Handy Fake
Member
Ah kna man. Bloody Mags telling wuh Mackems the obvious.
-
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.
- Status
Md Ray
Member
I'd say games designed with next-gen console's SSD in mind will completely eliminate pop-in. WD2's engine is still current-gen. I too think this is bs.
True. That's correct, it's not the same as Steam's FPS indicator, the one in the vid looks different to Steam's.
Last edited:
geordiemp
Member
Ah kna man. Bloody Mags telling wuh Mackems the obvious.
I am a geordie originally, lived down south of London for so many years i need a babel fish myself.
I like the avatar, makes me appear stupid, draws them in., a semi engineer avatar would not be cool.
A bit like being on GAF
geordiemp
Member
does that last tweet from Sweeney implies U5 demo uses kraken for more then 5.5GB/s BW for streaming? or it would use it anyway even if it less then 5.5GB/s ? more then 5.5GB/s would make sense considering it uses 8K textures if so then it's most likely other next-gen platforms will use 4K textures, i expect that will be main difference between them.
Maybe but they can also run that demo at 60 FPS with less detail on Ps5, so the SSD speed and more important Latency / coherency is still necessary to go double frame rate is my take
Last edited:
FranXico
Member
The engine is fully scalable, so not only can one scale back and/or disable features for old hardware, one can also tweak settings to optimize performance.
The question is whether graphical performance settings are going to somehow become standard on consoles next gen.
Last edited:
farmerboy
Member
Maybe not as wild as you think.
HAL-01
Member
if every single frame the engine is streaming in new tris, that means we could take the per frame poly count (20 mill) x frame rate (30) and get 600 mill tris of unique geometry per second. Find out how much data an 8k asset takes per million tris and you could get a fairly good estimate of the data throughput required for this demo.
Or I could be horribly wrong doing some bro math in which case please dismiss.
Or I could be horribly wrong doing some bro math in which case please dismiss.
Moses85
Member
The Franchise is a Joke. Part 1 was ok, Part 2 trash and the upcoming looks na as well.
Lunatic_Gamer
Member
What Microsoft and Xbox can learn from Sony and Nintendo gameplay reveals
www.windowscentral.com
What Microsoft and Xbox can learn from Sony and Nintendo gameplay reveals
This is why Microsoft should pay more attention to how Sony and Nintendo present their games.
www.windowscentral.com
Md Ray
Member
I welcome this wholeheartedly. More choices, the better.
jackets with pockets
Banned
The Epic developer was talking about providing those knobs for developers, not gamers.
Last edited:
geordiemp
Member
There will be infinite settings for devs on dev kits for sure.
For consumer hardware and games....a few (2 +) options are common on current console games, like GOW 4k30 on pro or performance at 1080p50 or so. But that allows devs to optimise 2 scenarios.
Dont think PC like tuning will ever be on consoles for consumers.
Last edited:
M-V2
Member
I like how the whole conversation shifted from Teraflops to the SSDs, smart call by Sony, that demo shifted the whole conversation to the strongest point in their hardware.
Even Xbox fanboys are talking about the SSD trying so hard to downplay it they literally forgot about the Teraflops (at least from what I'm seeing)
Smart f**** Sony
Even Xbox fanboys are talking about the SSD trying so hard to downplay it they literally forgot about the Teraflops (at least from what I'm seeing)
Smart f**** Sony
Last edited:
geordiemp
Member
I like how the whole conversation shifted from Teraflops to the SSDs, smart call by Sony, that demo shifted the whole conversation to the strongest point in their hardware.
Even Xbox fanboys are talking about the SSD trying so hard to downplay it they literally forgot about the Teraflops (at least from what I'm seeing)
Smart f**** Sony
Its not smart marketing, this SSD and coherency, cache scrubbers and fast and narrow approach with attacking IO latency has been planned meticulously for 3 years. or more.
