This post is turning into the exact thing that Sony would love: specs, specs and more specs. Nintendo will own as it has always done. For every 3-4 good game on either Sony or Microsoft's machine(s). Nintendo have one great game that kills it to pieces. The next-gen wars will break at least one of Sony/Microsoft but Nintendo may sell less units but definitely make more money.

Sometimes it is forgotten that most Nintendo console owners have the same console for years while Sony (in particular) owners will have replaced their original console at least once.

Nintendo came outta left field and owned E3, if the crowds hadn't followed, Sony or Microsoft woulda thought they won based on "stats" and tech specs. Guys it's called "games" for a reason. Fun rules!!!

@codec:

Your projects are not crap. Just remember me when the millions come rolling in! ;-)#

The explosions interfering with the gameplay is what I was talking about when i spoke aboput obscuring of vision, and then went on to blablabla about the difference between incidental effects on gameplay and design features, but it seems you werent listening to me at all! pls re-read my post, I'm anticipating what your reaction will be.

AI: i dont know what ut does but its definitely a lot more complex than quake. however it will still be implemented as a direct function of core game logic, no question. A fully autonomous client AI (as seen in chess programs) is NOT something that realtime gaming can expect to see anytime soon. Nonetheless stuff like fuzzy logic, genetic algorithms, neural networks and so on can still be applied in this context! witness black and white. However ont mention spore yet, it hasnt come out and games like this tend to be exaggerated.




@spikyn

the spec debate is basically just an attempt to clear the air on some of the smoke and mirrors that sony has been spraying into the air. Thank God they were caught scamming everyone with killzone or else we for never hear di end of di thing.

I personally have been excited about the wii-remote since it was announced (outside e3) last year. And its great to see that the public shares my enthusiasm! PLUS the wii is cheap enough for me to actuially buy! PLUS the games are on dvd so the alaba boys may be able to crack em this time, meaning i can actually play games on it! and finally, it plays gamecube games, so I can finally play all those great cube games i always wanted to play but couldnt justify the unreasonable expense! (for a console that has no pirated games. this is Naija, afterall. original games are too expensive because we dont have localised releases)

ok my bad. i guess i mixed things up. i tend to glance over posts instead of patiently reading them. my bad. but i think you're mixing up the definition of autonomous here:
an autonomous agent is a system situated within and part of and environment that senses that environment and acts on it, over time, in pursuit of its own agenda and so as to effect what it senses in the future.
FPS bots, characters in sports games are examples of autonomous agents.

oh yes, though i haven't played it, i hear black and white is really good. i dont think any game has used neural networks because debugging them is hell. their output is unpredictable just like real-life situations. unlike a FSM, where u know your agent is either running or still, either shooting or not. oh well,

anyway, thanks man. if millions (which i doubt :p) roll in, i go holla. u on msn or yahoo? enjoying the chat.

I'm going to be a bit of a pretentious dickhead here, so I apologize in advance!

I didnt say 'autonomous agent'. I said 'autonomous *client* agent'!

What I mean is a system that is seperate from the core game logic, sending commands exactly as a human player would.

For example, imagine if when coding your football game, you code it as a 2-player game with no 1-player functions, then code a seperate system which will PLAY THE GAME exactly as a human would, ie with 8 directions and a set of fire buttons. This system will need real intelligence, because it will be unable to cheat, and it will have to analyse the data sent to it via the screen or whatever channel you use and base its strategies on that.

As you know, the way you are going to write it is that the game engine itself will 'pretend' to play by moving the players around in a completely different way than the human. Its like in the simple 'chaser' game that everyone uses to learn game programming. The code for the human is something like

(routine a)
if input(up) {decrement(y)};
if input(down) {increment(y)};

etc


but for the computer it is

(routine b)
if (cpu_x < x ) {decrement(x)};
if (cpu_x > x) {increment(x)};

when it SHOULD be

(routine(c))
if (superduperAIroutine_says_go_left) {sendinput(left)};
if (superduperAIroutine_says_go_right) {sendinput(right)};

which would send messages to routine (a) above. This is what makes it a client AI. It is not part and parcel of the game logic as in routine b. But as you can see, superduperAIroutine would be a bitch to code. How does it know where the human is? It would need to take a dump of screen memory, analyze it and figure it out. Why bother when you can just reach into the variables themselves and compare, then directly change the position of the agents involved? ah but then, you are merely masquerading as a player. And THIS is how games are made today.

In streetfighter Guile has to pretend to charge his flashkicks. Imagine how hard it is for the prorammers o ensure that sopmehow, guile has been in a position where down is held for 1 second before executing the flashkicks! End result: a lot of instant flashkicks. cheating, shebi? call it a limitation of the AI

anyway the final point is that routine (a) and routine (c) can exist on seperate cores, but not routine (a)_ and routine (b) because routine b is part of the thread that includes responding to human input. For us to remove AI from the main thread we should remove it from core game logic.

Unless of course, the code we use for AI is so complex, we can afford to use the 'tic' method of calculation, and merely poll the AI thread from time to time for results that ill be used to decide what to do. If the results are not available in a particular tic, we do some generic action, otherwise we do what the AI says. This I believe will not be feasible until the AI routines become SO complex that treating them in this way will actually be a performance gain.




black and white supposedly uses a neural net to evolve the creature's behaviour over time, based on its experiences in the game world and its interactions with you. The results are supposedly unpredictable, also. I'm currently playing the game (the original) and really enjoying it!

well it all depends on program design and you can seperate routine (b). the answer is keeping the shared data centralized. let me try to use an example, (i'm not good at explaining things)

say you're chasing object A. A has it's x and y coordinates, oh Bleep it.

that tabletennis demo i did, right? i had up to 3 or 4 threads accessing and modifying the ball position. i had an AI thread that simply 'chased' the ball, querying it's position. i had a physics thread that did the necessary ball motion simulation updating it's position, i also had a networking thread that queried the ball for it's position.

i could give your the source if you want. i guess it'll be clearer in code.

black and white could've used a "rule-based AI system". i think that's what tekken and most fighting games use.

Possibly you multithreaded your tabletennis demo, BUT,

you have to justify it! would a single threaded approach not be faster?

the domain of solutions only includes those where for a given problem, the multithreaded solution is faster than the singlethreaded solution. If we want we can multithread hello world! But we are only concerned with situations where performance is actually enhanced. Many times multithreading causes horrible performance hits.

I imagine you are locking and semaphoring and mutexing like ker-razy to maintain sync in your program. Hope u know all these are proc cycles down the garbage chute! Cool tech demo but in practice, is it workable?

the rule-based thingy isnt what black and white uses. It specifically went out of its way to implement a learning neural network- at least thats what all their pr blahblahblah said. Of course I havent seen the code!

