-
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.
Graphics are no big deal, claims Sony UK boss...AI is the future for games.
- Thread starter jko
- Start date
StingerNLG
Member
So wait.....
Resistance is really 22 gigs worth of AI?? AWESOME!!! 
From what I understand, isn't the Xbox 360's multi threaded cpu better for AI than the Cell? I would have thought Cell is better for things such as physics and HDTV processing, where as the Xbox 360 cpu is better for AI and other out of order operations. Correct me if I'm totally wrong though.
Zombie James
Banned
Martoo said:
The future is to improve on everything, why should AI be exempt?
xaosslug
Member
I agree. AI and Physics are what I'm looking forward to the most this gen, as well as my graphics.
Judging by Sony's 1st and 2nd party offerings alone, it seems that's what Sony's pushing for. From what I've read/seen of Resistance, LAIR, Heavenly Sword, and MotorStorm AI/physics factor heavily into those games.
Judging by Sony's 1st and 2nd party offerings alone, it seems that's what Sony's pushing for. From what I've read/seen of Resistance, LAIR, Heavenly Sword, and MotorStorm AI/physics factor heavily into those games.
giant_frying_pan
Member
Could have fooled me!
Moderation Unlimited
Banned
_leech_ said:
QFMFT
GoldenEye 007
Member
To be fair, Wii does have an overall processing advantage over GameCube to the tune of 2-4 times, right? So, if a developer actually tries to tap into that added power, you may see some nice results. Not on the X360/PS3 level, but a jump nonetheless.Dragona Akehi said:
i'm by no means a techie, so i'm not even sure what allthis means, but a poster at gonintendo.com posted this:GoldenEye 007 said:
http://gonintendo.com/?p=5035#comment-76309
Logan Cano
Banned
jko said:
Fanboy BS.
GoldenEye 007
Member
jko, really didn't want to get into that type of speculation at all. Way too many variables for that. I wanted to keep it very simple in that the 2-4x comment would represent 2-4 times the possible work in the same amount of time. Pretty much that much more number-crunching avaliable over GameCube. If a developer actually takes time to learn the Wii architecture, they should be able to get a noticable bump in not only graphics but also physics and AI and other gameplay applications.
I've said it before, but I think a big reason why a lot of Wii games look underwhelming(to some people) and also don't seem to be very complex in other areas(to some people) is partly because few developers took the time to learn GameCube's architecture. If they half-assed GameCube development, what makes them suddenly become an expert on the Wii front? Few are qualified to take proper advantage of the system to allow for that added AI, physics, sound, and graphics. I'd imagine Square, Capcom, and Nintendo are among the few that get GameCube development. The rest are stuck playing catch-up.
And another part is the fact that Nintendo isn't exactly pushing this ultra-AI intensive gameplay. Still, that shouldn't stop a developer from trying such a thing out.
I've said it before, but I think a big reason why a lot of Wii games look underwhelming(to some people) and also don't seem to be very complex in other areas(to some people) is partly because few developers took the time to learn GameCube's architecture. If they half-assed GameCube development, what makes them suddenly become an expert on the Wii front? Few are qualified to take proper advantage of the system to allow for that added AI, physics, sound, and graphics. I'd imagine Square, Capcom, and Nintendo are among the few that get GameCube development. The rest are stuck playing catch-up.
And another part is the fact that Nintendo isn't exactly pushing this ultra-AI intensive gameplay. Still, that shouldn't stop a developer from trying such a thing out.
Deku said:How advanced. I also think PS3 would be more successful if it could handle the fun of the Wii
I suppose you have a point, if your definition of 'fun' involves being forced to flail your arms around like you're having a seizure in order to play games. (Even then, the footage of the Warhawk demonstration proves that you can get your spazzing-out fix with the PS3 too, if you're so inclined.
)(And seriously, wtf? 'The PS3 would be more successful if it could handle the fun of the Wii'? Now Nintendo has a monopoly on fun? Games can't be fun without a motion sensor? Give it a rest, Deku.)
DCharlie
And even i am moderately surprised
and every Wii game is going to be exactly like swimmng with the controller? enough already, all the anti-Wii statements are sounding similar to the anti-DS crap we had to put up with.
why not just wait and see?
sonys late inclussion of the motion sensors makes me think that Sony think Nintendo might be on to something!