It was bold, I can imagine Cerny saying hey I dont want more CUs and flops, I want IO silicon to do this and this and people looking at him in bemusement, but why ?
Lets face it, every body thought he had gone crazy in the GDC speach, like what on earth. Underpowered, overheated whats going on...
I can imagine Cerny smiling underneath thinking just you wait, not long now .....Its not just the intelligence, its the will and tenacity to see it through. I am impressed as F.
Its a paradigm shift in concepts, I love it.
Last edited:
M-V2
Member
I'm not talking about the marketing but the narrative to be more accurate. The narrative in the past few months was about Tflops now it's about the SSDs. No matter how we spin it, that was a smart move from Sony or epic to shift that stupid narrative to what's really going to matter next gen.
Omega Supreme Holopsicon
Banned
I wonder what he means by lower detail. Lower detail on assets? Going back to older downscaled rendering method? Or just dropping resolution to 1080p, which would be meaningless if the image can be that clean.
The thing is that nanite has to crunch 1Billion triangles per frame into those 20 Million. So unless it is precrunched in some data structure on the ssd, you need that Billion triangles to be streamed from the ssd to the gpu so that nanite can do its crunching.
SportsFan581
Member
My biggest question regarding the UE5 demo is how much of this detail can be maintained at 90 or 120fps. VR will be very compelling on PS5 if they can keep a lot of this detail in place. No more PS3 looking VR games, hopefully.
Thirty7ven
Banned
I like how the whole conversation shifted from Teraflops to the SSDs, smart call by Sony, that demo shifted the whole conversation to the strongest point in their hardware.
Even Xbox fanboys are talking about the SSD trying so hard to downplay it they literally forgot about the Teraflops (at least from what I'm seeing)
Smart f**** Sony
I don’t think it’s about being smart or not. Microsoft got ahead of the narrative and sold their vision for next gen and people latched into the number they understand, and that’s 12 Teraflops. Why? Because they believe that was what set apart the PS4 and the Xbox One, therefore it stands to reason in the mind of the uneducated that it’s as simple as that. 12 teraflops, the end.
Except it’s not, and the point of hardware evolving is to break barriers and create new paths. People just got so used to the idea of more of the same because of PS3/360 -> PS4/Xbox One. They got so used to it that now even when big time developers talk about the breakthroughs and how revolutionary they are, you get an army of posters who go to war against that.
Here’s the kicker though, XBox Studios and MS itself has put more focus on Velocity Architecture and CPU than they have on the 12 TF of power angle. That’s mostly on the fanboys and shills, fed hook line and sinker. And right now it doesn’t take a genius to understand why.
SportsFan581
Member
geordiemp
Member
His words were lower QUALITY at 60 FPS, so quality must mean they toned down the 8K to something like 4k assets or it means they ran at 1080p60. Input or output quality, we do not know.
If the answer is the former, lower than 8K assests can enable 60 FPS, that is the way forward imo.
Last edited:
M-V2
Member
Let's see how is that gonna reflect on their sales, they already sold the narrative of Tflops when they launched the Xbox One X but it didn't help them. Let's see if they could change that with their games.
Elog
Member
I feel as if this discussion has deteriorated a lot. We are all gamers and want the best possible experience.
When the two new consoles were released. Many made a big number out of the lower Tflops of the PS5 compared to the Microsoft one. Then Cerny had his speech. Many saw that speech as damage control - an attempt to mitigate the lower Tflops number by talking about 'secret sauce'.
Cerny's point was that what we see on the screen is actually a result from all components off the system and we have implicitly accepted the limitations within the PC architecture (current) to brute force graphics over the GPU and then of course the Tflops number is key. However, if you can stream high quality assets in real time off your HD not only would you off-load the GPU, the drivers behind GPU performance would be different since there would be many more minor calls per time unit so that frequency and work distribution between CUs would drive performance to a significant extent in addition to Tflops as such i.e. a one Tflop is different from another Tflop in terms of what we see on the screen.
