AI spell casting priorities
 

Hi,

Seems like some spells might be too popular.

I've heard about:
Astral Shield
Arrow Fend (recently heard of this, so I'm not sure how bad it is)

Any others since last patch?

Re: AI spell casting priorities
 

Bone Grinding (typically over Undead Mastery)

Re: AI spell casting priorities
 

Kristoffer:

Black list! We want the ability to set a list of spells that our mages are forbidden to cast!

Re: AI spell casting priorities
 

Seriously. This would fix most of the problems.

Jazzepi

Re: AI spell casting priorities
 

+1 vote for blacklist.
If you really want to fix the problem, it will be definitely the best fix

Re: AI spell casting priorities
 

Okey, people. I suspect KO has already heard about the blacklist idea, and that it won't be very easy to implement.

While a blacklist might fix some issues, that's not the point here. KO is asking specifically for spells that have too high spellcasting priority, probably with the intention of fixing that. Wether or not at one point in the future there might possibly be thought of maybe adding a blacklist feature, it won't be today, nor tomorrow. In the meantime having the priority of the ridiculously popular spells lowered would be wonderful, so the least we can do is help the guy out when he asks for input instead of throwing requests for a new feature in his face.

Having said all that, I can't think of one not already mentioned right now, but I'll definitely keep my eyes open.

Edit: Oh, and thanks, KO, for caring.

Re: AI spell casting priorities
 

KO, would the simplest solution be to boost the bonus added to the scripted spell?

At least it looked like that was what was going on in the debug log.

Re: AI spell casting priorities
 

Here are some battlefield spell annoyances:

Priests casting 'sermon of courage' on archers instead of the nearby melee troops marching forward. Melee troops should always be targeted first over archers.

Mages casting elementals like 'living earth' when the enemy is already retreating. This becomes a total waste of gems and shouldn't be casted while the enemy is retreating.


---
I agree with Natpy... spellcasters with 12 hitpoints or less should never cast fireshield unless scripted.

Re: AI spell casting priorities
 

Fire shield

Re: AI spell casting priorities
 

Targeting is a different issue.

Lowering priorities for Summons, especially gem using summons would be nice. Slow summons while the enemy is already retreating are pointless.

I'd say the AI over prioritizes summons in general. Raise Dead spam is nice, but usually summons aren't that effective.

No gem use when the enemy army is routed? I can't think of a reason. As long as beserk or other unroutable units are considered.

Re: AI spell casting priorities
 

Thanks Amhazair http://forum.shrapnelgames.com/images/smilies/happy.gif

I'm just looking for particular spells atm, since I think I might be able to fix that. It should be quite simple to change factors on priority, but I don't think I can manage major changes. Thus I want only to know spells that seem to be overly popular.

Re: AI spell casting priorities
 

Fire flies seems to get prioritized an unfortunate amount. At the very least fire darts should almost always trump it. As mentioned, fire and astral shield are probably too high.

Also, it doesn't matter much to most players, but it would probably help new players and the AI if all the summon ____power spells were very highly favored.

Re: AI spell casting priorities
 

I have to agree w/ summoning, especially after the enemy has routed. If possible, I'd like gem-cost summoning only if scripted.

And while I'm wishing, I really would like a blacklist. http://forum.shrapnelgames.com/images/smilies/happy.gif I've studied the debugging output... and maybe, hypothetically, some things not entirely in compliance with the game's licence... http://forum.shrapnelgames.com/images/smilies/wink.gif and it doesn't look too hard. The real question I think is whether it's worth the Devs' time vs. the long list of bugs and other feature requests.

To get back on topic, is there any way of reducing the priority of Wooden Warriors/Mass protection when my army is facing an army with a lot of heat/fire effects, Abysia especially? I can certainly see that adding checks for this one may be too hard.

Re: AI spell casting priorities
 

Astral shield seems rather high priority too.

Re: AI spell casting priorities
 

Blink should also not get too much priority or at best not be autocasted by the AI at all. It is just not that funny when your yogis (or any other S1 mage i suppose) blinks in the path of retreating enemy archers and gets slain.

