575 réponses à ce sujet

[Targonis1]

For a second or third playthrough, you have higher difficulty settings to make things a bit more challenging. The key issue as others have said, is that you want the game to continue to be a challenge as you progress through it. So, it would really suck to make the game where the enemy in some places were so easy it was pathetic.



Now, Risen was a game where things don't scale at all, though things change at various points as you progress through the story. That was a game where you do NOT want to go into a dungeon early in the game because the monsters really are so dangerous you have to be higher level to stand a chance of surviving against them.



So, look at Dragon Age...the point is that you have a horde of monsters invading. So what do you do to make things more difficult, just throw so many mobs at the player that you run out of spell points and stamina halfway through the fight and you have to run around while resting your combat abilities? It isn't like you fight from the easy stuff in front to the more difficult later either, but more that as the enemy forces advance, they send out more and more raiding parties and such.



So, where would the "more powerful" stuff be compared to the easy stuff? For the most part, you would have bands of wimpy(to the player) mobs to terrorize the people in the countryside, and you would have the more powerful in their lairs. Would you want to even be bothered to save the farms after the early levels since there would be no challenge in doing it?



Or, the game could send you deeper and deeper into the Deep Roads, fighting more and more powerful forces, but then you end up in a more linear story to get the feeling that the game gets more difficult as you progress. This is what most games do, they guide you by the nose from easy and "tame" areas into more and more dangerous areas, and prevent you from even getting to the dangerous stuff until much later in the game.



No one wants to feel like they are on a linear path through the game, and that is the trade-off. The only real way to solve the problem would be to have the game change encounters based on the order you go through the content. So, if you go to one area, you don't save the people on the farm, and if you go to the farm, the people are now dead, and there is a fortified camp. If you don't take care of the camp, it will now have walls and fortifications later when you finally DO decide to go there, and the longer you take before dealing with that area, the more powerful the forces are to deal with.



That may be the way to go, but it also means that there would be a lot more work to set up the different potential side-effects of these missions. Picture if you decide NOT to stop in Lothering. What would happen to the people there? Some might flee on their own, and show up with new quests in the other areas. Or, what if you go to Redcliff before Lothering and you stop the events there and can get enough forces to defend Lothering that the Darkspawn attack the Dalish Elves instead? The number of potential scenarios would be very very high, and the developers would have to account for too many to let the game be very long.



Of course, isn't that what we want, to see that choices in missions will have a wide-scale impact on how the rest of the game plays out? What if your Origin Story not only changed your character background going into Ostagar, but also your first mission against the Darkspawn? So, maybe you COULD have saved the day at Ostagar with one origin story, but there are Darkspawn attacks from different places, so you save one place, but lose the others, and each one has its own tactical change to the game world.

[Zem_]

Paromlin wrote...

If you're talking about AI developed by Bioware then yes. If you're talking in general then no.
Just take a look at the wonderful work TonyK has done for NWN2 AI. If a single person can do it, then you'd expect a company that produces video games can do it as well...

No, I do mean in general.  I haven't played with TonyK's AI and I am sure it is an improvement but AI is still just AI.  It's not unpredictable or creative.  It follows a set of rules and learning those rules eventually means you have the upper hand.   Always.   Whether you are level-scaling or laying out fixed-level spawns, you have to consider giving your NPC enemies an increasing advantage either in levels or abilities ABOVE that of the player in order to keep pace, because the player will always be able to outsmart the AI.  How trivial would many of the key encounters in NWN2 and DAO be if they did not both make liberal use of the "cutscene ambush", for example, to put you in a bad starting position, lock the door behind you (that you would never have been stupid enough to walk through in the first place), etc?

[Baskervore]

Zem_ wrote...

Paromlin wrote...

If you're talking about AI developed by Bioware then yes. If you're talking in general then no.
Just take a look at the wonderful work TonyK has done for NWN2 AI. If a single person can do it, then you'd expect a company that produces video games can do it as well...

No, I do mean in general.  I haven't played with TonyK's AI and I am sure it is an improvement but AI is still just AI.  It's not unpredictable or creative.  It follows a set of rules and learning those rules eventually means you have the upper hand.   Always.   Whether you are level-scaling or laying out fixed-level spawns, you have to consider giving your NPC enemies an increasing advantage either in levels or abilities ABOVE that of the player in order to keep pace, because the player will always be able to outsmart the AI.  How trivial would many of the key encounters in NWN2 and DAO be if they did not both make liberal use of the "cutscene ambush", for example, to put you in a bad starting position, lock the door behind you (that you would never have been stupid enough to walk through in the first place), etc?

Wait, is there an AI mod to improve the enemies of the game? Hmmm I might not need it though, I have enough trouble as it is, lol.

[Zem_]

Baskervore wrote...
Wait, is there an AI mod to improve the enemies of the game? Hmmm I might not need it though, I have enough trouble as it is, lol.