Then the Unreal Engine 5 came - and right now - Cerny seems to know what he is talking about. I love it.
I am sure games will look great on both consoles but if I had to bet money right now on something - that something would be that Cerny was bang on the money and that many hardware comparisons between the two next-gen consoles declared a victor prematurely.
When the two new consoles were released. Many made a big number out of the lower Tflops of the PS5 compared to the Microsoft one. Then Cerny had his speech. Many saw that speech as damage control - an attempt to mitigate the lower Tflops number by talking about 'secret sauce'.
Cerny's point was that what we see on the screen is actually a result from all components off the system and we have implicitly accepted the limitations within the PC architecture (current) to brute force graphics over the GPU and then of course the Tflops number is key. However, if you can stream high quality assets in real time off your HD not only would you off-load the GPU, the drivers behind GPU performance would be different since there would be many more minor calls per time unit so that frequency and work distribution between CUs would drive performance to a significant extent in addition to Tflops as such i.e. a one Tflop is different from another Tflop in terms of what we see on the screen.
Then the Unreal Engine 5 came - and right now - Cerny seems to know what he is talking about. I love it.
I am sure games will look great on both consoles but if I had to bet money right now on something - that something would be that Cerny was bang on the money and that many hardware comparisons between the two next-gen consoles declared a victor prematurely.
Omega Supreme Holopsicon
Banned
the statues were identical, but all the rocks in the demo had similar complexity with millions of polygons per asset. At the point of the demo when they said 1 billion polygons crunched into 20 million there was a cave with varied assets, not the 500 statues.
BadBreathOfTheWild
Member
The magic bullet for VR is eye-tracking with foveated rendering. Sony patents suggests it’s an active area of research for them.
Once a game knows where a player’s fovea is pointing, VR can become better quality than TV. That isn’t an exaggeration, either.
AI upscaling like DLSS is all the rage at the moment, as are other forms of intelligent image construction. Guess what? Your brain already does this for free while adding no extra watts on the APU! Your literal blind-spot is filled in by your brain making stuff up. Pretty much everything outside of your fovea is simulated in your head outside of significant movement and large visual structures.
It’s something people aren’t really aware of unless they start playing around with magic tricks and learning about human fallibility.
Most illusion tricks and sleight of hand is done by pinning an audiences fovea where you need it at certain times.
A lot of vision relies on assumption. A lot of the human experience relies on assumption.
Even the effect of moving your eyes to look somewhere else is hidden from you by the simulation your brain makes. It’s why that illusion where you look at a clock and it seems like the first second hand tick takes a lot longer than the subsequent ones. Your experience of vision is literally turned off until the eye is back on target and your brain time shifts the experience to make it seem seamless.
For VR this means although you need to render a scene twice (recent improvements by Nvidia remove most of the CPU overhead for this) you only need to render what’s under your fovea at high resolution and with any kind of expensive filtering.
TVs cannot do this, they don’t know where you’re looking.
This means that a VR headset could have 8K resolution or more, and as long as it can track your eyes and has fast enough latency it might only need to render a patch of it at “8K quality”, and the rest progressively lower down to barely anything, and you’d never know. Dart your eyes around all you want and it’s sharper even than a TV, with perfect filtering, and rendering costs comparable with (even lower at points) than a 4K TV where every inch of it is at full detail, resolution and filtering.
Expect future VR to pull well ahead of TV quality and performance.
Cannot wait to see where Sony go with it. This is the kind of next-gen I want. Not the same old formula at higher resolutions and frame-rates.
I want new worlds and experiences, not sharper versions of the same worlds and experiences.
Imagine how good Ironman could be flying around a city with PS5 IO. Or the next Spider-man. I want him to go faster and for there to be streets that span the city in length. I don’t want the same carefully designed maze with slow floaty swinging but now with sharper edges and shiner effects.
