so good because they can use all the RAM for what you see on the screen because it's a demo & you're not going outside of what is on the screen in the small scene. The fast SSDs allow them to use this trick but you can actually go beyond the small space because it can be streamed into RAM as you go.
 

Meowzers

Member
Anyone that doubts the gameplay from UE5 tech demo, should see how good that Battlefront game really is.

That game from like 2016 or so easily tops anything that Xbox showed off last week.
 

Sethbacca

Member
I don't get why everyone is so concerned about the performance in that demo. That shit was a triangle torture test for the ps5. I can't imagine that most studios will be using assets containing even half as many polys as the cinema quality assets used in that video. Even at quarter to half the poly density, this shit would look spectacular with the lighting model and everything else going on. I think 1440p/60 is going to be easily achievable for a lot of games depending on art direction.
 

Shin

Banned
Fair comparison because it's demo vs. demo and from the same company.

 

Hendrick's

If only my penis was as big as my GamerScore!
Fooling someone by telling someone that it's not made just for PS5 & that it's scale-able so that it can even work with slower HDD?


Are you that emotionally attached to console brands that you think everything someone say is to hurt your console brand?


look back at the post that I replied to
I am not attached at all I just like to keep things honest. To your point though, perhaps I had you mixed up with another poster, so my mistake.
 

Portugeezer

Member
Fair comparison because it's demo vs. demo and from the same company.

UE4 was an uninteractive demo and we got games which look better than it. And for some reason people doubt the new UE5 demo; we will get better looking games, biggest factor will probably be studio budgets.
 
It won't run worse but it will surely have lower quality assets. Sweeney specifically stated the top tier ssd is the reason this demo was possible. But lets ignore that and starts the spinzz

Sweeney was talking about the PS5 can do, not what the XsX can't do. We don't know if the XsX can't also do it. I am sure we will find out.
 

ethomaz

Banned
Some interesting quotes in the article.

"The vast majority of triangles are software rasterised using hyper-optimised compute shaders specifically designed for the advantages we can exploit," explains Brian Karis. "As a result, we've been able to leave hardware rasterisers in the dust at this specific task. Software rasterisation is a core component of Nanite that allows it to achieve what it does. We can't beat hardware rasterisers in all cases though so we'll use hardware when we've determined it's the faster path. On PlayStation 5 we use primitive shaders for that path which is considerably faster than using the old pipeline we had before with vertex shaders."

So they made use of Primitive Shaders (Mesh Shaders) to run faster on PS5.
Nanite uses software (CPU?) to generate de triangles/details instead to use the GPU rasterizers... it will only use them when needed.... that is why they said it won't affect GPU performance in the interview.

"Lumen uses ray tracing to solve indirect lighting, but not triangle ray tracing, "

Lumen is based in ray tracing but jut for indirect lighting... that is probably a way to avoid performance hits with use of ray-tracing.
 

Sethbacca

Member
The demo performed poorly?

There are many posts in the other threads discussing this topic that are whining about 1440/30 with some frame rate drops. It's an unoptimized early build of an engine that's a year or two out from general usage in a demo that's showing us the absolute most hardcore case scenario for engine usage with respect to polys and cinema quality assets. The fact that it's performing as well as it is under these conditions tells me that games are likely going to be fucking amazing.
 
Top Bottom