Not that I know of.  TonyK's AI mod is for the Neverwinter Nights games. 

[ladydesire]

There isn't one for DA, yet; the one we are talking about is for Neverwinter Nights 2. Even if there was one for DA, it likely wouldn't affect lower difficulty levels as much, if at all; that's the way I'm thinking about doing it, anyway.

[Upper_Krust]

Zem_ wrote...

Upper_Krust wrote...

The main reason is that Levelling Up itself (in tandem with all the benefits like magic items which increase in power) makes the disparity between the PCs and the Enemies greater and greater.

Giving them better gear would help, but I'd say AI is an ever bigger issue, and I am not even talking about exploits.  The more abilities you add, the more choices you have in combat, and thus the more complex the decision making process is. 

The main problem with Dragon Age is not the flat difficulty though (although it is indirectly), its that the game requires virtually zero thought. There are no worthwhile choices to make for either players or enemies. That is because the game is, on the one hand, far too forgiving, and on the other hand simply doesn't make the differences pronounced enough. 

I mean what weapon do I use on a Golem? A mace should be better, but you don't get punished for using a sword against them to any degree so why bother (especially if your magic sword is a better weapon in general)? Choosing what spell or weapon to use SHOULD make a difference to the outcome. Thats what this game lacks, consequence of combat actions. 

AI is simply not able to make the best use of those abilties.  Just look at rogues.  Is the AI even capable of making sure rogues are positioned properly for backstabs against the highest threat foes?  Will they send a stealthed assassin to flank and take out your mage while the rest of them provide a diversion?  Are their mages smart about using AoEs, placing ground effect spells to hinder your progress at choke points?

Both those could be handled with a sentence or two of code.

Aside from certain scripted events, they pretty much are not able to employ advanced tactics or to use their abilities wisely.  They rely on numbers (and liberal abuse of the "cutscene" )  But however you slice it, the more abilities you get as you gain levels, the harder it becomes for the AI to out-think a human player, even if they have the same bag of tricks. 

You are thinking about this in the wrong way. AI shouldn't have to outthink the player.

1. Enemies should have a small list of attacks which they use dictated by what the heroes do (in the same way the secondary players are controlled with tactics).

2. Enemies should be set a Behaviour pattern (again the same way secondary PCs can be controlled)

3. Enemies should have a Morale rating.

a. Cowardly...run away when outnumbered (if they escape add these enemies to a future encounter)
b. Standard...will surrender when the last man standing (gives the player the choice to free them or kill them)
c. Stubborn...will fight to the death
d. Courageous...gains a bonus to attack/damage when outnumbered and a further bonus when the last man standing

4. Enemies should have a strategy.

Just to differentiate this from tactics, a strategy would be a statistical bonus (of some kind) which the enemy would gain as long as it follows its original plan.

ie. Protect the leader, attack the mage, block the doorway

5. Most (of the exotic) enemies should be strong/tough in some areas but weak/vulnerable in others.

These force players into making choices about (a) their own attacks, ( their own defences and © their positioning.

[Paromlin]

Zem_ wrote...

No, I do mean in general.  I haven't played with TonyK's AI and I am sure it is an improvement but AI is still just AI.  It's not unpredictable or creative.  It follows a set of rules and learning those rules eventually means you have the upper hand.   Always.   Whether you are level-scaling or laying out fixed-level spawns, you have to consider giving your NPC enemies an increasing advantage either in levels or abilities ABOVE that of the player in order to keep pace, because the player will always be able to outsmart the AI.  How trivial would many of the key encounters in NWN2 and DAO be if they did not both make liberal use of the "cutscene ambush", for example, to put you in a bad starting position, lock the door behind you (that you would never have been stupid enough to walk through in the first place), etc?

Then I think you're wrong.

Yes, AI is AI - human brain is human brain. But, there's a but... this is a game with a ruleset. You act in the boundaries of these rules and I must say that DA's ruleset ain't so complicated. If they manged to refine chess AI so that it's able to beat the best chess players in the world.. I imagine they can do it as well for a game; at least to a degree that gives a real headache (in a positive sense) to the player. AI doesn't need to be unpredictable or creative - you can do only so much in a computer game; some things work and some things don't.

Who will have the upper hand depends entirely on the skill of the player and the AI (and difficulty level).

While I dislike the cutscene ambush, I don't see anything wrong with closing doors. I mean, why would *they* come and not you? I am all for having no escape to safety and being forced to fight to death. Adds to the challenge.

You must have in mind that DA's enemy AI doesn't even use potions... but they instead have 5x as much HP as you do. :/ That was Bioware's way to tweak the difficulty anyway.

[AlanC9]

Upper_Krust wrote...
I mean what weapon do I use on a Golem? A mace should be better, but you don't get punished for using a sword against them to any degree so why bother (especially if your magic sword is a better weapon in general)? Choosing what spell or weapon to use SHOULD make a difference to the outcome. Thats what this game lacks, consequence of combat actions.