Faster IO and VR is the next-gen I’m most excited about.
Self_Destructive
Member
8K textures used in the demo seems needlessly excessive, and essentially just to prove a point of how much throughput is required to do so.
Actual games will be much smarter about this.
geordiemp
Member
Yup, hence the demo running at 60 FPS with 4K assets (or whatever the scaling is) seems a better way forward, still requires the same high streaming SSD and delivery as your working on smaller assets but at twice the frame rate.
Got to admit, the cgi quality assests made a strong point.
Last edited:
zubairali2
Member
they'll never give up
Shmunter
Member
Imtjnotu
Member
Lol okay.... Epic China.....
xacto
Member
Omega Supreme Holopsicon
Banned
what do you mean baked? They said you can use the raw zbrush and quixel scan assets in your game without normal maps or lods. This particular demo apparently made heavy use of raw quixel photogrammetry assets, and there was a wide variety of assets.
Last edited:
kuncol02
Banned
It's obviously not saved as raw zbrush files. It need to be saved into some sort of treelike structure which allow renderer to quickly grab only polygons which are needed to display. And there is question if that structure will allow for references to single model or every instance will need to have it own data in that tree. First solution would have lot of problems with for example overlapping models.
SportsFan581
Member
Has Sony patented anything on varifocal actuation? That's really the missing piece to VR, what stops you from focusing on near and far objects without blurriness. As it stands now, none of the current headsets can adjust for the changing shape of the eye as you shift focus. Varifocal actuation + foveated rendering would be out of this world.
bitbydeath
Member
I’m starting to think the assets is what Sony was referring to when mentioning 8K support.
SportsFan581
Member
I’m starting to think the assets is what Sony was referring to when mentioning 8K support.
Nah, that's regarding the next-gen Tetris or pong ports.
Darius87
Member
that's the whole point of PS5 SSD to stream highest quality assets without creating LOD's to cut dev time, it's like 2 hits with one shot (highest possible details and reduced dev time) doesn't pro and xonex already uses 4K textures in some games?
everything seems planed by sony in advance while making PS5. i'm not saying 8K textures will be in every game but i think moving on it could become standart.
geordiemp
Member
I’m starting to think the assets is what Sony was referring to when mentioning 8K support.
Thats probably referring to Hdmi 2.1 new standard output ?
azertydu91
Hard to Kill
Ahhh come on... mine was innocent.
Neo Blaster
Member
I thought it didn't use RT at all.
FeiRR
Banned
Sure. But those moments could require replacing most of assets in the memory. It should take less than 2 seconds to swap all VRAM content and those corridor sections are more or less that long. It's still a lot better than a loading screen.Isn’t every step you take a loading section?
PS5 SSD Is 'Far Ahead' of High-End PCs, Epic Games CEO Says - IGN
The PlayStation 5 breakthroughs on SSD and storage are ahead of any PC alternatives.www.ign.com
B... but where are we going to learn that you shouldn't eat yellow snow, then?
MasterCornholio
Member
Handy Fake
Member
On a slightly tangential note, I'd love to see No Man's Sky II on this engine. I wonder how it would work with procedural generation.
zubairali2
Member
i think no mans sky on ps5 would be INSANE
Roronoa Zoro
Gold Member
By user experience do you mean like showing the UI and how to change certain settings and the store fronts and stuff? I love seeing new UIs don't know whyI’m starting to think the assets is what Sony was referring to when mentioning 8K support.
Last edited:
Handy Fake
Member
The asset streaming alone would make it incredibly stable.
Thirty7ven
Banned
Like Clockwork
Wow, that's a damn good point. I'd never thought of that.
FeiRR
Banned
Or someone's been playing with AI. I felt a bit weird when I saw this first.
Have you by any chance read Peter Watts' books? He explores the ideas of perception and consciousness in a very engaging way. I also found this very interesting since DualSense is getting a heartbeat sensor.
Last edited:
- Status