Future console discussion and thoughts

[Shifty Geezer]

He'll be lucky if in five years the majority of the developed world is up to 720p displays, is what.

He may have been talking about the top-end technology, rather than what's commonly adopted. I doubt TV's will go beyond 1080p @ 60 Hz for many years though. 4k x 2k only really makes sense for movies, and > 60 Hz is highly improbable as that gives smooth animation already. There's no need for faster. You're just guzzling up storage for no benefit. Perhaps movies would want to be at higher refreshes to elliminate judder on huge screens, but they haven't yet even got beyond 24 fps! Most games aren't hitting 60 fps. TV broadcasts faster than 60 fps would require stupid BW above the already stupid demands of 1080p60. Who wants to compress the image quality lower to get faster framerates when 60 fps is plenty fast enough? Compression artefacts would look far worse.

The next gen consoles should be aiming for 1080p60, constant, and using rendering power to add image quality.

 

[pc999]

Photorealism is boring, what you want is photosurrealism

That should have signature honors.

 

[Crossbar]

Kutaragio smokes a lot of crack too. None of that is going to happen.

He'll be lucky if in five years the majority of the developed world is up to 720p displays, is what.

Well those things might happen at some point in time, the 240 fps figure, well maybe not.

I wonder why Kutaragi at times seem to grab some semi-random number that are beyond belief, obviously he is a very smart guy, judging from his achievements. Is it the vision he is aiming for, is it all just some kind of distorted information just not reveal their real plans for the competetion or is it just PR-BS to grab some headlines?

 

[Cheezdoodles]

Its just PR-BS to grab headlines.

I doubt the PS3 ever had any plans to support dual 1080p.

He allso threw the 120fps @ 1080p number around, which we all know will never happend in a game on the PS3

 

[nonamer]

PS3 can actually output two HD video streams.

 

[Laa-Yosh]

I realise that this is a hardware related site but I believe that the next generation will have less emphasis on hardware. We're already getting diminished returns, especially compared to development costs. Who's gonna make four times as many textures as we've seen in Gears? Who's gonna care about the difference? It's not nearly as big as the jump from 2D to 3D.
Also, cutting costs on hardware will become a lot harder as we hit the limits of manufacturing technology. You just can't expect to cut PS4's costs from $ 8-900 down to $ 80-90. So consoles will have to become cheaper to build, which also means more simple hardware.

I'd expect more important changes with the controllers, and on the software and services side.

 

[TheChefO]

Who's gonna make four times as many textures as we've seen in Gears? Who's gonna care about the difference?

I agree that there is limt to how big a budget can be for a single game and this will limit the amount of artist's time in content creation but I also think that gamers will still crave and desire for another big leap visually from their ps4/xbox720 over their predecessors. This is where I think proceedurally generated content will help greatly. Not that I think everything will be 100% procedural but adding it as a variable to help add variety and detail is what I imagine it's role will be in the next 5 years. Also applying this same thought process into the animation and geometry side of things as well will assist creating more detailed worlds without spending exponentially more for content creation.

 

[Laa-Yosh]

100% procedural approaches just won't work for anything other than heavily stylized games, like Spore. It can help you build an entire city or variations of a certain character for an army, but it'll still require a lot of assets to work from. There's a reason why it hasn't been used that widely in movie VFX and animated CG productions either, even after decades of research. It won't solve the whole problem, just some subsets of it that people usually don't really care about.

For example, everyone accepts that all the Locust drones look the same in Gears. Epic could develop technology to make each drone different in Gears 2. A software tool could randomly combine different body parts, even change texture colors using image processing, and apply some deformations to change the shape and size of the character, although it'd be complicated to make it work with a single set of animations. So, they could make a thousand individual drones using such a tool, but they'd still have to model at least 5-8 variations for each body part. The Warhammer fantasy RTS has this feature for the armies, and it took a lot of time to build all the assets, many times as much as it took them to write the army builder tool...

But you can't really program a tool that would model and texture a Locust drone, or a car, or a house. There are some tools for plants and trees, but as far as I know even Speedtree needs a lot of assets to work with.

 

[TheChefO]

100% procedural approaches just won't work for anything other than heavily stylized games, like Spore. It can help you build an entire city or variations of a certain character for an army, but it'll still require a lot of assets to work from. There's a reason why it hasn't been used that widely in movie VFX and animated CG productions either, even after decades of research. It won't solve the whole problem, just some subsets of it that people usually don't really care about.

For example, everyone accepts that all the Locust drones look the same in Gears. Epic could develop technology to make each drone different in Gears 2. A software tool could randomly combine different body parts, even change texture colors using image processing, and apply some deformations to change the shape and size of the character, although it'd be complicated to make it work with a single set of animations. So, they could make a thousand individual drones using such a tool, but they'd still have to model at least 5-8 variations for each body part. The Warhammer fantasy RTS has this feature for the armies, and it took a lot of time to build all the assets, many times as much as it took them to write the army builder tool...