ANyway so you get what b&w is all about, its a god-game like populous except that it has a virtual pet thrown in. like all godgames, you indirectly rule your people, rather than giving direct orders like in civ or AoE, you impress them with miracles, help them with their harvest, and create disciples to create specialization and make your people more independent. The more people that blieve in you, the more prayer power you generate to workl nmore miracles and convert more villages to your cause.

Where the AI comes in is the creature. You have an animal that wanders around the countryside interacting with your people. It develops and grows by learning from its experiences. You can interact with it, rewarding it for certain actions and punishing it for others. You can also use a 'leash of learning' which forces it to watch you and learn from your behaviour. Based on your behaviour, the creature eventually develops a personality of its own, and will act independently of you. Do your job right and it becomes a powerful ally, because it has unlimited miracle power and any villagers it copnverts will belong to you. Do your job wrongly and it may even develop a personality that contradicts your own, sabotaging your efforts, or it may become lazy, stupid or otherwise useless.

Its an intersting concept! Be sure to check it out. And no, its not a smoke-and-mirrors thing like Fable's good vs evil noinsense which was just a lot of branching possibilities.

@darkmanzero

Reason #2: Metal Gear Solid 4: Guns of the Patriots
Remember when we said that the Xbox 360 doesn't have a killer app yet? Well the PlayStation 3's Metal Gear Solid 4 is definitely a killer app. Game creator Hideo Kojima has shown time and time again that he knows how to craft an excellent videogame, and after witnessing the impressive real-time demonstration of Guns of the Patriots (and its accompanying trailer) at TGS this past September, we're already convinced that the PS3 will be capable of some amazing things. A great pedigree plus great technology? Sounds good to us!

http://ps3.ign.com/articles/668/668446p3.html

about EUPHORIA:
April 19, 2006 - Since the release of the Xbox 360 and even more so with the approaching PS3 launch, there's been a hot debate regarding what qualifies as "next-gen." With all the added processing power, the next wave of consoles promises better graphics and enhanced sound, but what about gameplay? Though games get slicker production values, do they really play any differently than in the past? From what we've seen so far, the answer is no. We've seen a lot of products released for the Xbox 360 that don't differ much from what was possible on current gen, save for the bells and whistles.

But with its upcoming Indiana Jones title, LucasArts may be onto something. Utilizing a run-time animation technology called euphoria, the game is poised to push the limits of what was previously possible. The company that developed the technology, NaturalMotion, has been working for years to hone euphoria to accurately replicate real-world physics such as strength, weight, and momentum with in-game character models. Whereas before, all animations had to be pre-programmed, euphoria allows for reactions and behaviors to occur in real-time -- allowing models to react realistically to whatever situation might arise.

To be clear, euphoria is not an AI program, it's an animation technology that allows for physically accurate behaviors. The AI is programmed by the developer, in this case LucasArts, into the euphoria models to control them. What we saw at LucasArts was the marriage of euphoria physics and LucasArts A.I. scripts.


While it may sound good on paper, LucasArts was eager to show off the technology. We were shown two interactive technical demos highlighting the system in action with Chris Williams, Project Lead at LucasArts and Peter Hirschmann, Vice President of Product Development on hand. The first glimpse we got, the Tower Demo, showed off AI reactions to being dropped through a series of planks and beams. The second, which we'll call the "Fight Demo," featured Indiana Jones beating up an enemy A.I. to showcase varying reactions to getting punched and tossed around. Both showed potential for a new kind of gaming experience.

"For us, this is bleeding edge stuff," said Hirschmann. "It's what next-gen is about. It's not creating a linear path. It's creating intelligent characters, really interesting setups and environments, and putting the player in there and letting them see what the hell happens."

The Tower demo showed off what LucasArts was talking about. It gave the player the ability to toss in AI-controlled character models at will, letting them fend for themselves as they tumbled through a Price is Right "Plinko-style" structure. While initially, it seemed like a rag-doll physics demonstration, it soon became clear that the AI models were actively trying to stop their falls. Reacting to their surroundings without the restrictions of pre-programmed animations, the models reached out to whatever beams happened to be closest. Since they were falling at relatively significant speeds, they often missed, but would try again when they passed by the next plank or beam.

Eventually, as they slowed themselves with attempts to grab on, they'd be successful. However, the exercise didn't end there. Some prevailed in pulling themselves up to a standing position, but others lost their grip and continued to fall. The demo achieved an even greater level of complexity when several models were thrown into each other. A hanging model hit by another wouldn't just hang there stupidly while the other ricochets off. Instead, it'll actually reach out its hand to grab the other. In the same way, the tossed model will reach out its hand to try and grab onto the hanging one's leg. As further proof of the advanced AI at work, the hanging model would even reach down and try to pull the other up to safety.

To add yet more depth, fire hydrants and other obstacles were tossed at the models as they frantically tried to hang on. As expected, they could do little to stop from getting pummeled. Surprisingly, they didn't give up efforts to take the least amount of damage possible. Though some models were knocked cold after getting smacked, others fumbled around to break their fall. To stop themselves from tumbling, the models would actually stick their hands out in a last ditch cushioning attempt. Depending on the layout of the environment, they would react accordingly. For instance, when they got slammed into a wall, their body reacted with a convincing thud. Part of the body impacted first, and the rest of the limbs went flailing around, impacting surfaces like they actually existed.

"You don't have to animate 5,000 different permutations," said Hirschmann. "You author one behavior and then [the character model] stumbles back, he'll try to grab, if there's nothing there he'll fall over. With this kind of AI and these kind of behaviors it creates these situations where no two players' experiences are ever the same."
While we've all seen that before with rag doll models, those only reacted as such when they were "dead" -- In other words, when you pumped enough bullets into them or broke enough of their bones to make their AI program switch off. With these euphoria-enhanced models, the A.I. is still very much alive and kicking while the models get slammed into objects.

"In our minds, the age of the rag doll is over," said Williams. "Now we layer on top of the simulation this intelligence, this awareness, this ability to respond, react, get back up and grab onto things."

At one point, a model got knocked over while standing on a plank that angled downward, toward the ground. After falling, it began sliding forward with gravity. Since its A.I. was still fully functioning, it was clawing wildly around to try and stop its slide. Unfortunately for the model, the plank was flat, but as it slid out off the edge, it managed to catch itself on the very lip of the lath. It dangled precariously for a moment, but was thereafter able to hoist itself back up to shelter.

The Fight Demo yielded insight into more of euphoria's subtleties. Taking place in large square rooms littered with random geometric objects and a broken-down car, this demo showed off how Indy's attacks knocked around enemies. The sole foe in the stage started off by walking up towards Indy and was promptly met with fists to the face. After a few punches, Indy was able to grab the guy by the shoulders and toss him backwards, which made it more obvious what kinds of animation possibilities existed.


