New experimental version

[Seravy]

Feel free to suggest a new condition but it can't be more detailed then the current "ranged>3" condition. There are literally zero free bytes left to use in the procedure. It also have to be broad enough to allow the AI to at least put it on something even in the early game when they don't have magicians to target.

[Nelphine]

The dreaded 'no space' strikes again.

Also yeah, magic weapons is just too common, and in most combats will be either equal to or better than focus magic (only if they, what, 9+ squares away? Will focus magic be better). Anything with .. Whatever ability catapults and longbowman have, long ranged? ... Will always end up worse with focus magic if they have magic weapons.

[Seravy]

I'm not sure if Focus Magic applies before or after magic weapons so that depends on that.

Something else. I noticed the AI is not attacking my settlers and empty floating islands. Not a surprise, these things have a 0 attack rating so their total rating is also 0 and a tile with 0 target value is considered empty.
What is a surprise is, I thought it was changed to "at least 1" a very long time ago - I'm quite sure the AI did attack my things...well, some time in the past. But I checked older versions and this was never implemented which I don't understand.

[Seravy]

Ran into some significant bugs today so one (hopefully last) update before the 5.0 release to be able to test at the full capabilities of the AI :

-Fixed : maintenance cost of Seismic Mastery and Fairy Ring were swapped.
-Fixed : Units with Life Steal were not valid AI targets for Blood Lust.
-AI can now target units that have a touch attack even if they aren't a priority choice for Focus Magic
-The AI is now not allowed to cast Focus Magic before turn 10 - it's too likely to find no targets and waste the spell, as well as significantly slowing down early expansion (80 can be used to make two hell hounds or a ghoul instead!)
-The AI is now allowed to cast Earthquake when Seimic Mastery is on if they don't have an expensive summoning spell available to trigger it.
-Fixed EXP14C bug : intercontinental attacks read the enemy land power from the matrix incorrectly, resulting in failing to attack targets sometimes.
-Fixed EXP14C bug : intercontinental attacks read the enemy strength of nodes/lairs from the matrix incorrectly.
-Fixed EXP14 bug : intercontinental stacks with a power divisible by 256 couldn't do intercontinental attacks.
-Each unit has a minimal AI rating of 1. This allows the AI to attack units with zero attack strength such as settlers or floating islands.

[Nelphine]

Just watched a chimera (in the aura of a chaos node) attack a magic spirit - and lose. I didn't think to have a save before it, so I can't offer it to you, but it seems very unlikely that spellcasting would make up for the difference in defense. However, the chimera was raised from the dead, so it was definitely the 6 death/5 life books that did it.

The opponent has 5 chaos/4 life. Is it possible there's something messed up with chaos/death books in strategic?

Confirmed it with my own game/troops. 1 chimera (5 chaos/1 life, 218 casting skill) vs 1 regular magic weapon nomad swordsmen (7 death/5 life, ~225 casting skill); the chimera loses every time.

Given that the chimera defensive rating is 4*8*4*(5+4) = 900 + 180 + 72 = 1152
And the swordsmen defensive rating is 4*1*6*(3+4) = 168
And the swordsmen ranged attack rating is (3-2)*6*20*1.32 = 158.4

And the casting skill does:
Chaos : 120 Attack
Life : 30 PercentATK, 50 Defense
Death : 90 Attack, 20 Defense

Each "Attack" adds 1 attack power for each (mana spent*books owned*attack for realm)/6/5.
Each PercentageATK adds 1% of the existing attack power for each (mana spent*books owned*attack for realm)/20/5.
Each "Defense" adds 1 defensive power for each (mana spent*books owned*attack for realm)/5/5.

Mana is spent over 6 turns.

So the AI is getting 90*(225/6)*7/30 = 21*37.5 = 787.5 attack rating from death
And 30*(225/6)*5/100 = 56.25% increase on the nomad attack, which is 89.1, from life
Therefore, the AI has a total of 1035 attack rating on the first ranged turn.
The AI gets 20*(225/6)*7/25 = 210 defense rating from death.
And 50*(225/6)*5/25 = 375 defense rating from life.
Therefore the AI has a total of 763 defense rating on the first ranged turn.