I dunno about weapon choice. I though the golf bag syndrome of D&D 3.5 was pretty stupid.

[dragon_83]

AlanC9 wrote...

. I though the golf bag syndrome of D&D 3.5 was pretty stupid.

What is golf bag syndrome?

[Upper_Krust]

AlanC9 wrote...

Upper_Krust wrote...
I mean what weapon do I use on a Golem? A mace should be better, but you don't get punished for using a sword against them to any degree so why bother (especially if your magic sword is a better weapon in general)? Choosing what spell or weapon to use SHOULD make a difference to the outcome. Thats what this game lacks, consequence of combat actions.

I dunno about weapon choice. I though the golf bag syndrome of D&D 3.5 was pretty stupid.

True. but what we are talking about here is a very small number of choices, but each making a tangible difference.
Runes slot into weapons in DAO so its not like you would overtly need more than 3.

Also remember that at certain parts of the game you may only have a limited choice of weapon types, so you get to see the good and bad of certain weapons on certain monsters. 

[Upper_Krust]

dragon_83 wrote...

AlanC9 wrote...

. I though the golf bag syndrome of D&D 3.5 was pretty stupid.

What is golf bag syndrome?

Basically in D&D 3.5, lots of monsters (in particular outsiders* and golems) had Damage Reduction linked to a certain type of material.

*Such as demons, devils etc.

So a devil might have Damage Reduction 10/silver. So unless you are using a silver weapon the devil takes 10 less damage from your attacks with that weapon.

The way players would get around this is by carrying half a dozen weapons of different materials. So you would have a silver weapon for devils and lycanthropes, adamantine weapons for golems, iron weapons for faeries and demons etc.

The way to avoid this in a video game (which isn't easy for PnP) is to use percentages (which are easily calculated on the fly by the processor).

So instead of something like 10/silver. You could have 25% vs. Golems or 400% vs. demons etc.

So you basically want (for example) maybe a base 1, 2 and 4 damage (lets say Crushing = 1, Slashing = 2, Piercing = 4) but then modify that for armour so that Crushing = 4, Slashing = 2, Piercing = 1.

So against flesh, Piercing Weapons are the best.
Against plants; Slashing might be the best.
While against golems, Crushing weapons might be the best.

Once you establish the basics then you play around with the rules.

Oozes might be 90% immune to all physical damage.
Skeletons might be 90% immune to piercing weapons.
Plants might be 90% immune to crushing damage.

etc.

[Zem_]

Upper_Krust wrote...

Zem_ wrote...

AI is simply not able to make the best use of those abilties.  Just look at rogues.  Is the AI even capable of making sure rogues are positioned properly for backstabs against the highest threat foes?  Will they send a stealthed assassin to flank and take out your mage while the rest of them provide a diversion?  Are their mages smart about using AoEs, placing ground effect spells to hinder your progress at choke points?

Both those could be handled with a sentence or two of code.

You can have a NPC rogue look for a mage class character and then pursue him, but what if my mage is running around or he's placed a rune of repulsion at his feet? Will your two lines of code enable the enemy rogue to abandon his backstab attempt and maybe try to shoot me instead? Or maybe the higher threat foe is not the mage, but the warrior charging YOUR mage.

I could go on, but you see where this is going? I can come up with scenarios all day because I can think. A computer can't. You have to think of all these ahead of time and program the computer with a huge series of if-thens to handle all these contingencies, to say nothing of the ones that require you to recognize tactical positioning and terrain features like choke points, target ground spells where they will have the greatest effect even if no one is currently standing there. Seriously, I want to see the "simple" game AI that can do this.

You are thinking about this in the wrong way. AI shouldn't have to outthink the player.

Didn't say they had to, but they have to do something to balance the fact that the player CAN outthink them. That was my point in response to the "they only need comparable weapons" argument. No... they need better weapons. Or better abilties. More HP (which they usually have). Because at the end of the day you can always run circles around them in the tactical dept. YOU are not limited to a simple list of if-then conditions. YOU can anticipate what they are doing. The computer can't necessarily anticipate what you will do. It won't target an AoE spell on a location you WILL be and if they tried to program in that ability you would quickly learn how to feint to get them to waste a spell.

The game may sound simple when you just think in terms of weapons and abilities and levels but it gets a lot more complicated when you add positioning and timing to the mix. When I see AI that can really surprise me in the tactical dept., I will change that opinion.

Modifié par Zem_, 30 mars 2010 - 08:02 .

[dragon_83]

Upper_Krust wrote...

dragon_83 wrote...

AlanC9 wrote...

. I though the golf bag syndrome of D&D 3.5 was pretty stupid.

What is golf bag syndrome?

Basically in D&D 3.5, lots of monsters (in particular outsiders* and golems) had Damage Reduction linked to a certain type of material.

*Such as demons, devils etc.