Instead of lapsing into some kind of pre-generated tumble, the enemy was at the mercy of every object in the environment. As he flew back, his body twisted ever so slightly so he could see where he was falling, and he'd try to brace for impact with his hands and shoulders as effectively as he could. Every environmental object he came into contact with was reacted to realistically. When getting thrown into a car, his body flattened against the side of the car. When getting tossed onto the edge of a wall, his reactions were different every time. Occasionally he would bounce and barrel roll to the ground, at others he would hit and lay out across top of the wall, and even come to rest angled from the wall to ground.

Again, he wasn't dead when this was happening, but rather still being driven by the adaptive animations that euphoria allows for. In every instance, the foe would get right to his feet, using his arms to push himself upright and come back at Indy. These unique reactions to in-game occurrences happened while getting punched as well. After Indy delivered a heavy smack, the opponent would go backpedaling backwards. While enemies in other games have pre-programmed reactions to punches, the euphoria enemies adjusted to their environment in real time. They would stagger, feet fumbling around differently every time to regain their balance. Any objects they happened to encounter while attempting to regain composure would affect them as one would expect it to in real life. If there was a low obstacle, they would trip. If there was a wall, they'd hit it and fall over. If there was an angled surface, they'd hit it and tumble off in whatever direction the physics demanded.

"The stumbling, the staggering, the notion that the characters really have this sense of situational awareness, and they have this sense of self preservation that really allows you to, given a circumstance, get a different reaction from them every time," said Williams.

More interesting possibilities were mentioned regarding team A.I. dynamics which we didn't get to see. "It's never two against one, it's two as one" said Hirschmann.

LucasArts mentioned groups of several enemies would be able to work together to overcome obstacles and more effectively attack you, leading to unique battles every time. They said they were hoping players would want to reload areas over and over again solely to see the different ways they could take out foes and watch them work together. In terms of specific examples, it was said enemies could pass weapons between each other and help each other up and over obstacles.

Hirschmann added, "We're working on a behavior right now where you can toss a guy into another guy and his head will actually track him and actually put his hands out to try and catch him."

While we saw a glimpse of this teamwork with the Tower Demo, we'll have to wait and see how it translates into actual gameplay.

Unfortunately we could not acquire media from any of the demos, so we've posted video from NaturalMotion's site that should give you a sense of what we're talking about. Note that this is not media from Indiana Jones, and actually from NaturalMotion's non-run-time product called endorphin. In it you'll be able to see how its technology manipulates a character model's physical reactions. Ideally, as this technology eventually spreads and development teams get more comfortable using it, we'll start to see strikingly more realistic A.I. behavior and reactions. While Indiana Jones won't release until next year, it's good to know new technologies are being pursued that could provide gamers with exciting, innovative experiences. We're hoping euphoria turns out to be one of them.

http://ps3.ign.com/articles/702/702423p2.html

i copied and pasted this cos of ur rant about how crap AI is and watnot. well heres a sample of nextgen AI. you are bein unnecessarily overcritical(one might even say cynical) about wat u have neither seen nor experienced. you base your arguments and "research" on the works of critics of the ps3 while i try to approach this from a common sense perspective with as little industry jargon as i can. now u said the MGS 4 trailer was not realtime: i remember folks sayin the same thing wen Sons of liberty was trailered and kojima had to re-issue ANOTHER trailer in which u can pan the camera durin cutscenes, suddenly all those who previously said "CGI" were now saying "woah!". and the truth is neither you, me, codec, john carmack nor even the folks at microsoft have seen the ps3 so on wat grounds are we arguing? and the pdf doc u referred me to didnt bolster ur reasonin, infact it serves to undermine ur position. yes that same article u gleefully referred to as written not by any games body but genuine techs. well, the article DID confirm the CELL is a multithreadable parallel processor that can do much for games. "The SPU units are designed for aggregation into an
array of processing elements. The programming model for
the SPU has not been predetermined, and the[b] BPA can support
both process pipelining and parallel processing.[/b] In a
pipeline implementation, a transformation on a data set performs
a specific task at each SPU and then passes the intermediate
results to the next SPU. The advantage of this technique
is that the code in each SPU is usually quite small and
the operations are easy to manage. The pipeline is also more
predictable. This predictability is important to the design
goals of the Cell processor and is one of the reasons the SPEs
have local memory and not cache memory. The disadvantage
is that it’s often difficult to make each stage equal in complexity
and time." this is quoted from the same article. and regarding your notion that the spe unique memory allocation is a disadvantage, well the article further says that in addition to havin individual memory (as well as a native l2 cache) " the memory model of the Cell processor supports the sharing of memory locations".
also the final fantasy demo (realtime also, not CGI) was crafted with the CEL alone independent of the GPU. also, u said the cell works on video encoding and has nothin to offer grafx, well accordin to ur article "The processing capability of the Cell processor is still being
explored. One demo that IBM has showed was a detailed 3D
contour map with satellite images imposed on the geography.
The Cell processor can render the ray-cast graphics at around
an order of magnitude faster than a contemporary PC processor.
The performance of only the 4GHz Power core without the
SPEs is harder to compare with the slower 2.5GHz G5 processor,
because of the Cell core’s simpler microarchitecture. The
higher clock speed of the Cell processor can still process many
math operations faster than the G5 can, so we would give the
performance edge to the Cell processor". in concluding the article, the author said "this chip is a quantum leap over the Emotion Engine and
with the parallel floating-point performance is a virtual
supercomputer on a chip.
Despite the excessive hype surrounding the Cell
processor at ISSCC, the possibilities of processing this much
media and graphics data are still being explored. The potential
impact is very exciting if the right programming model
can be found to utilize the capabilities. We have only
scratched the surface of the chip and its potential. Luckily,
we can count on an army of game programmers to dig deep
into this chip, and we’ll hopefully be able to learn from them
how much capability can be exploited. The next challenge
for the STI Design Center (after building the chip and systems)
will be to find a way to make this architecture accessible
to programmers beyond the aforementioned gamer
developers.
Editor’s Note: The Cell architecture deserves more exploration
and MPR will have additional coverage on it in the near future,
including more details on the architecture and programming model issues."

Sony, Toshiba and IBM worked together to develop the Cell processor. It's their answer to the growing trend toward multi-core processing, in which manufacturers place as many processors as possible onto one chip. The Cell processor is scalable for different performance needs. The one used in the PlayStation 3 crams 234 million transistors onto a single die. For comparison, one of the most powerful desktop PC CPUs available in 2005, the $1,000, dual-core Pentium Processor Extreme Edition, just barely breaks the 200-million-transistor mark.



The setup of the Cell processor is like having a team of processors all working together on one chip to handle the large computational workload needed to run next-generation video games. In order to understand how the Cell processor works, it helps to look at each of the major parts that comprise this processor.

The "Processing Element" of the Cell is a 3.2-GHz PowerPC core equipped with 512 KB of L2 cache. The PowerPC core is a type of microprocessor similar to the one you would find running the Apple G5. It's a powerful processor on its own and could easily run a computer by itself; but in the Cell, the PowerPC core is not the sole processor. Instead, it's more of a "managing processor." It delegates processing to the eight other processors on the chip, the Synergistic Processing Elements.