Re: AI spell casting priorities
 

I'm going to say fire flies and fire darts. They are pretty useless spells in MA and LA.

I don't think sermon of courage is a problem. It seems to be always cast by an indy priest and that is all he can cast and doesn't have the range for melee troops.

Blink should be removed from the AI list. I do not see it often, only rarely, but I never have seen it used well. That is a spell for a thug or SC, not some 10 HP unequipped mage.

AI does skelly spam and other undead well. Elemental summoning is almost always done poorly.

I do not pay attention to buffs that the AI casts so I do not know how applicable they are.

I do not think I see enough long range damage spells other than smite and a very occasional paralyze.

Re: AI spell casting priorities
 

High level astral mages cast paralyze (at SC's)too often, when soul slay (or enslave mind) would usually be a much better choise.

This is in another ballpark as arrow fend and bone grinding and the like of course. I've never yet seen paralyze overwrite a scripted spell.

Re: AI spell casting priorities
 

Marble Warriors seems to take priority over Army of Gold, strangely.

<font color="red">Edit: This is not necessarily true, as I was using Army of Gold improperly. </font>

Re: AI spell casting priorities
 

Is it possible to lower priority for gem-using spells in general? This is the reason I need to park gems in some scout/fighter commander &amp; transfer tham to mage each turn - or else hust not use gems... http://forum.shrapnelgames.com/images/smilies/frown.gif I don't think that this amount of micromanagement is much fun...

Re: AI spell casting priorities
 

I hate Touch of Madness. With a passion. It should never be cast unscripted.
I never leave a mage unscripted, but I know that if the AI decides that a nature
spell is not worth the gems, it will default to Touch of Madness, and send
archers or mages into the fray. At the end of the battle, the number of
casulties precisely matches the number of targeted units.

And not a single gem should be used against a retreating enemy, ever. What
really sucks, and I have seen it happen at least once, is when the AI decides
that it does not need Arrow Fend when I am storming a castle, but once the
enemy has routed, summoning some Air Elementals is a damn good idea. Arrow
Fend, by the way, is one of the spells I'd love to see as "scripted only".
If you can't be bothered to script it, you should not benefit from it.

Re: AI spell casting priorities
 

If you can't be bothered to script it you shouldn't benefit? That's a weird thing to say. For one thing, who would ever benefit from arrow fend being cast unscripted? Only the AI would have air mages of sufficient skill kicking about with handfuls of gems and no script and most of the time it would still be a bad choice of spell.

Also, why the hostility towards people who don't script? I can't be arsed with the micro hassle of it most of the time. I script when I want certain powerful spells cast or have a definite battlemage script assigned to a number, but that's about it.


I personally think all spells which require gems (other than blood spells) should be 'script only'. The only problem there is the AI nations obviously wouldn't cast any combat spells requiring gems, which is a pretty big problem I admit.

It would definitely be good if any spells which are likely to either be a waste of gems/time or harmful to your own side (like berserk causing spells) were either really low priority or script only. Niche spells like ones only useful to thugs (shields, breath of winter etc), ones which summon neutral creatures (lammashtas, horrors etc) and so on should all be low priority or script only. That way unscripted mages would be more likely to cast something generally useful.

Re: AI spell casting priorities
 

For myself, I can think of at least one instance where summoning elementals during a 'rout' could be useful; that is, summoning air elementals to hit the rear to cause more casualties. However, this would be contingent on summoned elementals attacking the rear, and only really applies to air elementals.

I'd like to second _any_ spell that causes creatures to go berserk is very unhealthy.

Re: AI spell casting priorities
 

Oh, that would be *so* nice! Then you could actually give your Antimagic caster 3 pearls when he is supposed to storm a castle, and he'd actually cast it in three battles (magic/break siege/storm), instead of blowing all the pearls in the magic battle phase.

Armies of the AI nations could be allowed to use gems unscripted -- the game knows which nations are AI and which are human.

Re: AI spell casting priorities
 

I don't know if it's that easy, allowing the AI to use gems unscripted. AI nations currently use the same battle casting AI as humans I believe.

