• 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.

PS4 Rumors , APU code named 'Liverpool' Radeon HD 7970 GPU Steamroller CPU 16GB Flash

Status
Not open for further replies.

onQ123

Member
Sega_Genesis_CDX_with_Official_6_button_controller.jpg

You forgot one more redesign: Sega Genesis CDX.

I knew it was something I was forgetting Sega went all out with the Genesis redesigns
 

onQ123

Member
PS1 games usually had 2000 polygons per frame. PS2 was capable of 166,666 polygons per frame (or 333,332 polygons at 30fps).

To get that leap again, PS4 would need to render 166 million polygons per frame (Uncharted 2 = 2 million polygons per frame).

I don't think there's a GPU in the world that can do that. You also have to remember diminishing returns. PS1 games were a blocky mess. We now have the rendering power to have characters with a full set of teeth, fingers etc so where would all those extra resources go? Perhaps making them smoother but that's it.

I think we are a long way from diminishing returns because we still need to reach the level where we can get real-time Ray Tracing\ Path Tracing in games.

Real-time photorealistic rendering: Urban Sprawl

soa1.png
 

USC-fan

Banned
PS1 games usually had 2000 polygons per frame. PS2 was capable of 166,666 polygons per frame (or 333,332 polygons at 30fps).

To get that leap again, PS4 would need to render 166 million polygons per frame (Uncharted 2 = 2 million polygons per frame).

I don't think there's a GPU in the world that can do that. You also have to remember diminishing returns. PS1 games were a blocky mess. We now have the rendering power to have characters with a full set of teeth, fingers etc so where would all those extra resources go? Perhaps making them smoother but that's it.

Just give you an idea of next Gen ps4. Couldn't find you 2 million number but anyway.


In the listing, Naughty Dog states that the character artist should have “Prior professional experience with normal map extraction of human models for next-gen games.”

Additionally, the listing mentioned, ”this digital sculptor will be responsible for making million poly models game ready, and support highest-res characters for renderings.”

What these quotes suggest is very simple: Naughty Dog is looking to develop a next-gen game with million polygon models, something that has clearly never been done before. In fact, nothing has come even close in recent years.

As PlayStation Lifestyle mentioned: To put this in perspective, Naughty Dog stated that Kratos’ model in God of War 3 was made of 20,000 polygons. Similarly, Uncharted 2 character models were made of 30,000 polygons.

50x ps3 models....

http://www.se7ensins.com/289/naughty-dog-working-on-a-million-polygon-game/
 
my expectations for PS4:

price €/$499 standard model
gpu comparable to HD 78xx/88xx series
4GB ram with 1GB set aside for the OS, 2GB vram
PS Home out of Beta, loaded at all times and the option to boot straight into it.
Browser loaded at all times, true multitasking

announcement at Playstation Meeting january 2013, full reveal E3 '13
 

USC-fan

Banned
that doesn't mean that the models in the game will be a million polygons because the models that they use to make games always have more polygons than the in game models.

Good point but that was the only info I could find on poly count. 1 million seems like way over kill. I doubt it would be over 100k.
 

i-Lo

Member
that doesn't mean that the models in the game will be a million polygons because the models that they use to make games always have more polygons than the in game models.

Precisely. That said, I wanted to ask whether you guys think we are approaching a point of irrelevance in terms of polygons in main character because they look pretty astounding at say around 20-30K polygons. I agree with with Dragonelite and Systemfehler with regards to the improvement in IQ (AA and AF), lighting and physics which still require improvements (especially, collision physics. clipping).

That said, with regards to the capacity of crunching more polygons, there should come an advantage on things like draw distance and reduction in object pop-ins and abrupt shifts in LoDs. These are what suddenly suspend the immersion.

Now, with regards to LoDs (and once again, I am THE n00b) can't tessellation completely eradicate the need to have multiple LoDs for an object whose display is related to the distance between it and player? I remember seeing the Endless City demo by nVidia and wondered if this would be now possible for next gen.

What do you guys think?
 

BKK

Member
nope!

1084124-800px_sega_genesis_original_model_super.jpg


46%20Sega%20Genesis%202.jpg



56%20Sega%20Genesis%203.jpg



sega_nomad_2s.jpg


& I guess the nomad counts as a Genesis too so that's 3 redesigns

Edit: & the redesigns are to save money in the 1st place so what does Sony money issues have to do with it? if they make the console smaller & lighter it will save them money on parts & shipping.