The computational workload comes in through the PowerPC core. The core then assesses the work that needs to be done, looks at what the SPEs are currently processing and decides how to best dole out the workload to achieve maximum efficiency.


The SPEs used in the Cell processor are each SIMD (Single Instruction, Multiple Data), 128-bit vector processors. Vector processors are designed to quickly process several pieces of data at once. They were commonly used in the 1980s in large, powerful, scientific supercomputers and were created as a faster alternative to the more common scalar processor. Scalar processors can only work one data element at a time. Despite this limitation, advances in scalar design and performance have made the use of vector processors very rare these days in most computers. However, because of the vector processor's ability to handle several data elements at once, IBM resurrected this design for the Cell. There are eight SPEs on the chip, but only seven of them handle processing. The eighth SPE is built in as redundancy in case one of the other seven fails.

The SPEs each come loaded with 256 KB SRAM. This high-speed memory helps each SPE crunch numbers quickly. The SPE memory is also visible to the main Processing Element. This allows the PowerPC Core to utilize the resources of each SPE in the most efficient way possible. All of this amounts to unprecedented power for a piece of consumer electronics.

Because graphics are so important to computers (and especially computers designed to play video games), there are microprocessors dedicated only to creating and displaying computer graphics. This processor is called the Graphic Processing Unit (GPU). One of the most anticipated aspects of the PlayStation 3 is the new GPU that was created for it -- the RSX "Reality Synthesizer."



Sony designed the RSX with graphics-card manufacturer Nvidia. The RSX is based on Nvidia's GeForce graphics technology. It's a 550-MHz, 300-million-transistor graphics chip. To put that in perspective, according to this Nvidia press release, the number of transistors on the RSX is "more than the total number of transistors in both the central processing units and the graphics processing units of the three leading current-generation systems, combined."

Unlike the GPU in the Xbox 360, the RSX is built on the traditional independent vertex/pixel shader architecture. Shaders are computer programs that determine the final look of what you see on the screen when you're looking at computer animation.



All of this translated to a level of graphic detail never before seen on a video-game console.

now this is definitely my last word on this ps3/xbox360 wahala cos u guys insist on wading into the murky territory of tech and obscure terms that mean little to majority of gamers. now though i can only claim a rather rimentary understanding of underlying CPU and semiconductor eevices architecture, i do know theres no way in hell darkmanzero can say all the power of the ps3 (yes i admit the cell is a litle overhyped, as they did with the Emotion Engine, but EE has so many shortcomings which sony has definitely learned from) is tripe and that the xbox stands to be potentially better. i still maintain that we do away wit all this speculation and instead think of how we can afford the darn thing! but one thing i do know: ps3 IS parallel-processing, multithreadable, and has better memory architecture and is multicore, so there!

@darkmanzero

Reason #2: Metal Gear Solid 4: Guns of the Patriots
Remember when we said that the Xbox 360 doesn't have a killer app yet? Well the PlayStation 3's Metal Gear Solid 4 is definitely a killer app. Game creator Hideo Kojima has shown time and time again that he knows how to craft an excellent videogame, and after witnessing the impressive real-time demonstration of Guns of the Patriots (and its accompanying trailer) at TGS this past September, we're already convinced that the PS3 will be capable of some amazing things. A great pedigree plus great technology? Sounds good to us!

http://ps3.ign.com/articles/668/668446p3.html

about EUPHORIA:
April 19, 2006 - Since the release of the Xbox 360 and even more so with the approaching PS3 launch, there's been a hot debate regarding what qualifies as "next-gen." With all the added processing power, the next wave of consoles promises better graphics and enhanced sound, but what about gameplay? Though games get slicker production values, do they really play any differently than in the past? From what we've seen so far, the answer is no. We've seen a lot of products released for the Xbox 360 that don't differ much from what was possible on current gen, save for the bells and whistles.

But with its upcoming Indiana Jones title, LucasArts may be onto something. Utilizing a run-time animation technology called euphoria, the game is poised to push the limits of what was previously possible. The company that developed the technology, NaturalMotion, has been working for years to hone euphoria to accurately replicate real-world physics such as strength, weight, and momentum with in-game character models. Whereas before, all animations had to be pre-programmed, euphoria allows for reactions and behaviors to occur in real-time -- allowing models to react realistically to whatever situation might arise.

To be clear, euphoria is not an AI program, it's an animation technology that allows for physically accurate behaviors. The AI is programmed by the developer, in this case LucasArts, into the euphoria models to control them. What we saw at LucasArts was the marriage of euphoria physics and LucasArts A.I. scripts.


While it may sound good on paper, LucasArts was eager to show off the technology. We were shown two interactive technical demos highlighting the system in action with Chris Williams, Project Lead at LucasArts and Peter Hirschmann, Vice President of Product Development on hand. The first glimpse we got, the Tower Demo, showed off AI reactions to being dropped through a series of planks and beams. The second, which we'll call the "Fight Demo," featured Indiana Jones beating up an enemy A.I. to showcase varying reactions to getting punched and tossed around. Both showed potential for a new kind of gaming experience.

"For us, this is bleeding edge stuff," said Hirschmann. "It's what next-gen is about. It's not creating a linear path. It's creating intelligent characters, really interesting setups and environments, and putting the player in there and letting them see what the hell happens."

The Tower demo showed off what LucasArts was talking about. It gave the player the ability to toss in AI-controlled character models at will, letting them fend for themselves as they tumbled through a Price is Right "Plinko-style" structure. While initially, it seemed like a rag-doll physics demonstration, it soon became clear that the AI models were actively trying to stop their falls. Reacting to their surroundings without the restrictions of pre-programmed animations, the models reached out to whatever beams happened to be closest. Since they were falling at relatively significant speeds, they often missed, but would try again when they passed by the next plank or beam.

Eventually, as they slowed themselves with attempts to grab on, they'd be successful. However, the exercise didn't end there. Some prevailed in pulling themselves up to a standing position, but others lost their grip and continued to fall. The demo achieved an even greater level of complexity when several models were thrown into each other. A hanging model hit by another wouldn't just hang there stupidly while the other ricochets off. Instead, it'll actually reach out its hand to grab the other. In the same way, the tossed model will reach out its hand to try and grab onto the hanging one's leg. As further proof of the advanced AI at work, the hanging model would even reach down and try to pull the other up to safety.

To add yet more depth, fire hydrants and other obstacles were tossed at the models as they frantically tried to hang on. As expected, they could do little to stop from getting pummeled. Surprisingly, they didn't give up efforts to take the least amount of damage possible. Though some models were knocked cold after getting smacked, others fumbled around to break their fall. To stop themselves from tumbling, the models would actually stick their hands out in a last ditch cushioning attempt. Depending on the layout of the environment, they would react accordingly. For instance, when they got slammed into a wall, their body reacted with a convincing thud. Part of the body impacted first, and the rest of the limbs went flailing around, impacting surfaces like they actually existed.