DCharlie said:
Hey, I wasn't the one snidely implying that Wii games are somehow going to be inherently 'more fun' than PS3 titles. And since the only real thing I can think of that the Wii could (arguably) bring to the table that the PS3 can't is the magical motion-sensing wand, that (and the whole gesture-recognition, 'pantomime this action to trigger something happening in the game' business associated with it) seems to be where he's suggesting the 'fun' is. I think that's pretty absurd. Then again, I also think it's absurd how many Nintendo fans suddenly 'realized' that they were 'getting bored with gaming' and 'desperately needed something new to revitalize their interest' after Iwata announced the new party line, too. I guess a more 'normal' attitude would be for me to lose all interest in playing games with a conventional controller now that I can wave around a wand or scribble with a stylus instead?
DCharlie said:sonys late inclussion of the motion sensors makes me think that Sony think Nintendo might be on to something!
Yes, the desire to make it possible for 3rd party devs to come up with multiplatform titles that make token use of motion sensing, so they can port them to both the PS3 and Wii with the same mechanics intact. *shrug*
Spider_Jerusalem
Member
I agree. I think games are much more stunning when showing great IA/animations/physics than only raw brute graphics power.
However, this statement sounded a little bit like someone who discovered that gfx isn't a field in which they can show their superiority and are now looking at other fields to pimp
However, this statement sounded a little bit like someone who discovered that gfx isn't a field in which they can show their superiority and are now looking at other fields to pimp
TAJ
Darkness cannot drive out darkness; only light can do that. Hate cannot drive out hate; only love can do that.
>>>720p without AA looks a *little* better than 480p.<<<
ROFLMFAO... Oh, my sides...
>>>But, 480p with good AA and all effects on, actually looks better than a game where a lot of the power has gone towards achieving the high HD resolution.<<<
A lot of variables to that one, but if he means 480p on Wii vs. 720p on 360, he's delusional.
>>>720p takes up 3x the bandwidth, <<<
True, but on 360, that comes out of a 256GBps pool, so it's really irrelevant.
>>>3x the memory <<<
Again, dedicated frame buffer. Try again, tard. And even if Xenos didn't have that, I think that extra FEW HUNDRED MBs of RAM would come in useful there, just maybe.
>>>and 3x the fillrate of 480p, but it doesnt look 3x better at all.
Even if Hollywood (the Wii VPU) is only half the overall speed of xenos, it would still be faster at filling its smaller resolution with similar quality pixels.<<<
Sigh. Even if I could get past the part about Wii having half or more the raw fillrate of 360, I'd have to remind this shit that if big theoretical fillrate numbers were all that mattered, PS2's GS would have stomped Xbox's NV2A. DX10-level hardware anyone?
ROFLMFAO... Oh, my sides...
>>>But, 480p with good AA and all effects on, actually looks better than a game where a lot of the power has gone towards achieving the high HD resolution.<<<
A lot of variables to that one, but if he means 480p on Wii vs. 720p on 360, he's delusional.
>>>720p takes up 3x the bandwidth, <<<
True, but on 360, that comes out of a 256GBps pool, so it's really irrelevant.
>>>3x the memory <<<
Again, dedicated frame buffer. Try again, tard. And even if Xenos didn't have that, I think that extra FEW HUNDRED MBs of RAM would come in useful there, just maybe.
>>>and 3x the fillrate of 480p, but it doesnt look 3x better at all.
Even if Hollywood (the Wii VPU) is only half the overall speed of xenos, it would still be faster at filling its smaller resolution with similar quality pixels.<<<
Sigh. Even if I could get past the part about Wii having half or more the raw fillrate of 360, I'd have to remind this shit that if big theoretical fillrate numbers were all that mattered, PS2's GS would have stomped Xbox's NV2A. DX10-level hardware anyone?
Spider_Jerusalem said:However, this statement sounded a little bit like someone who discovered that gfx isn't a field in which they can show their superiority and are now looking at other fields to pimp
The same noises are coming out of certain people who traditionally are very graphics-centric, at least with regard to the technical gains that can be made going forward versus other things..