Also;

QqBG0.jpg


ewfsW.jpg


5B43E.jpg


UGGde.png


YJhEk.png


0BS85.jpg


5lOJX.png


nS8LZ.jpg


V0xaD.png


Of course there's a ton more if you start getting into portable and combination CD/TV/LD etc models.
 
my expectations for PS4:

price €/$499 standard model
gpu comparable to HD 78xx/88xx series
4GB ram with 1GB set aside for the OS, 2GB vram
PS Home out of Beta, loaded at all times and the option to boot straight into it.
Browser loaded at all times, true multitasking

announcement at Playstation Meeting january 2013, full reveal E3 '13
That's happening (Browser always loaded by default) for the XMB at the end of the year from two cites indicating "New Device PS3 WebMAF" and speculation that there will be a XTV PS3.5 model with HDMI pass-thru and low power modes similar to the Xbox361.

WebMAF is the Mozilla Application Framework which in part is a STANDARD for IPC (Inter Process Communication D-Buss standards) and APIs for HTML5 web applications. For instance a Calendar or Contact list should communicate with a Chat program.

There is a new PS3 coming from multiple cites.
 

Clear

CliffyB's Cock Holster
Interlaced video blows. Using a progressive scan method always results in superior IQ, so its hardly a loss.
 
Isn't the 360 in the same position.
It has a combined ram pool but most of its games can't do 1080p either without the hardware scaler?


Scratch that, my brain read p not i, lol.

360 is in the same position. The problem isn't that the ram is split, its the fact that there is not enough of it and developers rather push other things. The fact that 360 has a unified pool doesn't help it much in that regard, while it has its own set of problems with EDRam. Alot of games fail to hit 720p/1080i while 1080p is about twice the data.
 
360 is in the same position. The problem isn't that the ram is split, its the fact that there is not enough of it and developers rather push other things. The fact that 360 has a unified pool doesn't help it much in that regard, while it has its own set of problems with EDRam. Alot of games fail to hit 720p/1080i while 1080p is about twice the data.

Every single 360 game can dispaly on a 1080i TV where as the PS3 downgrades the picture to 480p. Developers have said many times it is due to the split ram on the PS3.
 
Every single 360 game can dispaly on a 1080i TV where as the PS3 downgrades the picture to 480p. Developers have said many times it is due to the split ram on the PS3.

Its to do with the HANA Scaler chip in the 360, not unified RAM. 360 can scale to many resolutions due to this chip, the game isnt rendered any differently though, the native resolution doesnt change.
 
Gemüsepizza;39410865 said:
If you connect your PS3 to an 1080i display which does not support 720p or 1080p, you are doing it wrong.

Many early HDTV's were 480p/1080i only. While not a major problem now, it was one oversight since launch that was never fixed and was caused by the split ram.
 

androvsky

Member
Many early HDTV's were 480p/1080i only. While not a major problem now, it was one oversight since launch that was never fixed and was caused by the split ram.

The PS3 can't normally support 1080i due to the lack of a scaler chip. There's a workaround in the SDK, but it takes more memory than normal. Split memory can be a cause of a lack of memory, but so can the PS3's OS taking more memory than the 360. Also, if the developers are already using all the memory they can, they're not going to want to make big changes to support a resolution almost nobody uses.
 
Every single 360 game can dispaly on a 1080i TV where as the PS3 downgrades the picture to 480p. Developers have said many times it is due to the split ram on the PS3.

Its called hardware scaling. There are a ton of loads of games on 360 that don't even render 720p/1080i natively. And again its mostly due to developer preference. They can make a big 1080p FB and scale back textures, polys and whatnot or they can sacrifice resolution for those things.

There is no inherent problem with split ram, PCs have been doing it for more than a decade. The problem with both the consoles is that they only have 512mb(even less because of OS overhead). Thats why they quickly backed off the 720p limit they bragged about at E3.
 

Globox_82

Banned
I don't want to open a new thread for this question. How long before the official reveal of 360 and or PS3 did specs or something legit leaked? I really can't remember. I am shocked that today when we have twitter and what not, there still isn't anything leaked that looks legit. Like PS3 slim for example.
 
I don't want to open a new thread for this question. How long before the official reveal of 360 and or PS3 did specs or something legit leaked? I really can't remember. I am shocked that today when we have twitter and what not, there still isn't anything leaked that looks legit. Like PS3 slim for example.