Re: AI spell casting priorities
 

I think it would be nice if after the scripted spells (assuming it did cast those) it will use those spells for the rest of the battle too.
(when my volva (+ light of northern star) are spamming soulslay for 5 turns... I don't want them to cast other spells after that.. there is a REASON I've scripted it 5 times.

Re: AI spell casting priorities
 

It just needs to set/track/compare a flag indicating whether the nation is controlled by human or AI.

If the nation is human, and spell fatigue is over 100 and it's not favspell, set score to zero.

Re: AI spell casting priorities
 

The problem with that is when you run into something you weren't expecting, and your scripted orders aren't as useful as other spells would be.

I've got one solo game where the script I'd used was fine against the indy provinces I was attacking, but the two battles that occurred when Ernor attacked the same provinces, on the same turn that I did, needed much different scripts. If my priests hadn't started banishing after their script ran out, I would have lost both battles. (I didn't even know Ernor was my closest neighbor until we blundered into each other the first time.)

Re: AI spell casting priorities
 

Just as useful as a blacklist in my opinion, would be a 'cast this only' spell, for skellespamming, soulslay spamming, frozen heart, etc. just like how you can set fire, or attack.

Anyway, to answer KO's question, the one's I've noticed a lot are spells others have mentioned, namely; Astral Shield, Fire Shield.

Fire Flies could probably be lowered some, if a mage can cast fire darts, it pretty much always should over fire flies.

Phoenix Power, Earthpower, and Eagle Eyes could use a raise in their value, to help the AI.

Re: AI spell casting priorities
 

On the idea of not using gems on retreating enemies: There is a case where they are still useful: spells that will transfer units to your side.

Re: AI spell casting priorities
 

One of the spells that sends my blood pressure to the roof is Flying Shards (Fire flies is another, for the same reason). Often gets priority over magma bolts and blade wind after scripts end, even over more useful spells like Strength of Giants, Legions of Steel and other things that could be cast.

Fireball and Acid Bolt seem to get much lower priority than they should from mages who can cast them, while Flare gets a much greater one (and is far worse for fatigue).

At the same time, it would be nice if the indie AI grimoire was improved some. It needs to have Fire Darts, Fireball and Magma Bolts added to it for the evocations so that mages other than Air can actually do something other than stand around watching grass grow (which is what flying shards and fire flies amount to most of the time).

Basically any spell that is in the shortlist with a mention of priority should perhaps be looked at. I know Arrow Fend is there and fire and astral shield are already listed as fixed on the progress page, so they're greenified.

Re: AI spell casting priorities
 

Having played MA ulm in single player recently, I can confirm that flying shards gets cast too much.

Summon earth power has too low a priority as my smiths would almost never cast it without scripting. While it is quite useful.

Another thing is that the smiths really prefer casting legions of steel and weapons of sharpness on my independent archers instead of my heavy infantry marching toward the enemy.

Re: AI spell casting priorities
 

Comments, please. Would this rule work like I think it would?

Gems won't be spent after round 5.

Scripted mages will follow their script, and not spend games after their script ends. Non-scripted mages and mages of AI nations can use their gems to cast gem-expensive spells.


About buff spells:

Banned buffs:

Only when scripted (because these are personal buffs more suitable for thug/sc units):

Astral Shield
Fire Shield
Personal Quickness
Blink
Flight (caster-only)
Breath of Winter
Water Shield
Charge Body
Soul Vortex
Barkskin (for poor Abysia)
Stoneskin (there are units with chill aura and Earth)
Invulnerability (Ironskin should be enough for mages, and there might be earth mages with poison aura)

Preferred buffs:
Buffs mages SHOULD cast, if they can:

Summon Water Power
Summon Storm Power
Phoenix Power
Summon Earthpower
Strength of Gaia
Air Shield
Personal Luck
Ironskin
Mistform

Spells I'm not sure about:
Mirror Image
Resist Magic
Iron Will
Personal Regeneration



Combat spells
Preferably, any buff spells/gem-expensive spells would have higher priority in the beginning, and would drop below direct damage spells of the same priority level in few turns. Thus, mages start by casting Luck/Body Ethereal, then quickly change into mind burn/paralyze or better spells.