And it's something Sony has been saying since day one with PS3, the importance of what's going on behind the graphics, what's driving the graphics. It's not to say graphics is a completely solved problem, but there are now IMO larger challenges to face elsewhere.
ShinNL
Member
I doubt the boss has seen any line of code (like some gamers who try to argue as techs). While processing power is nice to have, most of the difficulty in AI lies in programming and the intelligence of the programmer himself/herself.
To make a dog react to 50 different things is easy.
To make a dog as intelligent as a dog is hard.
The latter requires less processing power, but a very very intelligent programmer.
Arguing about power = more physics = fine with me.
Arguing about power = better AI = go home and study first.
To make a dog react to 50 different things is easy.
To make a dog as intelligent as a dog is hard.
The latter requires less processing power, but a very very intelligent programmer.
Arguing about power = more physics = fine with me.
Arguing about power = better AI = go home and study first.
Soneet said:
This is true. But I think he's implying that..when he says "We are no longer interested in graphics per se, because graphics chips can do that for us", I think that's a recognition of the programming challenge shifting elsewhere. Graphics has a fairly well defined future path, A.I. certainly does not. He's touting Cell here because all else being equal, more power is better, as it places fewer limits on the potential solutions that will be attempted going forward. The effectiveness of those solutions is very much dependent on the intelligence of the approach and their design, but you want hardware to be as small a potential limiter as possible. But I think what he's talking about here isn't simply A.I. but a range of issues that contribute to what he wants to see achieved, which he's just labelling in a broad sense as "artificial intelligence".
ShinNL
Member
Yeah, he just used an acronym totally the wrong way. I think he's trying to explain emotional facial expressions and social reactions.gofreak said:
However, I don't totally agree that's an improvement for every game. I still believe games shouldn't be about humans and much less about every day's life.
So in that sense, I don't believe he knows what true interactive art means. For example, I see more art in Okami and Zelda than in Shenmue and Grand Theft Auto.
I think AI is really an overused term that really doesn't apply much to games. Most games simply use patterns or events to trigger pre-set actions. I don't think we will see real AI until we move past binary. You can create the most complex branching logic but in the end the computer will choose one of the pre-programed options. It can't really improvise or form a new strategy beyond what it is told to do.
Soneet said:
I don't think advances in these areas would be the sole domain of games about humans and/or true-to-life games, anymore than advances in rendering are only applicable to photorealistic games. Fantastical games need to provide interaction that doesn't break believability (versus realism) as much as any other kind of game. And I think the way games are made needs to become more sophisticated as far as the interaction side is concerned this generation, and going forward (and that covers a lot of things - simulation, physics, AI, interface, game logic etc.).
Flo_Evans said:
Depends what you consider AI to be, which is still a matter of debate I guess. There's the end-goal of AI (which I'm not sure everyone is agreed upon yet, even) which some may think of when they hear "AI", and then there's simply the field of AI and the current state of it. I think much of game AI is probably very unsophisticated though. But there's no point in drawing from other more sophisticated techniques if you can't apply it in a compelling fashion in a game, I guess. Things need to step up though.
Flo_Evans said:
Well, there's a certain amount you have to give your AI to start with. There are techniques that allow an AI to compose and try new strategies etc. but they are all derived from the building blocks you give it. We could debate how different that is from how human intelligence doed it, though..the mechanisms are different, obviously, but do we start with nothing?
i agree.Flo_Evans said:
i don't think advanced a.i. is the future of gaming and disagree with this article's point.
i think it's just a way of trying to say 'don't worry investors, we're not just using better graphics to make more money we're going to use better a.i.', and i can't see it working.
it'll scratch the itch of some of the hardcore but you can't base an industry on this audience. even speaking from a hardcore perspective i question how good it will be in games. for example i'm a real tennis fan, i play it real life, regularly watch it and love playing tennis videogames but the reality is the best version i've ever played is final match tennis on the pc engine.
this 8 bit looking game has the best physics, the best opponents, the best a.i., the best feeling of actually playing tennis and it was made on 15 year old hardware. i've played just about every tennis game, including the top spin and virtua tennis series' and they're good, with glorious visuals, but when it comes to simulating the idea of playing tennis, they're nowhere near the quality of final match.
games simulate a.i., and imo the quality of this depends more on the talent of the development staff than the power of the hardware.