But you can't really program a tool that would model and texture a Locust drone, or a car, or a house. There are some tools for plants and trees, but as far as I know even Speedtree needs a lot of assets to work with.

Agreed - I said procedural would help. Not 100% procedural - but using it for adding variation and insignificant detail mostly to natural objects. Modelers and texture artists will of course still be needed but combining their efforts with procedural tech will help add variation and detail that would otherwise take either many times more artists or many times more hours (as you stated in your previous post).

It is not a cure all to rid the gaming world of graphic artists but it will be an invaluable tool to enable the artists to be much more productive in the quest to generate more lifelike and detailed worlds.


Question for you Laa-yosh - What do you think would be more valuable in a ps4/x720: 15x more ram or 15x more processing power? Obviously in either case the ram and processing power will both increase but which do you feel would be more valuable?

 

[Fran]

Photorealism is boring, what you want is photosurrealism

Nice

 

[Shifty Geezer]

I agree that there is limt to how big a budget can be for a single game and this will limit the amount of artist's time in content creation but I also think that gamers will still crave and desire for another big leap visually from their ps4/xbox720 over their predecessors.

A big graphical leap might not be possible due to diminishing returns. A simple quadrupling of poly counts, textures, etc, will only add a marginal improvement. To get a big and noticeable improvement would need a lot more power.

I think though, that some realtime GI tech and very high IQ, with existing levels of graphics, would fit the bill okay, if that's not achieved this gen (which is where your idea of procedural content may fall down. Fingers crossed, that happens this gen, so there's little room for a big advancement in that field.)

 

[Fafalada]

Photorealism is boring, what you want is photosurrealism

Like a photo run through saturation + 300% filter?
You know, like all those fantastic holywood blockbusters (Driven, Charlies Angels, DOA, FAF (1-3)...).

Anyway, judging by the sales of GoW, I fear we're doomed to another 10 years of Tom Forsyth's 'prophecies' about producers demands from 2001.
"Gritty! Deep shadows! Photorealism! Anisotropic normal Blinn maps with spline vectors! Whatever Tim Sweeny's doing!"

That said you could always develop exclusively for non US market

 

[Johnny Awesome]

Every few months people talk about diminishing returns and then something like Gears of War proves that gamers care about these supposed incremental improvements after all.

If next generation can deliver 1080p w/4x AA and huge amounts of textures etc... it will be enough for consumers. Gamers want their Gears, Halo, GTA, etc... games to look lifelike, but budgets will be the limiting factor.

Look at the PS3: Technically more poweful, but X360 has Gears of War because MS gave enough money to Epic to create the art assets that would deliver the best graphical experience in 2006. Next year when the PS3 gets MGS4, we'll see what it can offer when budgets are high enough, but for the most part art assets are going to be the big limiter for a very long time IMO.

That's why MS built the console they did. They looked at game budgets being realistically 10-20 million max for 95% of next-gen titles and build hardware that could exploit those budgets. PS3 is probably more power than publishers are willing to spend money to exploit in 2006. Maybe not in 2009 though when the PS3 userbase gets large enough.

So I think MS will concentrate mainly on IQ and other improvements that won't require a lot more money spent by developers for 95% of titles. They won't make a beast that requires 50-100 million game budgets to fully exploit. I'm not sure what Sony will do.

 

[pc999]

Every few months people talk about diminishing returns and then something like Gears of War proves that gamers care about these supposed incremental improvements after all.

IIRC most people (like me) said beyond UE3/this gen that would happen, not that it is already happening.

Anyway think there will probably be a good jump in gfx althought not anything near PS2-PS3, but I think that most advances will be in others areas (physics, AI, animation, interfaces...).

Like I said some good time ago I think that reaching photo realism (as a overall goal) will be abadoned in the next gen (PS4...), or even sonner, as long as it can give a good final result that our brain see as close to what is hapening in the real world (eg GI, accurate shadows...), but devs will start to use gfx (or anything) to get a new kinds of looks (photo-surrealism) that while our brain will see this as a good 3D piece of work it will not be anything real. (eg imagine Killer 7 with GI, accurate shadows..., well the lightning is in part responsible by the game look but I guess you understand what I mean).

 

[Laa-Yosh]

It is not a cure all to rid the gaming world of graphic artists but it will be an invaluable tool to enable the artists to be much more productive in the quest to generate more lifelike and detailed worlds.

The thing is, we already use such tools on the content creation side, there's an entire software industry built on it Ingame implementations are a different case though, procedurally generating scenery and such isn't a simple thing and takes quite a lot of CPU power as well.

 

[TheChefO]

A big graphical leap might not be possible due to diminishing returns. A simple quadrupling of poly counts, textures, etc, will only add a marginal improvement. To get a big and noticeable improvement would need a lot more power.