"You don't have to animate 5,000 different permutations," said Hirschmann. "You author one behavior and then [the character model] stumbles back, he'll try to grab, if there's nothing there he'll fall over. With this kind of AI and these kind of behaviors it creates these situations where no two players' experiences are ever the same."
While we've all seen that before with rag doll models, those only reacted as such when they were "dead" -- In other words, when you pumped enough bullets into them or broke enough of their bones to make their AI program switch off. With these euphoria-enhanced models, the A.I. is still very much alive and kicking while the models get slammed into objects.

"In our minds, the age of the rag doll is over," said Williams. "Now we layer on top of the simulation this intelligence, this awareness, this ability to respond, react, get back up and grab onto things."

At one point, a model got knocked over while standing on a plank that angled downward, toward the ground. After falling, it began sliding forward with gravity. Since its A.I. was still fully functioning, it was clawing wildly around to try and stop its slide. Unfortunately for the model, the plank was flat, but as it slid out off the edge, it managed to catch itself on the very lip of the lath. It dangled precariously for a moment, but was thereafter able to hoist itself back up to shelter.

The Fight Demo yielded insight into more of euphoria's subtleties. Taking place in large square rooms littered with random geometric objects and a broken-down car, this demo showed off how Indy's attacks knocked around enemies. The sole foe in the stage started off by walking up towards Indy and was promptly met with fists to the face. After a few punches, Indy was able to grab the guy by the shoulders and toss him backwards, which made it more obvious what kinds of animation possibilities existed.


Instead of lapsing into some kind of pre-generated tumble, the enemy was at the mercy of every object in the environment. As he flew back, his body twisted ever so slightly so he could see where he was falling, and he'd try to brace for impact with his hands and shoulders as effectively as he could. Every environmental object he came into contact with was reacted to realistically. When getting thrown into a car, his body flattened against the side of the car. When getting tossed onto the edge of a wall, his reactions were different every time. Occasionally he would bounce and barrel roll to the ground, at others he would hit and lay out across top of the wall, and even come to rest angled from the wall to ground.

Again, he wasn't dead when this was happening, but rather still being driven by the adaptive animations that euphoria allows for. In every instance, the foe would get right to his feet, using his arms to push himself upright and come back at Indy. These unique reactions to in-game occurrences happened while getting punched as well. After Indy delivered a heavy smack, the opponent would go backpedaling backwards. While enemies in other games have pre-programmed reactions to punches, the euphoria enemies adjusted to their environment in real time. They would stagger, feet fumbling around differently every time to regain their balance. Any objects they happened to encounter while attempting to regain composure would affect them as one would expect it to in real life. If there was a low obstacle, they would trip. If there was a wall, they'd hit it and fall over. If there was an angled surface, they'd hit it and tumble off in whatever direction the physics demanded.

"The stumbling, the staggering, the notion that the characters really have this sense of situational awareness, and they have this sense of self preservation that really allows you to, given a circumstance, get a different reaction from them every time," said Williams.

More interesting possibilities were mentioned regarding team A.I. dynamics which we didn't get to see. "It's never two against one, it's two as one" said Hirschmann.

LucasArts mentioned groups of several enemies would be able to work together to overcome obstacles and more effectively attack you, leading to unique battles every time. They said they were hoping players would want to reload areas over and over again solely to see the different ways they could take out foes and watch them work together. In terms of specific examples, it was said enemies could pass weapons between each other and help each other up and over obstacles.

Hirschmann added, "We're working on a behavior right now where you can toss a guy into another guy and his head will actually track him and actually put his hands out to try and catch him."

While we saw a glimpse of this teamwork with the Tower Demo, we'll have to wait and see how it translates into actual gameplay.

Unfortunately we could not acquire media from any of the demos, so we've posted video from NaturalMotion's site that should give you a sense of what we're talking about. Note that this is not media from Indiana Jones, and actually from NaturalMotion's non-run-time product called endorphin. In it you'll be able to see how its technology manipulates a character model's physical reactions. Ideally, as this technology eventually spreads and development teams get more comfortable using it, we'll start to see strikingly more realistic A.I. behavior and reactions. While Indiana Jones won't release until next year, it's good to know new technologies are being pursued that could provide gamers with exciting, innovative experiences. We're hoping euphoria turns out to be one of them.

http://ps3.ign.com/articles/702/702423p2.html

i copied and pasted this cos of ur rant about how crap AI is and watnot. well heres a sample of nextgen AI. you are bein unnecessarily overcritical(one might even say cynical) about wat u have neither seen nor experienced. you base your arguments and "research" on the works of critics of the ps3 while i try to approach this from a common sense perspective with as little industry jargon as i can. now u said the MGS 4 trailer was not realtime: i remember folks sayin the same thing wen Sons of liberty was trailered and kojima had to re-issue ANOTHER trailer in which u can pan the camera durin cutscenes, suddenly all those who previously said "CGI" were now saying "woah!". and the truth is neither you, me, codec, john carmack nor even the folks at microsoft have seen the ps3 so on wat grounds are we arguing? and the pdf doc u referred me to didnt bolster ur reasonin, infact it serves to undermine ur position. yes that same article u gleefully referred to as written not by any games body but genuine techs. well, the article DID confirm the CELL is a multithreadable parallel processor that can do much for games. "The SPU units are designed for aggregation into an
array of processing elements. The programming model for
the SPU has not been predetermined, and the[b] BPA can support
both process pipelining and parallel processing.[/b] In a
pipeline implementation, a transformation on a data set performs
a specific task at each SPU and then passes the intermediate
results to the next SPU. The advantage of this technique
is that the code in each SPU is usually quite small and
the operations are easy to manage. The pipeline is also more
predictable. This predictability is important to the design
goals of the Cell processor and is one of the reasons the SPEs
have local memory and not cache memory. The disadvantage
is that it’s often difficult to make each stage equal in complexity
and time."  this is quoted from the same article. and regarding your notion that the spe unique memory allocation is a disadvantage, well the article further says that in addition to havin individual memory (as well as a native l2 cache) " the memory model of the Cell processor supports the sharing of memory locations".
also the final fantasy demo (realtime also, not CGI) was crafted with the CEL alone independent of the GPU. also, u said the cell works on video encoding and has nothin to offer grafx, well accordin to ur article "The processing capability of the Cell processor is still being
explored. One demo that IBM has showed was a detailed 3D
contour map with satellite images imposed on the geography.
The Cell processor can render the ray-cast graphics at around
an order of magnitude faster than a contemporary PC processor.
The performance of only the 4GHz Power core without the
SPEs is harder to compare with the slower 2.5GHz G5 processor,
because of the Cell core’s simpler microarchitecture. The
higher clock speed of the Cell processor can still process many
math operations faster than the G5 can, so we would give the
performance edge to the Cell processor". in concluding the article, the author said "this chip is a quantum leap over the Emotion Engine and
with the parallel floating-point performance is a virtual
supercomputer on a chip.
Despite the excessive hype surrounding the Cell
processor at ISSCC, the possibilities of processing this much
media and graphics data are still being explored. The potential
impact is very exciting if the right programming model
can be found to utilize the capabilities. We have only
scratched the surface of the chip and its potential. Luckily,
we can count on an army of game programmers to dig deep
into this chip, and we’ll hopefully be able to learn from them
how much capability can be exploited. The next challenge
for the STI Design Center (after building the chip and systems)
will be to find a way to make this architecture accessible
to programmers beyond the aforementioned gamer
developers.
Editor’s Note: The Cell architecture deserves more exploration
and MPR will have additional coverage on it in the near future,
including more details on the architecture and programming model issues."