LiveFromKyoto
Member
Soneet said:
Let's all stop watching movies unless they're about animals or elves. It's more artistic that way. Humans and real life situations are so limited.
And he didn't use the acronym the wrong way, emotional reaction can be a part of an AI routine as well. And he's not just talking about PS3's raw processing power, but the advantages of having several threads available which can each be running their own AI routines, reacting off one another.
gofreak said:
well you need to give your NPC some skills of course! it seems some developers miss this and thier NPCs are unable to even navigate through thier own game world
You kind of have to consider game balance and design when talking about AI. It would be real easy to make an enemy that dodged all your bullets everytime while headshoting you in the blink of an eye. AI in games have to have exploitable flaws for you to defeat them.
What I want to see more of is NPCs reacting naturally. Hell even conversations with them. We have software than can recognize your voice (BLOOOO), but does it understand what it means? Typical RPG senario: you walk into a tavern and start talking to a wizard. What if instead of correctly navigating through a multiple choice menu, you actaully talked to the wizard and convinced him to join you (or paid him, or theatened him?). What if he hears word that his hometown is being attacked and he has to decide on his own to continue with you or go help his town - all based on your interactions with him.
But again it comes down to game design and how the designer wants to present the story. I am not a big GTA fan but you have to recognize that the future of games are 'sandboxes' GTA is a fun little world to cause mayham in, but the NPCs are like slot cars on a track. If you murder some woman in the street thier husband doesn't plot revenge and hunt you down. You are forgoten as soon as you turn the corner.
Kojima hinted at some of this for MGS4. You can make allies on the battlefield or more enemies. I still remember ins MGS1 walking through the snow and saying to myself "o neat look at my foot prints!" Then a sentry walked over and goes "huh? whos footprints are these?!" what if instead of sounding the alarm the sentry started to stalk you? what if instead of going in guns blazing he watched and waited and snuck up behind you while you were using your scope and broke your neck?
Come on, we are not talking about chess here, where you need as much CPU power as you can muster to do build a gigantic tree of possible moves. We are talking about action games where the enemies that can react in a smart way to human players, and all you need is just some improved algorithms. Developers could have done that with PS1. They just couldn't/didn't bother to come up with better algorithms, that's all.
linsivvi said:
There's far more to improving AI than 'coming up with better algorithms', and the gaming possibilities are far broader and more exciting than you make them sound. I'd eventually like to see AI in games evolve to the point where NPC's would be able to interact with the player (and one another) in meaningful ways, acting on personal goals and fears, and using natural language processing to have an actual dialogue with the player rather than just spewing canned lines on cue. Such AI-intensive games are still decades in the future, but developers are already starting to take baby steps in that direction today. I'm hoping that as processing power continues to grow, the ambitions of developers with respect to AI will grow along with it.
Flo_Evans said:well you need to give your NPC some skills of course! it seems some developers miss this and thier NPCs are unable to even navigate through thier own game world
You kind of have to consider game balance and design when talking about AI. It would be real easy to make an enemy that dodged all your bullets everytime while headshoting you in the blink of an eye. AI in games have to have exploitable flaws for you to defeat them.
What I want to see more of is NPCs reacting naturally. Hell even conversations with them. We have software than can recognize your voice (BLOOOO), but does it understand what it means? Typical RPG senario: you walk into a tavern and start talking to a wizard. What if instead of correctly navigating through a multiple choice menu, you actaully talked to the wizard and convinced him to join you (or paid him, or theatened him?). What if he hears word that his hometown is being attacked and he has to decide on his own to continue with you or go help his town - all based on your interactions with him.
I think Jarosh mentioned in another thread that voice recognition technology won't be up to par for what you're describing in the foreseeable future. Unfortunate.
Too bad the PS3 and, to a lesser extend X360, are far more good at building gigantic trees than doing complex branching computations...linsivvi said:
Cell is good for manipulating sequential datas (for graphs, videos, and some kinds of physics) but quite bad as soon as you have to do a lot of tests and conditional jumps. Several programmers have already said that those two "next-gen" consoles show very bad performances for those kind of tasks.
Even the developpers of Cell said that programmers and/or compilers would need to implement branch prediction directly in the source code to avoid big performance hits...