It'll come eventually as more and more developers get more advanced kits. I still think something will come before the end of the year.
 

mrklaw

MrArseFace
I'm almost glad the PS3 has certain areas causing problems. Hopefully it means Sony will address them suitably for PS4.

I don't care if it has unified ram or not, just have enough of it to do the things developers need.
Make the OS efficient enough not to compromise games (or put more ram in to accomplish the same thing)
Cross game chat etc as standard
Smart solutions to AA built in
 

jaypah

Member
Did not know there were people who thought that. I wonder what they think of another free service that's at least as good as live, like steam.

I have seen zero people claim that on GAF. or anywhere really. If I had to think really hard about it I could maybe see it appearing in YouTube comments but, really, once you get to YouTube comments you don't really have much of an argument.

I think XBL is worth a few bucks a month but I would also pay that for steam. The only difference is that PC gamers already had all of this stuff for free so I'll never have to. MS struck the console market at the right time with their fee. I'm sure that PSN is glorious if your friends are there but mine aren't so I don't play online games as much on PS3 but the service itself isn't bad at all.
 

JordanN

Banned
I think we are a long way from diminishing returns because we still need to reach the level where we can get real-time Ray Tracing\ Path Tracing in games.
You're missing the point. I'm not talking about diminishing returns in general, I'm talking about diminishing returns in models.

PS1 games required artists to use their imagination when modeling because not all the time they had enough polygons to properly represent human anatomy. PS2 gave developers more than enough polygons to work with to properly represent humans.

So how would PS4 represent this leap again? By reinventing human anatomy?

Gemüsepizza;39405587 said:
Meaning that while it won't be a generational leap regarding polygon count, it will be a generational leap regarding effects. Still a generational leap to me.
But that could be said about any gen.
 

onQ123

Member
You're missing the point. I'm not talking about diminishing returns in general, I'm talking about diminishing returns in models.

PS1 games required artists to use their imagination when modeling because not all the time they had enough polygons to properly represent human anatomy. PS2 gave developers more than enough polygons to work with to properly represent humans.

So how would PS4 represent this leap again? By reinventing human anatomy?


But that could be said about any gen.

PS3 models look good enough to you now but once you start to see the PS4 models more & more you will see all the flaws & where they cut corners.

how about taking skin from looking like wax to looking like skin.
 
You're missing the point. I'm not talking about diminishing returns in general, I'm talking about diminishing returns in models.

PS1 games required artists to use their imagination when modeling because not all the time they had enough polygons to properly represent human anatomy. PS2 gave developers more than enough polygons to work with to properly represent humans.

So how would PS4 represent this leap again? By reinventing human anatomy?


But that could be said about any gen.

Hair, clothing, real hands and fingers, realistic joints, ...
 

rdrr gnr

Member
You're missing the point. I'm not talking about diminishing returns in general, I'm talking about diminishing returns in models.

PS1 games required artists to use their imagination when modeling because not all the time they had enough polygons to properly represent human anatomy. PS2 gave developers more than enough polygons to work with to properly represent humans.

So how would PS4 represent this leap again? By reinventing human anatomy?
I was under the impression most character models are first created in ultra-high quality and then made to work in limited specs. So, for one, we would have better looking models in-game. Secondly, there is more to human anatomy than limbs. Skin (and its hard to replicate properties), hair and its follicles, and so much more. Games look good, but they are far from photorealistic. Clothing, muscles, and so, so much more.
 

thuway

Member
You're missing the point. I'm not talking about diminishing returns in general, I'm talking about diminishing returns in models.

PS1 games required artists to use their imagination when modeling because not all the time they had enough polygons to properly represent human anatomy. PS2 gave developers more than enough polygons to work with to properly represent humans.

So how would PS4 represent this leap again? By reinventing human anatomy?

mangudon'tevenknow
 

coldfoot

Banned
Many early HDTV's were 480p/1080i only. While not a major problem now, it was one oversight since launch that was never fixed and was caused by the split ram.
It had nothing to do with Split ram, but with the feature set of the RSX with limited scaling capabilities. Also TV's that are 1080i only without 720p are garbage and should be replaced. I hope next gen consoles only support HDMI out at a minimum of 720p. No analog video output under any circumstance, and analog audio outputs only if you buy an extra add-on that converts optical to analog. They should not have anything besides HDMI,optical,network,power and USB ports.
 
Plus, wasn't in-game chat something that MS patented, or there was some other bottleneck. Ram is irrelevant, it's how it's used that matters.