So a devil might have Damage Reduction 10/silver. So unless you are using a silver weapon the devil takes 10 less damage from your attacks with that weapon.

The way players would get around this is by carrying half a dozen weapons of different materials. So you would have a silver weapon for devils and lycanthropes, adamantine weapons for golems, iron weapons for faeries and demons etc.

The way to avoid this in a video game (which isn't easy for PnP) is to use percentages (which are easily calculated on the fly by the processor).

So instead of something like 10/silver. You could have 25% vs. Golems or 400% vs. demons etc.

So you basically want (for example) maybe a base 1, 2 and 4 damage (lets say Crushing = 1, Slashing = 2, Piercing = 4) but then modify that for armour so that Crushing = 4, Slashing = 2, Piercing = 1.

So against flesh, Piercing Weapons are the best.
Against plants; Slashing might be the best.
While against golems, Crushing weapons might be the best.

Once you establish the basics then you play around with the rules.

Oozes might be 90% immune to all physical damage.
Skeletons might be 90% immune to piercing weapons.
Plants might be 90% immune to crushing damage.

etc.

Thanks for the explanation. I was familiar with this D&D game mechanic, but I didn't know it is called golf bag syndrome by the players.

[ladydesire]

Paromlin wrote...


You must have in mind that DA's enemy AI doesn't even use potions... but they instead have 5x as much HP as you do. :/ That was Bioware's way to tweak the difficulty anyway.

No need to use potions when most encounters have a Mage; I don't think I've seen too many Darkspawn encounters without at least one Emissary present and it's rare for encounters with Human or Elf to not have one as well.

[Paromlin]

ladydesire wrote...



No need to use potions when most encounters have a Mage; I don't think I've seen too many Darkspawn encounters without at least one Emissary present and it's rare for encounters with Human or Elf to not have one as well.

You do realise healing spells can't simply substitute potions and that a combination of both is much better than having only a mage, with limited mana and spell cooldowns, to heal the entire enemy squad?

[Upper_Krust]

Zem_ wrote...

You can have a NPC rogue look for a mage class character and then pursue him, but what if my mage is running around or he's placed a rune of repulsion at his feet? Will your two lines of code enable the enemy rogue to abandon his backstab attempt and maybe try to shoot me instead? Or maybe the higher threat foe is not the mage, but the warrior charging YOUR mage.

We have seen how flexible the tactics in DAO can be. Theres no reason to suggest there couldn't be a tactics stipulation of...

"If you can't injure PRIMARY TARGET (within 'x' amount of seconds), switch to SECONDARY TARGET."

I could go on, but you see where this is going?

Absolutely. You are nitpicking.

I can come up with scenarios all day because I can think. A computer can't. You have to think of all these ahead of time and program the computer with a huge series of if-thens to handle all these contingencies,

'A huge series of if-thens' that would take a competent individual less than a day to come up with, nevermind a game YEARS in development with MULTIPLE designers.

to say nothing of the ones that require you to recognize tactical positioning and terrain features like choke points,

Again easily handled. On each encounter map, all you have to do is indicate where a specific feature (like a choke point) is. Then the AI will recognise it.

target ground spells where they will have the greatest effect even if no one is currently standing there. Seriously, I want to see the "simple" game AI that can do this.

Any game with an ounce of thought put behind it could handle everything you have stated so far.

Didn't say they had to, but they have to do something to balance the fact that the player CAN outthink them.

...and that something is choice. The bottom line is that the players SHOULD outthink the AI. This isn't Kasparov vs. Deep Blue we're talking about. The goal is not for the computer AI to win. The goal is for the computer AI to challenge the player to outthink a given situation, not that AI is constantly adapting from one move to the next. 

That was my point in response to the "they only need comparable weapons" argument. No... they need better weapons. Or better abilties. More HP (which they usually have). Because at the end of the day you can always run circles around them in the tactical dept. YOU are not limited to a simple list of if-then conditions. YOU can anticipate what they are doing. The computer can't necessarily anticipate what you will do. It won't target an AoE spell on a location you WILL be and if they tried to program in that ability you would quickly learn how to feint to get them to waste a spell.

The game may sound simple when you just think in terms of weapons and abilities and levels but it gets a lot more complicated when you add positioning and timing to the mix. When I see AI that can really surprise me in the tactical dept., I will change that opinion.

Unfortunately Dragon Age doesn't attempt any real surprises. Which is why you can beat the game with basically the same tactic in every fight.

Modifié par Upper_Krust, 30 mars 2010 - 10:45 .

[mikeburns]

I played through all of DAO and the DLC on Xbox 360 (on Hard) and then almost all again on PC (also on Hard.) I feel like even with the location based enemy difficulty scaling, there is still a very palpable sense of character growth, at least imho.