Astral:
never unless scripted: Blink, Returning, Vortex of Returning, Solar Brilliance, Astral Healing, Soul Drain, Unraveling,

Priority classes, from lowest to highest. The spells in the same priority should be cast over spells of lower priority, but should be pretty much interchangeable against different armies.

Star Fires, Healing Light, Horror Mark
&lt; Mind Burn, Body Ethereal, Luck, Nether Bolt, Arcane Bolt,
&lt; Paralyze, Solar Rays, Stellar Cascades, Battle Fortune,
Astral Geyser
&lt; Soul Slay, Enslave Mind, Light of the Northern Star, Nether Darts, Astral Fires, Control, Opposition

AI nations should give these high priority, otherwise low priority/never cast. If my suggestion in the beginning would happen to be integrated, these would have fairly high priority.
Doom, Battle Fortune, Will of the Fates, Antimagic, Arcane Domination, Master Enslave, Astral Tempest.



I did this list by going through the manual's list of Astral spells. I'm not that experienced in competitive multiplayer, so please correct me if I'm wrong. If someone would like to help in making similar lists for the rest of the paths, I'd be very grateful.

Re: AI spell casting priorities
 

If you're going to include barkskin, you might as well include Protection, Wooden Warriors, and Mass Protection.

Jazzepi

Re: AI spell casting priorities
 

I didn't go through Nature spells yet. Protection could be included, perhaps, but Abysia shouldn't have enough nature to cast Wooden Warriors or Mass Protection any way. Also, if Protection has lower priority than other spells the mage might cast (if it's just N1 random on an abysian mage), it probably won't be cast any way.

Re: AI spell casting priorities
 

for astral you have light of the northern star quite high.. it cost a gem right? and it's not like that gem will be worth it most of the time (if it only gives +1 astral magic). IMHO light of the northern star should be one of the never cast unless scripted for powerfull casters that benefit from it in multiple paths)

Re: AI spell casting priorities
 

Power of the Spheres and Light of the Northern Star are different. I think I didn't have PotS at all. Light of the Northern Star gives +1 Astral to everyone on the battlefield, and because it's battlefield enchantment it can only be cast once per battle, so even in the worst case a lone S3 mage wastes 1 gem to become an S4 mage.

Re: AI spell casting priorities
 

Ahhaahahhahaa

Try telling that to sandman. His random nature mage there to cast relief for his abysian army cast mass protection out of nowhere. I don't think I should have to tell you what happened afterwards. Most of his mages had also cast the fire spell that makes them explode when they die, and he lost like 23 mages in one round of combat.

Jazzepi

Re: AI spell casting priorities
 

That Barkskin is mostly the case when fighting against Abysia, especially with Fire Storm up.

Re: AI spell casting priorities
 

Just for that particular reason I never include ANY nature mages in my Abysian armies. So if anything should get fixed it is this.

Oh, also ironskin casting Eagle Kings...

Re: AI spell casting priorities
 

Agreed, Flare should have lower priority than Fireball.

Re: AI spell casting priorities
 

Hum there should perhaps be exceptions for some evocation spells (namely shadow blast, fire storm, niefel flames, shimmering fields). A player may want to have someone (probably a communion master or high level pretender) chain casting large area evocations as long he has gems, even after 5 rounds. As well there should be an exception for astral healing.


I disagree. They should be very low priority but not blacklisted. As they can't harm you they are something a mage should cast if nothing else -nothing not worse for your troops- can be done. And flight, giving more survavibility to a mage if his army is routed, should rather be in the "usefull buffs" list.

These ones may be harmful for the mage or your troops and should be blacklisted. All protection spells give weakness against one element, so I include ironskin (especially bad in endgame, when the ennemy expect you to use army of lead/gold and so use shock dammage as much he can) only the player should decide to use one of them.

Mistform should be higher priority than anything else (and airshield in land battles). Resist magic is usable only if the army isn't protected by antimagic, so should be high priority too as it may save the mage. In + magic buffs, summon earthpower should be 1st priority as it gives reinvigoration. Flight should get a good value when the mage's army as suffered a lot of losses.




Astral Healing is one of the rare gems spell someone may hope to see used after round 6, except that agree.