I hope Sony and/or MS put aside enough for the OS to be full featured (eg 256-512MB), but still leave enough for games. The rumours of PS4 having 2GB aren't great, but if that is even lower after the OS takes its cut, then you're at WiiU area levels of ram.

The PS3 did not originally require that all games support voice chat. It wasn't a RAM issue, they just did not require it while Microsoft required that ALL games on Xbox Live support voice chat in some form (whether it be party chat through Xbox Live or whatever).

The PS3 OS, originally, did take up a huge portion the memory. It originally took up 120 megs of memory (out of the 256 reserved for gameplay), but was later reduced down to 50 megs. Source: http://www.dailytech.com/Sony+Shrinks+PS3+OS+Gives+Devs+70MB+of+Extra+Memory/article17775.htm
There are other sources too. That is just the first one that I could find. With a 256/256 split memory pool, that OS was a huge memory hog when the system first came out.
 

JordanN

Banned
PS3 models look good enough to you now but once you start to see the PS4 models more & more you will see all the flaws & where they cut corners.

how about taking skin from looking like wax to looking like skin.
What's stopping devs from doing that now? I'd say real skin approximation was already possible last gen if you look at Silent Hill's old man model.

systemfehler said:
Hair, clothing, real hands and fingers, realistic joints, ...
Out of all those, only hair would come the closest.
 
That's happening (Browser always loaded by default) for the XMB at the end of the year from two cites indicating "New Device PS3 WebMAF" and speculation that there will be a XTV PS3.5 model with HDMI pass-thru and low power modes similar to the Xbox361.

WebMAF is the Mozilla Application Framework which in part is a STANDARD for IPC (Inter Process Communication D-Buss standards) and APIs for HTML5 web applications. For instance a Calendar or Contact list should communicate with a Chat program.

There is a new PS3 coming from multiple cites.


Seems plausible, a new super slim PS3 to breath in fresh life in the ps3. But I dont see XTV happening, neither a new X361.
 
Resources. I think the technique is called sub-surface scattering or, at least, one of them.

Yes it is separable sub-surface scattering - runs pretty good on a GTX 580 ;-)
This technique also needs other high-ned effects/filters like HDR, blood and high quality AA - so the more power we get the easier such really good looking faces are possible.
 

Clockwork

Member
Its to do with the HANA Scaler chip in the 360, not unified RAM. 360 can scale to many resolutions due to this chip, the game isnt rendered any differently though, the native resolution doesnt change.

HANA is not a scaler chip.

In any case, sometimes it's funny reading misinformation. No 1080i because of a split memory pool? Hahahah, man....
 
Seems plausible, a new super slim PS3 to breath in fresh life in the ps3. But I dont see XTV happening, neither a new X361.
800 million NEW CE network connected devices from now till 2016. Both Xbox and PS3 to have full browsers for the first time ever!

Unless a refresh has low power modes it will not be able to be sold in California and several other countries in a year or so. To provide low power modes requires CPUs that can be turned off, portions of the GPU able to be turned off, I/O able to go into low power modes, Detection and turn-on from WiFi and network and on and on. This is not possible with the old PS3 or Xbox hardware, it all has to be replaced. Cell is difficult to shrink.

A cheaper PS3 with low power modes is going to be a SOC made with new hardware. Since we know a new console is coming, is it such a stretch to a HDMI Pass-thru to fully support XTV considering 800 million new CE network connected devices from now till 2016.

BOTH must have low power modes using the same hardware. The only difference between them would be the CPUs, Xbox with one more PPC CPU than PS3 and PS3 with 6 more TINY SPUs.

If HDMI pass-thru is to have overlay ability, HDMI IN has to be converted to an image in a frame buffer then converted to HDMI OUT again. In between the image in the frame buffer can be manipulated or could be codec compressed and saved making the PS3 a DVR or encripted and served making it a media server for other network connected platforms. Lowest power mode, HDMI in is connected directly to HDMI out.
 
That's more about how light bounces off a character rather then the character actually looking like a doll.

Well if you only talk about actual polygons on the model we probably already hit the sweet spot and after that it will be really difficult to notice any differences.

You notice the difference between a sphere consisting of a handfull polygons and a few thousands but I doubt the human eye can differentiate between 10 or 50 million polygons (just some random figure). So in that context you are right but luckily next-gen is not all about polygons but also about their manipulation and their final appearance.