Sony, Toshiba and IBM worked together to develop the Cell processor. It's their answer to the growing trend toward multi-core processing, in which manufacturers place as many processors as possible onto one chip. The Cell processor is scalable for different performance needs. The one used in the PlayStation 3 crams 234 million transistors onto a single die. For comparison, one of the most powerful desktop PC CPUs available in 2005, the $1,000, dual-core Pentium Processor Extreme Edition, just barely breaks the 200-million-transistor mark.



The setup of the Cell processor is like having a team of processors all working together on one chip to handle the large computational workload needed to run next-generation video games. In order to understand how the Cell processor works, it helps to look at each of the major parts that comprise this processor.

The "Processing Element" of the Cell is a 3.2-GHz PowerPC core equipped with 512 KB of L2 cache. The PowerPC core is a type of microprocessor similar to the one you would find running the Apple G5. It's a powerful processor on its own and could easily run a computer by itself; but in the Cell, the PowerPC core is not the sole processor. Instead, it's more of a "managing processor." It delegates processing to the eight other processors on the chip, the Synergistic Processing Elements.

The computational workload comes in through the PowerPC core. The core then assesses the work that needs to be done, looks at what the SPEs are currently processing and decides how to best dole out the workload to achieve maximum efficiency.


The SPEs used in the Cell processor are each SIMD (Single Instruction, Multiple Data), 128-bit vector processors. Vector processors are designed to quickly process several pieces of data at once. They were commonly used in the 1980s in large, powerful, scientific supercomputers and were created as a faster alternative to the more common scalar processor. Scalar processors can only work one data element at a time. Despite this limitation, advances in scalar design and performance have made the use of vector processors very rare these days in most computers. However, because of the vector processor's ability to handle several data elements at once, IBM resurrected this design for the Cell. There are eight SPEs on the chip, but only seven of them handle processing. The eighth SPE is built in as redundancy in case one of the other seven fails.

The SPEs each come loaded with 256 KB SRAM. This high-speed memory helps each SPE crunch numbers quickly. The SPE memory is also visible to the main Processing Element. This allows the PowerPC Core to utilize the resources of each SPE in the most efficient way possible. All of this amounts to unprecedented power for a piece of consumer electronics.

Because graphics are so important to computers (and especially computers designed to play video games), there are microprocessors dedicated only to creating and displaying computer graphics. This processor is called the Graphic Processing Unit (GPU). One of the most anticipated aspects of the PlayStation 3 is the new GPU that was created for it -- the RSX "Reality Synthesizer."



Sony designed the RSX with graphics-card manufacturer Nvidia. The RSX is based on Nvidia's GeForce graphics technology. It's a 550-MHz, 300-million-transistor graphics chip. To put that in perspective, according to this Nvidia press release, the number of transistors on the RSX is "more than the total number of transistors in both the central processing units and the graphics processing units of the three leading current-generation systems, combined."

Unlike the GPU in the Xbox 360, the RSX is built on the traditional independent vertex/pixel shader architecture. Shaders are computer programs that determine the final look of what you see on the screen when you're looking at computer animation.



All of this translated to a level of graphic detail never before seen on a video-game console.

now this is definitely my last word on this ps3/xbox360 wahala cos u guys insist on wading into the murky territory of tech and obscure terms that mean little to majority of gamers. now though i can only claim a rather rudimentary understanding of underlying CPU and semiconductor eevices architecture, i do know theres no way in hell darkmanzero can say all the power of the ps3 (yes i admit the cell is a litle overhyped, as they did with the Emotion Engine, but EE has so many shortcomings which sony has definitely learned from) is tripe and that the xbox stands to be potentially better. i still maintain that we do away wit all this speculation and instead think of how we can afford the darn thing! but one thing i do know: ps3 IS parallel-processing, multithreadable, and  has better memory architecture and is multicore, so there!

@dakmanzero:

my demo does perform better in multithreaded mode than single threaded. i tried it. these subsystems are updated at a slower rate than the graphics. for instance, the physics only updates once every 40 millisecond.
techniques like these are what will be applied to bigger games. there's one of such techniques, can't remember the name now, but it's similar to the concept of double buffering in graphics. even a triple buffering technique int threads exists.
concurrent programming is a neccessity for next-gen or companies are going to bite the dust.

seems B&W used a combination of decision trees and some other stuff. no neural nets in there.
others --> http://www.gameai.com/games.html.

@neoteny

I was going to just glance at your HUGE post (well done!) , assuming it was just copy-pasted, but Iread it and am glad I did, because it seems you made your arguments very clear- thus making it easier for me to get my points across! So thats what I'll do.

The mgs4 gun-in-mouth trailer is NOT a gamplay trailer. Before you roll your eyes in exasperation, allow me to explain! When you launch a game like Halo, or deus ex 2 on a PC, do you notice how the realtime cutscenes are so much faster than the game itself? Thats because a cutscene is a custcene, whether rendered realtime or not. You are comparing what amounts to a puppet show to a complete game! Definitely, the full game will be toned down considerably.

Also, we have not been shown any gameplay demos showing new or interesting gameplay. All we have seen is cutscenes- and cutscenes do not make a game. Finally, you give kojima too much credit. He scored major hits with MGS3 and the different versions of MGS1 (including twin snakes) but MGSA was rubbish to most people (we can blame the psp if we want) while Sons of liberty, despite positive reviews, scored 'most overrated game' awards by gamespot and Edge (and possibly everyone who had to suffer the games idiotic savepoint placement and snake-less gameplay. I want my snake!!!!!)