For me, especially on my first playthrough, towards the end, where my blood mage would be destroying groups of enemies with mass paralysis, death cloud, and blood wound, turning their own mage against them with blood control, I definitely felt a growing sense of ability and power, I think to the point that the sense of progress was something that kept me playing the game.



Additionally I think there is a certain sense of wavering difficulty that is unique to Dragon Age. Later in the game a big group of Darkspawn might not pose a challenge, but every time I saw a Revenant I was like ugh crap. Anyway. I thought the difficulty balance was great and additionally, it's kind of a "what do you expect" situation. I think given the open ended nature of the game they did a great job.

[Zem_]

Upper_Krust wrote...

Any game with an ounce of thought put behind it could handle everything you have stated so far.

Name ONE that does.  I've been gaming for quite some time and I've never been surprised by computer AI.  It is nearly always the case that the solution used to challenge the player is by gifting the NPC enemies with advantages the player doesn't have.  Greater numbers, more hit points, higher defenses... something.  All I am saying is this is done because the advantage a human player has in tactics is significant.

No game out there can pit you against a party equal in every way to yours and still challenge you.   Contrary to popular belief, not everyone resorts to PvP in games just to be a sociopathic twit.  They do so for a real challenge.

Didn't say they had to, but they have to do something to balance the fact that the player CAN outthink them.

...and that something is choice. The bottom line is that the players SHOULD outthink the AI. This isn't Kasparov vs. Deep Blue we're talking about. The goal is not for the computer AI to win. The goal is for the computer AI to challenge the player to outthink a given situation, not that AI is constantly adapting from one move to the next.

When did I ever say the goal was for the computer to win? You pointed out one reason that the game gets easier and easier as you level up: Gear.  I don't dispute that, but I am just pointing out that the AI is a big factor too.  The more abilities you get the more options you have in combat and thus the more complex it is.  The more choices there are, the more difficult it is for an AI with a finite list of rules to challenge you.  So you need to scale them up more in later levels than you did in earlier levels. 

Then I get hit with "Oh no AI is easy".  And that's what we've been arguing since.

Seriously, if AI was as easy as you claim why isn't anyone doing it?  Instead of having to throw greater numbers of enemies at you each with 5x your hit points or something, they could have smaller encounters with really smart enemies who are more equal to your characters and everyone would gush about how good their AI is.   Yet it doesn't happen.  Nearly every game I play has a forum with complaints about "bad AI".  Have I somehow missed playing all the RPGs with good AI?

[Upper_Krust]

[quote]Zem_ wrote...

[quote]Upper_Krust wrote...

Any game with an ounce of thought put behind it could handle everything you have stated so far.[/quote]

Name ONE that does.[/quote] 

I can't. I don't play enough RPGs to have a spectrum of games to draw upon.

That said I am familiar enough with RPGs to know that no game (at least none I have seen) has you battle the same enemies all the way through (except DAO).

However, your point is something of a red herring.


[quote]I've been gaming for quite some time and I've never been surprised by computer AI.[/quote]

This is the red herring I'm talking about. Its not the goal of AI to surprise. Its the goal of a situation to surprise or present a puzzle/challenge. Then the player learns that puzzle and works out a solution, moving on to the next challenge.

In Dragon Age, combat doesn't really require any thought. It almost never presents any problems that cannot be surmounted with simply attacking an enemy.


[quote]It is nearly always the case that the solution used to challenge the player is by gifting the NPC enemies with advantages the player doesn't have.  Greater numbers, more hit points, higher defenses... something.  All I am saying is this is done because the advantage a human player has in tactics is significant.[/quote]

Theres nothing wrong with NPC enemies having abilities or powers players can't have. Their reason for existing is solely to provide an obstacle/challenge to the player. When that obstacle is inconsequential it renders any challenge impotent.


[quote]No game out there can pit you against a party equal in every way to yours and still challenge you.   Contrary to popular belief, not everyone resorts to PvP in games just to be a sociopathic twit.  They do so for a real challenge.[/quote]

Again, more silliness on your part. A game can be challenging and fun, many videogames are. Dragon Age Origins is to a degree (despite its flaws: once you 'learn' the game its no longer much of a challenge). But Awakening is not challenging at all.

[quote]When did I ever say the goal was for the computer to win?[/quote]
 
You seem pre-occupied with the notion AI needs to be able to outthink or at least match a player in order to be challenging or rewarding.


[quote]You pointed out one reason that the game gets easier and easier as you level up: Gear.  I don't dispute that, but I am just pointing out that the AI is a big factor too.  [/quote]

Depends on your definition of AI. The enemies just need a few basic pointers to provide a challenge.

[quote]The more abilities you get the more options you have in combat and thus the more complex it is.  The more choices there are, the more difficult it is for an AI with a finite list of rules to challenge you.

So you need to scale them up more in later levels than you did in earlier levels.[/quote] 

Something which isn't done.


[quote]Then I get hit with "Oh no AI is easy". And that's what we've been arguing since.[/quote]

Argument is healthy for the mind.


