This is what Sony would like people to believe - but is not the case according to most game developers. . . .
Sony's current APIs are really, really inelegant. I expect that they will mature quickly. Programmers will have to change the way they think, which will also take a little bit of time, but adapting to increasingly parallel forms of software development is an industry-wide phenomenon right now.
So really at the end of the day, we'll have to wait a year or so until developers get used to the PS3/Cell. But they will, and when that happens I think you'll find that the PS3 will have a marked advantage over the rest of the market.
So really at the end of the day, we'll have to wait a year or so until developers get used to the PS3/Cell. But they will, and when that happens I think you'll find that the PS3 will have a marked advantage over the rest of the market.
bit like when the ps2 came out then. i remember the getaway as quite a good looking game when it came out. now, UGLY!! lol its all blocky and rubbish!
....So really at the end of the day, we'll have to wait a year or so until developers get used to the PS3/Cell. But they will, and when that happens I think you'll find that the PS3 will have a marked advantage over the rest of the market.
Around the start of 2008 hopefully there'll be a great nVidia GeForce 9 series that's like 65nm and 50Watts on load. That will kick the PS3's a55 to Mars and back.
Just like the resolution and quality of the PS2 is well surpassed by mid-high-end ATI / nVidia cards. But yes, PS2 is hella fun to play.
I think the XBOX360 is on the back foot now, with Microsoft having to dump more money on it to stave off the PS2, PS3, Wii.
I don't envy the developers and publishers(?) -- having to code for what, PS2, PS3, XBOX, XBOX360, Wii, WinXP, WinVista, OS X -- not all of them at once and only some, ever, except for the major titles which would target *almost* all of them.
bit like when the ps2 came out then. i remember the getaway as quite a good looking game when it came out. now, UGLY!! lol its all blocky and rubbish!
XBOX360 games right now - A year out, some of the titles don't look that great. Again, I have to throw in the PC gamer perspective and say for the titles released in the first 6 months of XBOX360, most now are easily surpassed by say, an nVidia 7900GT 256mb card.
Sony's current APIs are really, really inelegant. I expect that they will mature quickly. Programmers will have to change the way they think, which will also take a little bit of time, but adapting to increasingly parallel forms of software development is an industry-wide phenomenon right now.
I doubt if they will mature. They never did in the past. Sony has a learning curve which is a indistinguishable from a flat line.
Quote:
Originally Posted by Splinemodel
So really at the end of the day, we'll have to wait a year or so until developers get used to the PS3/Cell. But they will, and when that happens I think you'll find that the PS3 will have a marked advantage over the rest of the market.
If creating games were all about floating point maths that would be true.
The 360 is just better at running code. And an easier platform in which to develop that code.
With the PS2 the nightmare of development actually worked in Sony's favour. Developers had to opt for the only viable platform. Had to comit to developing for the PS2 - and having made that decision could not really afford to do anything special for the more capable Xbox.
The same might happen again. But the 360 has year's head start. And finally there is a game or two which does not suck.
I doubt if they will mature. They never did in the past. Sony has a learning curve which is a indistinguishable from a flat line.
...
If creating games were all about floating point maths that would be true.
The 360 is just better at running code. . . .
Sure, but the PS3 is very different. It's much more open. I will be very surprised if mature APIs and even IDEs don't surface from Sony or other players in the market.
...
The problem is that expert systems have dominated AI routines for a long time, as we've had fast, sequential ALUs for a while now, and haven't had good FPUs until very recently. However, there are many, many transform-based ways to get quality AI, and they actually tend to provide more desirable learning characteristics at the expense of requiring lead developers that are more math-savvy than previously. So there are ways to get high-cost routines to be FPU bound, and there's certainly plenty of financial motive right now for developers to learn to use these.
....The problem is that expert systems have dominated AI routines for a long time, as we've had fast, sequential ALUs for a while now, and haven't had good FPUs until very recently. However, there are many, many transform-based ways to get quality AI, and they actually tend to provide more desirable learning characteristics at the expense of requiring lead developers that are more math-savvy than previously. So there are ways to get high-cost routines to be FPU bound, and there's certainly plenty of financial motive right now for developers to learn to use these....