Euphoria: read between the lines. Euphoria falls under what I described as 'cosmetic physics'. The article clearly states that euphoria is NOT ai, but is rather a system to allow a ragdoll-type to interact properly with objects in its environment- for increased VISUAL appeal. Nothing to do with gameplay. Paralellizable YES, but does not improve gameplay. For the physics to improve gameplay, like in HL2, you need to integrate within the main thread. But codec has made some good arguments against that, I'll need to read a bit more before reinforcing that position. Nontheless Euphoria is NOT an example of this type of phys/AI- it is purely cosmetic.

i copied and pasted this because of your rant about how crap AI is and watnot. well heres a sample of nextgen AI. you are bein unnecessarily overcritical(one might even say cynical) about what u have neither seen nor experienced. you base your arguments and "research" on the works of critics of the ps3 while i try to approach this from a common sense perspective with as little industry jargon as i can.

Dude! cool down! please don't write off my research as 'research' in quotes! It hurts my feelings!

I posted a link to the microsoft article because you said MS accepts the cell is faster. I do not recall basing any other research on ps3 critics. And as for industry jargon, dude, you just copy-pasted a perfect example of marketing glockenspiel into your very post! Hey chill, they are just games, and sony fanboys are probably going to buy the ps3 in droves anyway, even if it turns out that sony built it from pigeon droppings! Lets cool down and have fun analyzing this cool technology, and seeing past sony's marketing hype.

Also, bros, did you REALLY read the ibm tech specs? You did read that the spe's are vector processors? And you DID read about their method of operation? You *DID* read about the method used to share memory and the recommendations on how they are meant to be used? Well, if you did, and you still maintain your position, I'll just keep quiet because anything I say will make me sound like an arrogant prick/pretentious bastard. I'll just say this: I guess its a matter of point-of-view. At the end of the day, the only thing we will have will be benchmarks, and you can be SURE they will all be fixed! so lets just forget about it, ok?

I'll just drop a clue: the keyword is 'predictability'. Perhaps that will help, ??


And finally, there's the major point: affordability!

It seems ps3 will be expensive, but what if sony is LYING? what if ps3 is actually going to be the same price as xbox360 and they just said this to mislead nintendo? U know nintendo was going to anounce the wii price at e3, but they didnt once they heard the ps3 price, and later came up with $250. What if they were planning to announce 199.99 or 239.99 and they decided to bump it up because of the high ps3 price?

I admit, its just wild speculation. But if this is what sony is really doing i will give them hand o!

@codec

check it out: (linked from your article)

http://www.gameai.com/blackandwhite.html

It is not a neural net, but not as simple as a decision tree either. It is a simplified neural net using symbolic rather than numeric data structures.

also, from

http://www.dansdata.com/blackwhite.htm

You Creature learns fighting skills, spells, how to impress Villagers in various ways, and, of course, where to poo, from you. You can stroke the Creature to tell him he's been good and beat him when he's been bad, and a conceptually simple neural net arrangement guides the animal's behaviour. You also get three Leashes - a neutral one that encourages learning, and two others that encourage compassion and aggression, respectively - which tell your creature to pay particular attention to what you're doing.

and then there's this

http://www.jadetower.org/muses/archives/000015.html

In fact, I've had my own experiences with the game that convinced me that my little ape had a temperament all of its own. Games like "Black & White" or "Creatures" have used neural network techniques to push the bounds of artificial intelligence. Indeed no one doubts the success neural networks have had in progressing the computer sciences.

So well, it seems like a lot of people think there are neural nets in there, ne? But maybe molyneux is just lying and throwing marketing hype around like Sony

lol. just saw this --> PS2 outselling 360.

meeen

thats bad o

it means the 'next gen' may end up being a colossal flop.

if ps2 sales are still so strong, the ps3 has a major opponent already.

I still put my money on wii. Its the only really interetsing new console. The other 2 are just a bundle of inflated tech specs and part 3, part 4, part 10 of the same games we've been playing since 1996 (10 years!)

Anyway ill prolly get a ps2 before then. I need to play god of war and res4 to see what the hype is all about.

God of War was okay but it didn't deserve Game of The Year. That was for Resident Evil 4.

surely, purely >>>xbox360!

Beautiful, beautiful! Can you send me one for free?

cant wait for PS3 to come out am trading ma xbox 360 for it, November aint far,

darkmanzero, you are absolutely right: you are an arrogant prick/pretentious bastard, lol. but i guess i must bow to a might greater than me. <bowing and mumbling "i'm not worthy">, and i did put  research in quotes because u didnt do any extensive studies on a physical ps3, did you? we are just trading numbers based on what both sides say of the darn consoles, based on our experiences, and mostly based on our opinions. but it all sickens me  mouthing off specs and tech so i'm done arguing. codec seems more real about this because his posts have this agreeable air of caution. and yes i copied and pasted some of it, but that's because u echoed john carmack and tons of what other people say. and by the way i do know what cutscenes are, i never claimed any mgs cutscene was in-game, all i said was it was rendered in real time, and such renders give an i dea of what the console is capable of. so keep your xbox360; a whole bunch of mankind is rooting for the ps3 and i KNOW they wont be disappointed.

and by the way, i never said euphoria was ai, i only said it augments AI. read: "To be clear, euphoria is not an AI program, it's an animation technology that allows for physically accurate behaviors. The AI is programmed by the developer, in this case LucasArts, into the euphoria models to control them. What we saw at LucasArts was the marriage of euphoria physics and LucasArts A.I. scripts".

Yeah, u are absolutely right.

Since i'm basing my research on published specs, i am actually also just parroting the words of others.

I just felt that internal ibm documents may be a bit closer to the truth than some marketer measuring penis length against that of bill gates or shigeru miyamoto, but the truth is that even that document is simplified, if I really wanted to claim elite status i should interpret the PCB's and OPcode lists!

ANyway just like u said, its all a waste of time. the only ppl that should care about it all are the devs anyway, and all WE should care about is what they produce!

the battle will be decided in november when ps3 launches! if it clears the board like ps2 did then microsoft and nintendo will be history!

well im glad that cleared the air then, cos im fagged out googling stuff. and wat i believe is dat despite the fact that wat we may end up wit are consoles that spout "cosmetic effects" rather than gameplay, people will still gun after them just for the wow effects. maybe the console devs are taking a lesson from hollyhood which still insists on pouring out a fair sized chunk of summer blocs laden wit techy effects and little sense. i dare say a lot of movies get made simply cos they offer ILM and co the possibility for showcasing their latest tricks ( superman, anybody?). wen last did the gaming industry actually crafted a revolutionary and daring new style of gameplay? its all sequels and redos, its sickening. also, and this is all my opinion so feel free to differ, i realize that the more power a console has, the less charisma it has. PSP is a cool piece of tech stuff but its so, impersonal, now the DS feels a lot warmer and closer to me almost as much as my phone. PSP feels like a sony CAN-DO project rather than the actual entertainment gizmo DS is but, what do i know, right. anyway cheers. and codec, keep it up man. u doin good.