I think though, that some realtime GI tech and very high IQ, with existing levels of graphics, would fit the bill okay, if that's not achieved this gen (which is where your idea of procedural content may fall down. Fingers crossed, that happens this gen, so there's little room for a big advancement in that field.)

I agree with your overall assessment that in order to get a percieved x improvement for x720/ps4 they would have to increase the hardware spec x10. (numbers obviously not important) This is why I think the ps4/x720 generation will be more about adding "life" to the game world than visual impact in a single screenshot. By "life" I mean things like concentrating heavily on physics, animation, ai, and subtle detail. These are the things that will give Sony and Microsoft a great "bang for the buck" in differentiating their new offerings in 2010.

Not to say they will invest nothing in the graphics department to improve their offerings. Things like improved aa (8x?) and hdr abilities should be expected but these features alone I don't think will be enough to sell new $400 console in 2010. In fact, I think they will have a pretty difficult time differentiating their new offerings in screenshots alone and will be heavily dependant on videos to showcase their wares as I imagine in 5 years devs will be producing some pretty compelling graphics on ps3/x360.

This is why I think they may not ramp linear in transistor count and may budget a bit more (percentage wise) for ram than they did this gen. A tiered system would make sense with 8gb of (inexpensive) 300mhz DDR, 512mb 1000mhz gddr and a smaller pool of embedded ram (32mb).

Any thoughts for embedding a ppu-like co-processor?

 

[Rangers]

A big graphical leap might not be possible due to diminishing returns. A simple quadrupling of poly counts, textures, etc, will only add a marginal improvement. To get a big and noticeable improvement would need a lot more power.

Sigh. This old myth...

Anyways, you'll see much greater than quadrupling of todays specs in the next gen, even if your statement was true...

Shoot, just look at the improvements in G80 alone. And remember how many years away from next gen we yet are! It's actually kind of scary.

If we would stay at one resolution, instead of tripling it, you'll also see bigger improvements (imagine what Xbox 360 could do at 640X480..) So that basically means, some depends on if 1080P is the next gen standard.

Even allowing for the huge resolution increase though, to me the current consoles easily demonstate a next gen leap (example, Gears on a SDTV).

 

[TheChefO]

The thing is, we already use such tools on the content creation side, there's an entire software industry built on it Ingame implementations are a different case though, procedurally generating scenery and such isn't a simple thing and takes quite a lot of CPU power as well.

Depends on how far one is looking to push procedural content. Also consider that both ps4 and x720 will likely both have dsp like functionality for rapidly ripping through procedurally generated content. Add to that the cost benefits of having hadware dynamically assist art/content creation and with this being the biggest budget eater for "blockuster" high budget games it makes sense that they would consider dedicating design resources to assist this aspect as much as possible and integrate these features into the artists workflow with pluggins to their design tools such as maya/max/photoshop.

I'm not sure if this is similar to what Carmack was/is doing with mega texture or not but something along the lines of describing a texture instead of it being a set of x/y pixels. For some things strict definition is needed but for others (most) a region or set of boundaries is necessary to describe a texture and it can be generated procedurally.

 

[TheChefO]

Sigh. This old myth...

Anyways, you'll see much greater than quadrupling of todays specs in the next gen, even if your statement was true...

Shoot, just look at the improvements in G80 alone. And remember how many years away from next gen we yet are! It's actually kind of scary.

If we would stay at one resolution, instead of tripling it, you'll also see bigger improvements (imagine what Xbox 360 could do at 640X480..) So that basically means, some depends on if 1080P is the next gen standard.

Even allowing for the huge resolution increase though, to me the current consoles easily demonstate a next gen leap (example, Gears on a SDTV).

True Gears is a great looking game but most of the launch titles on a sdtv look very close to xbox games. I think Shifty is spot on and screenshots I imagine will be very difficult to differentiate between old and new in 2010.

 

[Rangers]

True Gears is a great looking game but most of the launch titles on a sdtv look very close to xbox games. I think Shifty is spot on and screenshots I imagine will be very difficult to differentiate between old and new in 2010.

I think this is false.

If you ever actually check out screens (even better, video) of Xbox 360 games tabbed by the media to look like their Xbox counterpart, you will find there is usually in fact a vast difference. You can try with say, Madden. It was quite eye opening.

In other words, they lie.

Look at it this way, if nothing else, do you think it's really possible to engineer similar looking games on a box with 512 RAM versus 64, even if all else was equal, unless you tried really hard?

The 512 box is going to have massively better qaulity textures, at the least.

I tried to pick two similar gameplay pics of madden, xbox then 360

http://media.xbox.ign.com/media/811/811805/img_3781772.html

http://media.xbox360.ign.com/media/811/811804/img_3761548.html

I happen to think it's a poor picture of 360 madden, but you start to get an idea of the massive upgrades we are talking about.

Not picking on Madden, I'll wager I can do the same comparo on most any game widely tabbed to look "just like Xbox version" on 360.