Whooooooooo that's enough brain food for a week. ...Curious, where would you place the Wii in terms of coding for it and in terms of its hardware engine?
Sure, but the PS3 is very different. It's much more open. I will be very surprised if mature APIs and even IDEs don't surface from Sony or other players in the market.
...
The problem is that expert systems have dominated AI routines for a long time, as we've had fast, sequential ALUs for a while now, and haven't had good FPUs until very recently. However, there are many, many transform-based ways to get quality AI, and they actually tend to provide more desirable learning characteristics at the expense of requiring lead developers that are more math-savvy than previously. So there are ways to get high-cost routines to be FPU bound, and there's certainly plenty of financial motive right now for developers to learn to use these.
It is interesting that you believe that improved AI will produce better games. I have heard this argument for a while - that realistic physics and more general purpose AI will result in more rewarding game experiences.
I don't think either argument holds up. Although I used to think the same way (and made costly mistakes trying to make it happen)
Physics in games uses a lot of FPU. It is great to see objects collapsing and exploding realistically. It is a great effect which enhances realism. But what games designers avoid is having gameplay outcomes determined by the unpredicatable nature of physical simulations. What game designers need to do is force outcomes - to engineer specific situations to arise in a very deterministic way. For instance in a popular Formula One racing game - the coolest thing was for a car ahead to spin out of control right in front of the player. The designers made this happen, not because of physics or AI but becasue it is the coolest thing.
Better AI would make better games - it seems like an obvious conclusion, but again players do not want their enemies to run-off, hide for a while and ambush them hours later. It would suck. Instead what players want is for enemies to look smart but act predicatable. Designers want to contrive winnable situations that play our predictably according to a script. The game must pave a way for the player to devise and learn a solution to the problem. If the game AI learned too - that would seriously screw up the dynamic.
Some players claim they hate the on-rails nature of current games. But the sales figures say otherwise.
At a very high-level I think what is best is a blend of "objective-driven AI". Physics and explosions and stuff could be extremely realistic. Its effects though should be limited to for example, damage inflicted on players and non-player-characters (NPCs)... There should also be safeguards in place that can check on the AI and then "force" as you mention, events to happen. If enemies go and hide out for a while, for example, there should be a random (within a certain range) timeout for them to come back and try and get you.
HalfLife2 and Quake4, taking FPSs as examples, are definitely very conventional and very lead-you-by-the-hand kind of games. F.E.A.R. as well, though the enemy AI was very interesting. They still had strong objective-driven elements (that is to circle around and kill you or outright ambush you and try and kill you).
Depends on the genre. Successful FPSs need pacing where action takes place at certain regular intervals alongside a plot. HalfLife2:Episode 1 did this usually with "set pieces", for example, a four-storey broken-down carpark with 3 holes constantly producing those AntLions. You could jump up and climb up the levels, and just keep on shooting AntLions, or you could close up the holes with cars via the gravity gun. Other HL2:Episode 1 set pieces includes zombies in a locked room that then bursts open when "not expected" (the zombies escaping is triggered by player movement further down into another room).
Controlled but not predictable plot-unravelling, objective-driven AI, and moving away from "set pieces" which are the same every time you play the FPS, would return computer gaming to more non-linear interesting gaming. But will the market like it? So far only online multiplayer has been very successful in providing very unpredictable and engaging experiences in a defined environment. Bots have a bit of a way to go, but CounterStrike bots are actually not too bad in terms of AI, IMHO.
RTS AI has taken some interesting turns but after several good sessions battling the AI teams, you get a feel for how they build and move and attack for a given map....I thought Command And Conquer: Generals though was very interesting in the various ways you could achieve your objectives, a good mix of interesting maps and diverse building-up of your forces.
....What game designers need to do is force outcomes - to engineer specific situations to arise in a very deterministic way. For instance in a popular Formula One racing game - the coolest thing was for a car ahead to spin out of control right in front of the player. The designers made this happen, not because of physics or AI but becasue it is the coolest thing....
This is still important but needs a light touch - in HL2 and HL2:Episode 1 for example, these "cool things happening" was for the most part very predictable and exactly the same each time you played through a level.