So we already moved passed the


/--\
\--/ circle to the O ;-)
 
I think we are a long way from diminishing returns because we still need to reach the level where we can get real-time Ray Tracing\ Path Tracing in games.

Real-time photorealistic rendering: Urban Sprawl

soa1.png

Nice image, but diminishing returns does not mean that progress will not be made, as we still have a long way to go with graphical detail. It just mean that it will take alot more power to make noticeable progress as we make further advancements. For example, Tim Sweeney stated that it will take graphical hardware that is capable of doing ~5000 Tflops for a "good enough" approximation of our reality. The Xbox360/PS3 clocked in around 0.25 Tflops. The next gen of consoles will be lucky to reach 2 Tflops. It will take alot of power and GPU-generations to get close to that 5000 number.
 

i-Lo

Member
Well if you only talk about actual polygons on the model we probably already hit the sweet spot and after that it will be really difficult to notice any differences.

You notice the difference between a sphere consisting of a handfull polygons and a few thousands but I doubt the human eye can differentiate between 10 or 50 million polygons (just some random figure). So in that context you are right but luckily next-gen is not all about polygons but also about their manipulation and their final appearance.

So we already moved passed the


/--\
\--/ circle to the O ;-)

This is exactly what I mentioned earlier (quoted below). Imagine if next gen, all the characters had the same amount of detail as say Nate, Chloe, Elena, Kratos (from SP) consistently. After seeing new techniques to smooth out curves (behind the ear) during Crytek's presentation of CE3, wouldn't that constitute as sufficient even for next gen when it comes to polycount of around 20-30K per person (alone)?

Precisely. That said, I wanted to ask whether you guys think we are approaching a point of irrelevance in terms of polygons in main character because they look pretty astounding at say around 20-30K polygons. I agree with with Dragonelite and Systemfehler with regards to the improvement in IQ (AA and AF), lighting and physics which still require improvements (especially, collision physics. clipping).

That said, with regards to the capacity of crunching more polygons, there should come an advantage on things like draw distance and reduction in object pop-ins and abrupt shifts in LoDs. These are what suddenly suspend the immersion.

Now, with regards to LoDs (and once again, I am THE n00b) can't tessellation completely eradicate the need to have multiple LoDs for an object whose display is related to the distance between it and player? I remember seeing the Endless City demo by nVidia and wondered if this would be now possible for next gen.

What do you guys think?
 
R

Rösti

Unconfirmed Member
For example, Tim Sweeney stated that it will take graphical hardware that is capable of doing ~5000 Tflops for a "good enough" approximation of our reality. The Xbox360/PS3 clocked in around 0.25 Tflops. The next gen of consoles will be lucky to reach 2 Tflops. It will take alot of power and GPU-generations to get close to that 5000 number.
To put this in perspective, the IBM Sequoia supercomputer which currently tops the TOP500 list achieved 16324.8 TFLOPS (16 PFLOPS) in the HPL benchmark. Peak is estimated at around 20 PFLOPS, but Sequoia can scale to 100 PFLOPS. The machine costs millions of dollars per rack. Development of exascale computing starts this year with suggested implementation 2016-2018.

It could certainly take quite many generations to get to those numbers for consumer applications. At least if going by traditional computers. If you could overcome some of the problems (security mostly) with quantum computing, perhaps that could be utilized in some way to get closer to reality in games faster. Of course, such is quite exotic but advancements are made rapidly. IBM made some breakthroughs earlier this year in this field: http://www.pcmag.com/article2/0,2817,2400930,00.asp
 
If devs aren't careful cost, not hardware power, will be the limiting factor next generation. Games like Watch Dogs show that graphics will be a sufficient representation of reality on next gen systems, and devs will have to ask themselves if getting any better than that is worth the extra time and money.
 

onQ123

Member
If devs aren't careful cost, not hardware power, will be the limiting factor next generation. Games like Watch Dogs show that graphics will be a sufficient representation of reality on next gen systems, and devs will have to ask themselves if getting any better than that is worth the extra time and money.

if they had hardware good enough for Real-Time Ray-Tracing the games could be cheaper to make.
 
Wait, what in there gave you that idea?
Seems like they are (according to these rumors) putting a lot more money and planning into it than I thought.
Uh... why don't you just post the username? There's more than one person who posts, unless you think every random person who posts on the internet is defined by the website or vice versa.

:p

Uh...because his posts didn't have a username attached to them?
 
Status
Not open for further replies.
Top Bottom