[quote]Seriously, if AI was as easy as you claim why isn't anyone doing it?[/quote] 

Here's the kicker. In DAO the secondary characters are controlled by Tactics and Behaviours. Why couldn't enemies be controlled in the same way? When do you see enemies heal others - never. When do you see enemies drink potions or use items - never.

[quote]Instead of having to throw greater numbers of enemies at you each with 5x your hit points or something, they could have smaller encounters with really smart enemies who are more equal to your characters and everyone would gush about how good their AI is.   Yet it doesn't happen.  Nearly every game I play has a forum with complaints about "bad AI".  Have I somehow missed playing all the RPGs with good AI?
[/quote]

Again, I can't remember ever being out thought by a game...but I certainly remember being CHALLENGED by videogames.

Dragon Age challenges up until the point you 'get' (as in understand) the games combat. From that juncture onwards (through lack of difficulty and variety) it no longer becomes a challenge.

Yet almost every other game I have played does a better job of maintaining the level of challenge even after you 'get' the game.

Do different people 'get' the game sooner or later than others? Of course they do. This is why we have different difficulty levels. Yet even here Dragon Age; or more specifically Awakening, disappoints.

Modifié par Upper_Krust, 31 mars 2010 - 02:24 .

[1varangian]

I hope they come up with a game engine that allows them to play with the numbers in an encounter instead of levels. And have a system where a mass of low levels would actually pose a threat to high level PC's when they surround and overwhelm them.



Fighting against an actual horde of enemies would not only be epic but would also reflect the scaling of power very well.



I wouldn't object to a more patient PC power curve either. Baldur's Gate and Pool of Radiance are great examples of awesome games that keep PC advancement on a more believable scale.

[tmp7704]

Upper_Krust wrote...

That said I am familiar enough with RPGs to know that no game (at least none I have seen) has you battle the same enemies all the way through (except DAO).

I'd think that's quite a standard actually -- most games have a (limited) selection of enemies they pit against you in either random or location-based order. These enemies can be sometimes given differently coloured textures and altered names to provide impression you are advancing from bashing Lame Rats to bashing Grey Rats and eventually get strong enough to battle Vicious Black Rats, but ultimately you're still collecting rat tails throughout the game. Or sith lightsabers.

There is also the issue of enemy types being to large extent dictated by both the universe backstory and the game plot. If the game involves a sudden twist and/or a 'big reveal' about nature of the threat the player is facing then it makes more sense to introduce different enemy types once that happens. But if the plot revolves about battling the darkspawn which is supposed to be encompassing threat and its nature is known from the beginning, then it's to be expected the fights will generally involve said darkspawn from the beginning to the end, rather than suddenly take a turn to fighting Princes Turned Frogs.

Modifié par tmp7704, 31 mars 2010 - 03:54 .

[Zem_]

Upper_Krust wrote...

I've been gaming for quite some time and I've never been surprised by computer AI.

This is the red herring I'm talking about. Its not the goal of AI to surprise.

Okay stop.  This is getting old.  Do you SEE me saying it is the "goal" of AI to surprise?  I am simply pointing out that this good AI you keep saying is so easy to do.... I've never seen it done.  Either you are a genius, everyone else in the professional game industry is an idiot, or good AI is, in fact, harder to implement than it is to armchair program in a forum.

Regardless, I am not arguing for or against AI changes.  That's never been my point. 

Look, let me try one more time to rewind this argument.  This started with the assertion that in the higher levels, like in Awakening, the game is just getting progressively easier.  You said, I believe, that this was due to lack of appropriate gear and could easily be solved by scaling up their gear.  My only point here is that this is not enough, because the OTHER thing which is getting progressively more outclassed as you level up and gain more abilities is... AI.  I have not argued that it is the job of AI to surprise anyone.  I merely point out that it never has and that is WHY you need to scale enemies up PAST the player.

Yes, DAO doesn't do this.  I am suggesting it could.  Whether you do this with automatic scaling or not.  If you don't make your enemies tougher or more numerous than the player, they won't challenge because they just aren't smart enough.  In DAO, this complexity is mainly due to the wide variety of magic spells available to the player.  Melee abilities tend to be much more straight-forward and so are a lot easier for the AI to handle.  But magic gets ridiculous pretty fast.

Theres nothing wrong with NPC enemies having abilities or powers players can't have. Their reason for existing is solely to provide an obstacle/challenge to the player. When that obstacle is inconsequential it renders any challenge impotent.

Did I say there was anything wrong with it?  I in fact am saying it is necessary.  So we are in violent agreement here, apparently.

When did I ever say the goal was for the computer to win?

You seem pre-occupied with the notion AI needs to be able to outthink or at least match a player in order to be challenging or rewarding.

No.  I have never said this.  I am saying quite the opposite.  I am saying that because it can't do this, you must scale enemies up past the player in ability to compensate and you must do so at an increasing rate the more abilities the player gains as they get into higher and higher levels.