I think part of the challenge now with "interactive gaming entertainment" is justifying the cost of developing and testing a high level of alternate scenarios which not everyone will experience unless they replay the game all the way through say 3-4 times... ?
Comments
This is what Sony would like people to believe - but is not the case according to most game developers. . . .
Sony's current APIs are really, really inelegant. I expect that they will mature quickly. Programmers will have to change the way they think, which will also take a little bit of time, but adapting to increasingly parallel forms of software development is an industry-wide phenomenon right now.
So really at the end of the day, we'll have to wait a year or so until developers get used to the PS3/Cell. But they will, and when that happens I think you'll find that the PS3 will have a marked advantage over the rest of the market.
So really at the end of the day, we'll have to wait a year or so until developers get used to the PS3/Cell. But they will, and when that happens I think you'll find that the PS3 will have a marked advantage over the rest of the market.
bit like when the ps2 came out then. i remember the getaway as quite a good looking game when it came out. now, UGLY!! lol its all blocky and rubbish!
....So really at the end of the day, we'll have to wait a year or so until developers get used to the PS3/Cell. But they will, and when that happens I think you'll find that the PS3 will have a marked advantage over the rest of the market.
Around the start of 2008 hopefully there'll be a great nVidia GeForce 9 series that's like 65nm and 50Watts on load. That will kick the PS3's a55 to Mars and back.
Just like the resolution and quality of the PS2 is well surpassed by mid-high-end ATI / nVidia cards. But yes, PS2 is hella fun to play.
I think the XBOX360 is on the back foot now, with Microsoft having to dump more money on it to stave off the PS2, PS3, Wii.
I don't envy the developers and publishers(?) -- having to code for what, PS2, PS3, XBOX, XBOX360, Wii, WinXP, WinVista, OS X -- not all of them at once and only some, ever, except for the major titles which would target *almost* all of them.
bit like when the ps2 came out then. i remember the getaway as quite a good looking game when it came out. now, UGLY!! lol its all blocky and rubbish!
XBOX360 games right now - A year out, some of the titles don't look that great. Again, I have to throw in the PC gamer perspective and say for the titles released in the first 6 months of XBOX360, most now are easily surpassed by say, an nVidia 7900GT 256mb card.
Sony's current APIs are really, really inelegant. I expect that they will mature quickly. Programmers will have to change the way they think, which will also take a little bit of time, but adapting to increasingly parallel forms of software development is an industry-wide phenomenon right now.
I doubt if they will mature. They never did in the past. Sony has a learning curve which is a indistinguishable from a flat line.
So really at the end of the day, we'll have to wait a year or so until developers get used to the PS3/Cell. But they will, and when that happens I think you'll find that the PS3 will have a marked advantage over the rest of the market.
If creating games were all about floating point maths that would be true.
The 360 is just better at running code. And an easier platform in which to develop that code.
With the PS2 the nightmare of development actually worked in Sony's favour. Developers had to opt for the only viable platform. Had to comit to developing for the PS2 - and having made that decision could not really afford to do anything special for the more capable Xbox.
The same might happen again. But the 360 has year's head start. And finally there is a game or two which does not suck.
C.
I doubt if they will mature. They never did in the past. Sony has a learning curve which is a indistinguishable from a flat line.
...
If creating games were all about floating point maths that would be true.
The 360 is just better at running code. . . .
Sure, but the PS3 is very different. It's much more open. I will be very surprised if mature APIs and even IDEs don't surface from Sony or other players in the market.
...
The problem is that expert systems have dominated AI routines for a long time, as we've had fast, sequential ALUs for a while now, and haven't had good FPUs until very recently. However, there are many, many transform-based ways to get quality AI, and they actually tend to provide more desirable learning characteristics at the expense of requiring lead developers that are more math-savvy than previously. So there are ways to get high-cost routines to be FPU bound, and there's certainly plenty of financial motive right now for developers to learn to use these.
