Dissapointed

[Arryn]

Originally posted by Kristoffer O:

quote:

---

Originally posted by Strages Sanctus:

--Number of targets hurting me
--Are the troops I am about to target immune to this type of spell damage?
--How many of the attacking enemey troops will I hit with this spell.
--Will it hurt me if I am in the AOE of the spell?
--How many friendlies will be hurt by this spell?
--Armor of troops attacking me
--Is the spell I am about to use more efficient at damaging or penetrating armor.
-- how much will thi spell fatigue me? will I go unconcious if I cast this spell.

The AI does consider these things with the possible exception of AOE. Which of the above do you believe the AI must rate as more or less important a factor and in what cases?

The placement of AOE>1 is randomized and thus difficult to calculate regarding risk to your self.

---

In order ...

--Are the troops I am about to target immune to this type of spell damage?
(if the target is immune, there is no point to casting the spell)

--how much will this spell fatigue me? will I go unconcious if I cast this spell.
(if the caster is in imminent danger, making the situation more dire is a bad idea)

--Will it hurt me if I am in the AOE of the spell?
(only consider hurting self if you can destroy appreciable numbers of foes, I'll leave it to others to argue over what's 'appreciable')

--How many friendlies will be hurt by this spell?
(never cause more than, say, 1/4 remaining damage to friendlies unless you are assured of victory by wiping out the opposition. no one enjoys pyrrhic victories)

--Number of targets hurting me
(if you are being hit by six enemies, casting a spell that only hurts one of them, and won't even kill that one, is silly)

--How many of the attacking enemey troops will I hit with this spell.
(see earlier comments)

--Armor of troops attacking me
--Is the spell I am about to use more efficient at damaging or penetrating armor.

[Coffeedragon]

Originally posted by Lord_Devi:
I should definatly be able to take more control over them - the AI just isn't capable of using it's magic to it's full potential.

More control would be bad for SP, I believe -it would make computer opponents even weaker.

Even improvements on the spellcasting AI would not be without setback entirely -I think Magic vs. Might in Dom2 is already a little bit too much on the side of Magic.

[Strages Sanctus]

First let me say: Great work so far guys; AI programming is the holy grail of gaming.
Let me add, there is always room for improvement .

Most important to least important:


1. Are my targets immune to this damage type? (if yes ignore this spell)

2. Do I have any spells I can cast that my targets are especially vulnerable to? (examine these first)

3. Do I have any AOE spells I am immune/resistant to but my enemies are succeptible to? (this should be chosen over AOE spells that I am not immune/resistant to)

4. How many IMMEDATE threats can potentially be eliminated with this casting.
4.1. Area of the spell
4.2. Damage potential
4.3. Armor penetration potential
(focus on damage/AOE for light armor enemies)

5.Consider fatigue impact. By this I mean you can come up with a fatigue to damage potential ratio.... The most efficient spell should be used.
Discount any spell that will make me go unconcious.

6. Will friendlies be hurt?


This is where things start getting recursive; and very difficult to weight. I don't envy your task at all .

7. Will the potential number of surviving enemies attacking me be capable of killing me next round?
7.1 Can I remove myself from danger without fleeing the battle? (hard to calculate I guess)
7.2 Can I protect myself from the ensuing potential damage?

8. Can I cast 2 spells a turn? Can I augment my abilities before considering which offensive spell to cast?


Thanks for reading, considering and communicating with your customers.

[ January 27, 2004, 17:06: Message edited by: Strages Sanctus ]

[IKerensky]

Johan, did idea of a spellbook list went into developpers ?

What was the pros/cons against it ?

I think it could probably solve 9 out of 10 problems with the current casting as you can easily switch you caster spell selection considering the opposition expected and avoid many of the IA mistake....

Any luck of it making way into Dom III ?

[Kristoffer O]

>Originally posted by Arryn:
>In order ...

>--Are the troops I am about to target immune to this type of spell damage?
(if the target is immune, there is no point to casting the spell)

The AI does not use spells that the enemies are immune to. Anyhting suggesting otherwise might be a bug and should be reported.

>--how much will this spell fatigue me? will I go unconcious if I cast this spell.
(if the caster is in imminent danger, making the situation more dire is a bad idea)

The AI casts less exhaustive spells if nearly unconscious, but it follows your script, unless there is strong reason not to (immune enemies, out of range etc). The issue of following scripts is a difficult one. Players dislike when the AI overrides scripting, but also dislike when the AI follows their orders and cast spells that inevitably will damage friends.

>--Will it hurt me if I am in the AOE of the spell?
(only consider hurting self if you can destroy appreciable numbers of foes, I'll leave it to others to argue over what's 'appreciable')

Not sure to what extent this is calculated.

>--How many friendlies will be hurt by this spell?
(never cause more than, say, 1/4 remaining damage to friendlies unless you are assured of victory by wiping out the opposition. no one enjoys pyrrhic victories)

In one spell? What is assured victory?

>--Number of targets hurting me
(if you are being hit by six enemies, casting a spell that only hurts one of them, and won't even kill that one, is silly)
>--How many of the attacking enemey troops will I hit with this spell.
(see earlier comments)
>--Armor of troops attacking me
>--Is the spell I am about to use more efficient at damaging or penetrating armor.

The AI sums up the probable dmg dealt to the enemies and this includes dmg, penetration, targets, distance etc. Targets close by are more likely to be targeted as they ar percieved as possible threats.

I rephrase my question. Which of the above do you believe the AI doesn't give enough importance.
Every possible spell is given a 'preference factor' when a mage is about to cast a spell (this is what takes time during battle replays if you have done a great deal of research). The spell and target with the highest 'preference factor' is cast.
So which of the criteria is not given enough influence on the 'preference factor' of a spell. What kind of spells are cast too much or too little and under what circumstances?

[Strages Sanctus]

The biggest issue then as far as my experiences go is that the mages can and do hurt themselves more than their attackers are, in the process of eliminating their attackers.

The next issue would be they favour defense over offense (which is being dealt with in next patch). You should always aim to eliminate your enemies, thus making defense unecessary.

Quickness should always be the first thing cast if available and not scripted already.


IMHO the player spell tools should NOT be made more powerful without making the AI smarter too. This would unbalance SP games immensely and favour magic too much. Improving the weightings of the AI routines would be enough.


SIDE NOTE: Why don't archers have a hold and fire routine available?

[ January 27, 2004, 17:18: Message edited by: Strages Sanctus ]

[Psitticine]

The only "problem spell" I seem to encounter is Breath of Winter. My Water Mages often freeze out their bodyguards and other nearby units by casting it when their script isn't right for the occasion.

I think a check should be made for that sort of situation: abort a defensive spell if other friendly units are too close.