Yet almost every other game I have played does a better job of maintaining the level of challenge even after you 'get' the game.

I don't find this to be the case in CRPGs myself.  There is such a focus on leveling up and getting all kinds of rewards that it almost always trivializes high level combat eventually, no matter what tricks they try to pull.  You have to limit yourself with self-imposed challenges to maintain that level of difficulty.

[Upper_Krust]

Zem_ wrote...

Okay stop.  This is getting old.  Do you SEE me saying it is the "goal" of AI to surprise? 

You said you had never been surprised by AI in games, implying that you thought this important.

I am simply pointing out that this good AI you keep saying is so easy to do.... I've never seen it done.

Because the term AI is somewhat misleading. The enemies don't need to be able to think, they just need some either/or stipulations. They don't need to be able to react to every situation, just a handful, if even that.

Either you are a genius, everyone else in the professional game industry is an idiot, or good AI is, in fact, harder to implement than it is to armchair program in a forum.

Well I have a high IQ, certainly bordering on genius and I currently work in the RPG (pnp) industry and at one stage I flirted with the idea of working in the video game industry. Just for relaxation purposes I enjoy designing and redesigning video games (on paper that is - I'm no programmer)...hence my posts on these and other forums. So I'd like to think I have a decent understanding of the medium, although I'll certainly defer on matters of programming  to an actual programmer.

I don't think game industry professionals are idiotic, far from it. I think they are a very intelligent, hardworking bunch of people. However, that said, we have two advantages over developers.

1. Hindsight
2. Unlimited budget constraits (time and money) to represent our ideas.

Is good AI easier to armchair program in a forum...of course it is.

But lets examine the facts as we see them.

Dragon Age already has a Tactics and Behaviour System controlling the secondary characters. Is there any reason why this couldn't be used for individual enemies - no. Is there any reason these choices could not be expanded upon - no.

So the question becomes why are the enemies not using these features? 

Look, let me try one more time to rewind this argument.  This started with the assertion that in the higher levels, like in Awakening, the game is just getting progressively easier.  You said, I believe, that this was due to lack of appropriate gear and could easily be solved by scaling up their gear.

Yes and no. Statistically improving an enemy (via gear is one method) will eventually offset the imbalance of player power. Thus you could maintain the same level of difficulty throughout the game.

However, statistical difficulty should be of secondary importance behind forcing players to make tactical choices in combat and having those choices be the difference between success and failure. This might not mean a single choice. But could be multiple choices where there is a beneficial and detrimental outcomes. Too many bad choices and the player should be punished.

- For example a choice might be to concentrate your attacks on the healer or summoner first.
- Or it might be to use fire spells on a monster weak to fire.
- It might be to let the leader live until the end because the Shrieks go berserk when the leader is killed.
- A monster might only be 90% invulnerable to all attacks from the front, so you have to keep two characters on opposite sides to it.
- Golems might be 90% resistant to magic so keep your mages healing the tank.
- Oozes might be 90% resistant to physical attacks.
- Dwarves might attack in Square formation, which is weakest at the corners.
- Leaders might use potions 5 seconds after they are reduced below 25% health so kill them quick at that point.
- While the enemy musician plays all enemies gain a bonus to attack/damage, so kill him first.
- Plants might be 90% resistant to all weapons except axes.

But in Dragon Age there are very few choices which ultimately matter. Using a mace on a golem instead of a sword maybe makes a 5% difference. But you can't set up a game like DAO to swing on a 5% difference. Thus the difference between using a mace and a sword becomes virtually irrelevant.

My only point here is that this is not enough, because the OTHER thing which is getting progressively more outclassed as you level up and gain more abilities is... AI.  I have not argued that it is the job of AI to surprise anyone.  I merely point out that it never has and that is WHY you need to scale enemies up PAST the player.

Then we pretty much agree. What I would say though is that while enemy AI cannot surprise the player. The situation CAN be set up to surprise a player...the first time its encountered. Prompting the possible defeat of the player. Who then has to work out

1. Why they got beat
2. Whats the best strategy to overturn the outcome

Yes, DAO doesn't do this.  I am suggesting it could.  Whether you do this with automatic scaling or not.  If you don't make your enemies tougher or more numerous than the player, they won't challenge because they just aren't smart enough.  In DAO, this complexity is mainly due to the wide variety of magic spells available to the player.

 

Automatic scaling seems to be an excuse for using the same enemies far too often in Dragon Age.

Melee abilities tend to be much more straight-forward and so are a lot easier for the AI to handle.  But magic gets ridiculous pretty fast.

This was a problem with 3rd Edition D&D too. They solved it with 4th Edition. 

Magic does seem to be overpowered in DAO. An attack against a single target in melee should be much more powerful (against a single target) than an attack which affects multiple targets from range.