....The problem is that expert systems have dominated AI routines for a long time, as we've had fast, sequential ALUs for a while now, and haven't had good FPUs until very recently. However, there are many, many transform-based ways to get quality AI, and they actually tend to provide more desirable learning characteristics at the expense of requiring lead developers that are more math-savvy than previously. So there are ways to get high-cost routines to be FPU bound, and there's certainly plenty of financial motive right now for developers to learn to use these....
Whooooooooo that's enough brain food for a week.
Sure, but the PS3 is very different. It's much more open. I will be very surprised if mature APIs and even IDEs don't surface from Sony or other players in the market.
...
The problem is that expert systems have dominated AI routines for a long time, as we've had fast, sequential ALUs for a while now, and haven't had good FPUs until very recently. However, there are many, many transform-based ways to get quality AI, and they actually tend to provide more desirable learning characteristics at the expense of requiring lead developers that are more math-savvy than previously. So there are ways to get high-cost routines to be FPU bound, and there's certainly plenty of financial motive right now for developers to learn to use these.
It is interesting that you believe that improved AI will produce better games. I have heard this argument for a while - that realistic physics and more general purpose AI will result in more rewarding game experiences.
I don't think either argument holds up. Although I used to think the same way (and made costly mistakes trying to make it happen)
Physics in games uses a lot of FPU. It is great to see objects collapsing and exploding realistically. It is a great effect which enhances realism. But what games designers avoid is having gameplay outcomes determined by the unpredicatable nature of physical simulations. What game designers need to do is force outcomes - to engineer specific situations to arise in a very deterministic way. For instance in a popular Formula One racing game - the coolest thing was for a car ahead to spin out of control right in front of the player. The designers made this happen, not because of physics or AI but becasue it is the coolest thing.
Better AI would make better games - it seems like an obvious conclusion, but again players do not want their enemies to run-off, hide for a while and ambush them hours later. It would suck. Instead what players want is for enemies to look smart but act predicatable. Designers want to contrive winnable situations that play our predictably according to a script. The game must pave a way for the player to devise and learn a solution to the problem. If the game AI learned too - that would seriously screw up the dynamic.
Some players claim they hate the on-rails nature of current games. But the sales figures say otherwise.
C
HalfLife2 and Quake4, taking FPSs as examples, are definitely very conventional and very lead-you-by-the-hand kind of games. F.E.A.R. as well, though the enemy AI was very interesting. They still had strong objective-driven elements (that is to circle around and kill you or outright ambush you and try and kill you).
Depends on the genre. Successful FPSs need pacing where action takes place at certain regular intervals alongside a plot. HalfLife2:Episode 1 did this usually with "set pieces", for example, a four-storey broken-down carpark with 3 holes constantly producing those AntLions. You could jump up and climb up the levels, and just keep on shooting AntLions, or you could close up the holes with cars via the gravity gun. Other HL2:Episode 1 set pieces includes zombies in a locked room that then bursts open when "not expected" (the zombies escaping is triggered by player movement further down into another room).
Controlled but not predictable plot-unravelling, objective-driven AI, and moving away from "set pieces" which are the same every time you play the FPS, would return computer gaming to more non-linear interesting gaming. But will the market like it? So far only online multiplayer has been very successful in providing very unpredictable and engaging experiences in a defined environment. Bots have a bit of a way to go, but CounterStrike bots are actually not too bad in terms of AI, IMHO.
RTS AI has taken some interesting turns but after several good sessions battling the AI teams, you get a feel for how they build and move and attack for a given map....I thought Command And Conquer: Generals though was very interesting in the various ways you could achieve your objectives, a good mix of interesting maps and diverse building-up of your forces.
....What game designers need to do is force outcomes - to engineer specific situations to arise in a very deterministic way. For instance in a popular Formula One racing game - the coolest thing was for a car ahead to spin out of control right in front of the player. The designers made this happen, not because of physics or AI but becasue it is the coolest thing....
This is still important but needs a light touch - in HL2 and HL2:Episode 1 for example, these "cool things happening" was for the most part very predictable and exactly the same each time you played through a level.
I think part of the challenge now with "interactive gaming entertainment" is justifying the cost of developing and testing a high level of alternate scenarios which not everyone will experience unless they replay the game all the way through say 3-4 times... ?