I would include Doom and Astral Tempest in the black list. Doom is a waste if you destroy the ennemies, astral tempest kill your own guys.

Will of the Fates should be highest priority if a mage can cast it, and Antimagic as well. Mass enslave spells aren't that good if the ennemy has used antimagic and the caster has no penetration item, and casting one probably mean a powerful mage will be uncounscious, so they should not be valued as much as a defensive BE.

Re: AI spell casting priorities
 

KO, would it help if you had a bundled, sorted collection of the turn files, map files, and the fatherland files for the individual battles in question where the spell casting goes awry?

I know I've kept most of my turns for sloth, and I can provide you with multiple instances of arrow fend being cast for no reason. Also, there was a huge battle where R'yleh had undead mastery ready to cast in a big communion with over 5k+ undead in the battle, but instead cast bone grinding http://forum.shrapnelgames.com/images/smilies/frown.gif

Maybe it would be helpful if you could modify the values for the decision making process, and then "rehost" the turn to run it through a real life example to see how it would change the outcome.

Jazzepi

Re: AI spell casting priorities
 

I know it isn't perfect, but would you rather have that or the current situation?

Thanks for the comments.

I thought the battlefield enchants would mostly be for AI. While giving defensive BE higher value than enslavements would be good, AI could use Mass Enslave against other AI nations.

Re: AI spell casting priorities
 

I've seen an army of vans, with a staff of storms, cast ghost wolves rather than lightning spells. I lazily left them unscripted after summon storm power thinking : what could they cast not to win this battle ?
So I'd put the worthless ghost wolves in the blacklist.

Re: AI spell casting priorities
 

Air:

Never cast: charge body, mist, wrathful skies, mists of deception, shimmering fields

phantasmal warrior
&lt; false fetters, thunder ward, shock wave,
&lt; Aim, Resist Lightning, Air shield, Mirror Image, shocking grasp, wind guide, ghost wolves, confusion, lightning bolt, flight
&lt; Summon Storm Power, Mistform, orb lightning, arrow fend, thunder strike, mass flight, storm warriors, Fog Warriors

Summon lesser Air elemental
summon air elemental
summon valkyries
phantasmal army
living clouds
storm - I'd probably give this rather high rating, perhaps somewhere in the second-highest category. It may hurt you (archer nation), but it helps Air mages (storm power), so it's a tough choice.


Blood:

I don't know much about blood battle magic.

Death:

Death already does fine. Bane Fire and Drain Life should be rather high.

Earth:
never: Curse of Stones, Iron Bane, Rain of Stones, Earthquake

Flying shards, Earth Might, Iron Warriors
&lt; Earth Grip, Fists of Iron, Earth Meld, Farstrike, Rust Mist, Armor of Achilles, Strength of Giants,
&lt; Iron Will, Magma Bolts, Destruction, Shatter, Marble Warriors, Legions of Steel, Blade Wind
&lt; Summon Earthpower, Magma Eruption, Gifts from Heaven, Army of Gold, Army of Lead, Petrify, Weapons of Shrapness

ones I'm not sure about:
Stoneskin, Ironskin, summon LEE, Invulnerability, summon EE, living Earth; the protection spells would probably go under "never cast these", and I have no idea about earth elementals.

Re: AI spell casting priorities
 

Phantasmal warrior or whatever it is, that's a pain in the arse as well. AI loves to cast that.

Re: AI spell casting priorities
 

Yup, and Ghost Wolves. From the little I can tell, the AI likes warm bodies on the battlefield, even if they're chaff.

Re: AI spell casting priorities
 

A thought on the bit of not spending gems after turn 5:

There's one case I disagree on this. A caster who was scripted to spend gems should still have those spells available even past turn 5.

One thing that could improve the scripting: In addition to "Cast Spells" there would be "Cast Buffs" and "Cast Offensively".

Re: AI spell casting priorities
 

I have to disagree, I think the non-gem spam spells are good defaults.

The ones that bug me are the Holy spells. It seems like one skelly anywhere on the battlefield will trigger banishment no matter how silly that is.

Re: AI spell casting priorities
 

Scripted to spend gems? How does that work Loren?