Melee Attack vs. Single Target = 100%
Ranged Attack vs. Single Target = 50%
Melee Attack Arc = 50%
Ranged Attack Cone = 33%
Ranged Attack Circle = 12.5 to 25% depending on the size of the area covered
Ranged Attack All Enemies on Battlefield = 6.25%

If we compare Fireball to Arcane Bolt in DAO, fireball does twice as much damage to all enemies in a decent sized circle AND knocks them down. Arcane Bolt, on paper looks about as good as a basic warrior power. Fireball is x2.66 costlier in mana, yet in real terms its probably about 10-20 times better (at least based on my initial appraisal of the math). Cooldown times don't really have that big an effect because you just switch to casting a different spell.

Did I say there was anything wrong with it?  I in fact am saying it is necessary.  So we are in violent agreement here, apparently.

Thats the internet for you.

I don't find this to be the case in CRPGs myself.

Maybe I haven't played enough CRPGs lately.

There is such a focus on leveling up and getting all kinds of rewards that it almost always trivializes high level combat eventually, no matter what tricks they try to pull.  You have to limit yourself with self-imposed challenges to maintain that level of difficulty.

That just seems a bit like lazy design to me.

[ladydesire]

Upper_Krust wrote...


Dragon Age already has a Tactics and Behaviour System controlling the secondary characters. Is there any reason why this couldn't be used for individual enemies - no. Is there any reason these choices could not be expanded upon - no.

So the question becomes why are the enemies not using these features?

They are; if you look at the AI Packages, they are exactly the same as the in game TActics and behavior system for companions. Not all enemies have minimal tactic choices; it's just a matter of how long they are allowed to exist that determines whether or not they are a challenge.

However, statistical difficulty should be of secondary importance behind forcing players to make tactical choices in combat and having those choices be the difference between success and failure. This might not mean a single choice. But could be multiple choices where there is a beneficial and detrimental outcomes. Too many bad choices and the player should be punished.

Agreed, but should those choice be dictated by the AI, or how the character chooses to build their character?

But in Dragon Age there are very few choices which ultimately matter. Using a mace on a golem instead of a sword maybe makes a 5% difference. But you can't set up a game like DAO to swing on a 5% difference. Thus the difference between using a mace and a sword becomes virtually irrelevant.

Which is because Golems in DAO are not magical constructs, but rather Living Constructs (like D&D Ebberon setting Warforged); weapon choice shouldn't be the difference between defeating one or dying.

There is such a focus on leveling up and getting all kinds of rewards that it almost always trivializes high level combat eventually, no matter what tricks they try to pull.  You have to limit yourself with self-imposed challenges to maintain that level of difficulty.

That just seems a bit like lazy design to me.

Lazy design? I don't think so, since it allows more players with less than super builds to experience a challenge without having to alter their build style significantly. I decided to see what all the complaints were about and built a character that put all the points in one attribute and I found I don't like the lack of real challenge; there is risk, but when the foe falls in about 4 hits, I find that I'd rather continue with a "gimped" build than play where there is no significant challenge.

[Upper_Krust]

ladydesire wrote...

They are;

...just not to any extent that would impact an encounter - which is surely the whole point of them.

if you look at the AI Packages, they are exactly the same as the in game TActics and behavior system for companions.

Thats as I would expect. Why build a competent Tactics and Behaviour modelling system and then not use it.

Not all enemies have minimal tactic choices; it's just a matter of how long they are allowed to exist that determines whether or not they are a challenge.

In 500+ encounters I have never seen an enemy heal another nor have I ever seen an enemy drink a potion. Thus Bioware must have set up the tactics wrong (or at least to be inconsequential).

Agreed, but should those choice be dictated by the AI, or how the character chooses to build their character?

I think the Boss AI should maybe have a plan A and plan B (plan B for the event that Plan A isn't working). Standard monsters may not need more than a plan A. Neither should be influenced by the character build.

Characters should adapt around the challenges set by the enemies in the game. Enemies shouldn't have to adapt to the player.

Which is because Golems in DAO are not magical constructs, but rather Living Constructs (like D&D Ebberon setting Warforged); weapon choice shouldn't be the difference between defeating one or dying.

That doesn't make sense. They are made from stone, metal etc. Why wouldn't weapon type be a factor and why couldn't all golems be 90% resistant to magic (and by that I mean they only take 10% damage from spells)? 

Lazy design? I don't think so, since it allows more players with less than super builds to experience a challenge without having to alter their build style significantly. I decided to see what all the complaints were about and built a character that put all the points in one attribute and I found I don't like the lack of real challenge; there is risk, but when the foe falls in about 4 hits, I find that I'd rather continue with a "gimped" build than play where there is no significant challenge.

You have justified everything I have said to date. There is no significant challenge unless you gimp yourself.

Even with an optimised min/maxed party the higher difficulty levels should still be a challenge but they just simply are not...thats lazy design.