Yeah, I agree about the blockbusters and the 'efefcts' games. I guess when a movie/game commands such a high budget, people are more willing to spend money on measurable things (number of effects shots, more polygons and textures) than abstract things like style , story and gameplay.

however dont be sickened yet- even amongst the sequels etc the gaming industry has managed to implement many innovations over the past few years, in fact there have been so many fresh ideas in gaming that it makes one wonder how long it will last! sadly, the new ideas are never advertised, only the polygon count, so it is difficult to know what you are missing. games you may want to check out:

killer 7 : cool visual style, excellent new method of blending action with cinematic cameras
viewtiful joe : showcase of merging animation events with gameplay
half life 2 : visual and event-driven storytelling at its best
advent rising: many problems, but a promising mix of narrative and gameplay
stubbs the zombie: 'growing group' action- quite entertaining!
psychonauts: this has to be seen to be believed
God of war: for the first time (pop:warrior within came close), a thirdperson combat system that works.

These improvements tend to be evolutionary, nopt revolutionary, but many gamers dont keep up with each released game so we may be pleasantly surprised when we pick up a new game only to see about 5 years of innovation piled up.

Gaming is NOT losing its innovation! at least i dont think so,



oh yeah, downgrade your psp to 1.5 firmware, uy a 2gig memstick and enter the homebrew scene. you just may start sleeping with it again!

earlier on someone mentioned neural nets, are they applicable to the gaming environment say to bolster AI? i dont know much about them. and what about distributed intelligence? think they can find a way to make bots as entirely unpredictable as we want them to?

I never believed in the concept of Artifical intellegence. I think a lot of research had gone into these area with very little succesful commercial applications conmesurate with the amount of funds of research.

in terms of consumer products, yes, but elsewhere it has a greater impact than u may be aware of especially in domains removed from consumer artifacts. Turing machines do exist-in labs, u may say, or assembly lines, even as concepts wether defense/military or medical-but learning systems are real. expert systems ( or microworld programs) are in wide commercial use in medicine, chemistry, credit authorisation, financial management, corporate planning, document routing in financial institutions, oil and mineral prospecting, genetic engineering, automobile design and manufacture, camera lens design, computer installation design, airline scheduling, cargo placement and they're based on AI. But of course most of us are unaware of what entails making financial statements, say, or designing a camera lens. but in games AI isnt much to write home about (see Darkmanzero's postings above) cos its mostly clever scripting-just bots running off severely limited algorithms. the problem with AI is that it doesnt really offer itself to practical commercial applicability but those who pour funds into research perhaps do it for other reasons.

clocky.

AI is a thriving field with many applications. Because robbie the robot does not yet exist does not mean it is responsible for 'very little succesful commercial applications'. Do some research into what modern AI research is like, and you may be pleasantly surprised. In particular pay attention to some of the areas Neoteny mentioned


Neoteny

One reason why AI in games remains simple is that it is, for the most part, not neccesary. Xaero is Quake 3 is a dumb scripted bot but I challenge you to beat hm on hardcore difficulty!

Earlier on c0dec and I did some digging to see if Neural nets are used in Black and white (conflicting statements from various sources). It turns out that they are! so there you have it- a neural net enhanced game. Most people do believe the creatures to be somewhat intelligent. I certainly was quite amazed by my Ape a few times.

i still dont know much about neural nets cos everything i see about them STRESSES von Neumannism and comparative analyses between biological nets and artificial neural nets. one thing i noted was how they talked of training the nets rather than programming but i guess however a bot plays is down to the programmer and not any intelligence on the part of the program. but then again we may see a time when artificial nets closely resemble their biological counterparts but this beggars the question: does great AI make a good game better? ive not played quake 3 yet but i dont know if a more competitive bot will extend the semblance of reality i expect in a game or whether it will bog down the experience by being frustratingly hard to kill. frankly if i want to fantasize im wiping off a swarm of aliens in some far off galazy, i'd rather my aliens were dumb and acting a script so i can leisurely finish 'em off in 50 different ways.

the idea of using a neural net in black and white comes from the fact that you have to bring up and train a creature in a virtual world. It grows and develops a personality based on its experiences, most of which may take place despite or directkly due to your interference with the wiorld! the game's really cheap now, you can get an ortigio for 1k5 at game (shoprite) and it works on mmost modern pcs so why not pick up a copy and see for yourself?

You are right about the question 'is it neccesary'. Well, the truth is that many multiplayer games are exciting because the opponents are 'intelligent' humans, not intelligent in the sense that they are unbeatable, but in the sense that they are forced to learn and adapt to survive under conditions similar to the players'.

A perfect AI will be a perfect substitute for a human opponent, and thus will enhance the game greatly! Also in RTS games, good AI means less micromanagement, which means a better game. Also in squad based games, you will feel more attached to your men if they display initiative rather than actuing like dumb bots.

However in real life, bots are dumb, and they cheat by having more information about the situation than the player. However, a well programmed bot can be a compelling and exciting opponent! When you finally figure out how zareo works, it can be an edge-of-your-seat experience trying to beat him because he's so fast, he almost never misses, he's deadly accurate and nearly unstoppable, but still a dumb bot that you can manipulate if only you keep your wits about you!

Also who can forget fighting AKUMA and M.BISON in street fighter! cheating galore but if you could beat them you were THE MAN!

i dont know precisely if being competitive passes for intelligence cos as u say the bots have more information than the human player and this really is cheating as at once a handicap is created by virtue of that inbalance. well whatever. but i have a question: given that the present and next generation of consoles cannot fully simulate a real world and at the same time simulate real behaviour (as in human-level AI in multi-agents) because machine resources are limited , should game developers concentrate on creating a more realistic world, or should they focus on developing advanced AI?
check this link out, its a demo of multi-agent bots running off UnrealScript and game engine.; apparently they found a way to separate game logic from core game engine.
www.planetunreal.com/gamebots

Well, a scripted not is a scripted bot so obviously it isnt intelligent. but it sure is competitive!

My opinion is that realistic worlds are not neccesary for a great game. If you made a perfect simulation of a city in which to play, it may or may not give an entertainigngame. However the maps of DOOM1 and DOOM2 took gameplay into account before ebven bothering to resemble any real structure in structural design I mean, i dont care about textures cos of course it was 1992 and textures were primitive then, (most of the levels in doom are not recognisable as being similar in design to any real buildings at all) . Of course the games were, and still are, amazing today because of that.

Therefore a game world should be a playing field first, then a simulation of reality second.


secondly, advanced AI is not neccesary if it cannot cause the player to believe his opponent is human. Since the AI we have today cannot do that, we are left with c0dec's cool mixture of scripted decison trees and AI concepts used to handle some tricky parts. Managed well, an immensely entertaining game can be created. This is the soul of games! I mean. are u telling me you cant remember the way you felt when you first discovered how to get sagat to tiger uppercut forever using dhalsim's standing hard kick?!?!?!?!?

Meeeen. Games are just too much. I feel like tanning sf now, even sef

oh please dont get all nostalgic on me, i always shed a tear whenevr i remember the glorious days of the SNES. "sonic boom", " shoryuken"<sigh>