My chimera gets 120*(218/6)*5/30 = 726.666 attack rating from chaos.
And 50*(218/6)*1/25 = 72.666 defense rating from life.
Therefore, the Chimera has 727 attack rating on the ranged turns.
And 1225 defense rating on the first ranged turn.

The Chimera shouldn't die from 6 ranged (with spells) combat rounds; nor after the next 4 ranged rounds ...

Oh, its the extra defense from death/life. Without getting any extra defense, chaos is getting brutally shafted in combats with low numbers of units.
The next 5 rounds of spells means the nomad swordsmen ends up with another 2925 defense rating.
The Chimera only gets 363 more.

Assuming average damage, that means the Chimera is around 1568-621-63 = 884 defense rating once melee begins.

The nomad swordsmen on the other hand is around 3688-436 = 3252 defense rating once melee begins.

The original Chimera melee attack rating is [(10-2)+(6-2)]*4*20*1.32 = 1267.2
The original nomad swordsmen melee attack rating is (4-2)*6*20*1.32 = 316.8

Of course both of those numbers should be modified based on how much damage they've already taken. The nomad will obviously have been reduced a lot less than the chimera.

Without any reduction the Chimera should win, barely. Since the nomad has less reduction, the nomad wins handily.


Very strange, but working as expected.

The defense rating from spells really goes wonky on combats with low numbers of units, as shown above.

[Nelphine]

Based on the above post, we may need to modify the defense stats for the realms. For one thing, attacking a neutral/node/lair, once the wizard has 100+ casting skill, will virtually always be a win (unless the attacker is close to mono chaos) simply due to the extra defense, and the AI assuming it can only attack with rough parity in the first place. This will also cause results such as Chimera (which actually has a pretty good strategic rating for a summoned unit) being defeated by Magic Spirits/Settlers/Engineers, simply because the defense being added is worth 3 units of chimera.

For a comparison: The one life book my Chimera is using, is giving roughly as much life as the 5 chaos books my Chimera has are doing damage. So 50 spell defense is actually worth almost 600 spell attack in the current strategic combat formula. (Although strictly it's not quite that good, since spell attack also reduces the opponents attack strength, whereas spell defense does not increase the defenders attack strength, but except in cases like this with 1 unit, the attack strength removed by the spell attack is not going to be a significant portion of their overall attack strength. So 50 spell defense is probably close to 575-590 spell attack.)

Unless I'm using outdated formula in the above numbers. I can't remember if I already included your multiplier to spellcasting damage in strategic combat. I thought I did but maybe I didn't? (But if I am using outdated data, then why is the Chimera losing?)

I think I must not have the updated values - the values above would result in the Chimera not being raised from the dead, and he definitely is being raised from the dead. So the death magic has to be doing more damage than that.

Edit: If I remember correctly, you basically multiplied spell damage by 6 for strategic combat.

That's going to end up with 50 spell defense being closer to 95 spell attack. Which means Chaos is still weaker than the other realms, even discounting the wide variation on the percentATK realms.

Given that chaos is meant to be The Combat spell realm, I think that probably needs to be fixed. I'd change chaos to 150 attack strength.

Then the relative realm strength (under the assumption 30 percentATK is equal to 50 spell attack) would be chaos > sorcery/life > death > nature.

Lets tweak a bit more. I think it should look like Chaos > Death > Sorcery > Nature/Life.

So, Chaos to 140 attack, sorcery and life to 40 defense, life to 25% percentATK, no change to nature or death.

(This would result in chaos at 140, death around 128, sorcery around 126, life around 118, nature around 114.)
(For further comparison, the current scale is life/sorcery around 145, death around 128, chaos around 120, nature around 114.)

Probably should only increase Chaos to 135 attack. Might also want to increase Nature to 55 attack.


TLDR: Increase Chaos spell attack to 135. Increase Nature spell attack to 55. Decrease Sorcery spell defense to 40. Decrease Life spell defense to 40 and life percentATK to 25%.

[Seravy]

Weird but I have lost a Chimera to a Swordsmen in normal combat before. With enough buffs on the swordsmen and recurring Healing, Life can beat it while Bless prevents the Chaos wizard from affecting it.

...Albeit it might have been a halberdier, idk.


Looking at the numbers, 1 Attack bonus is worth 1 attack per 5 mana, and 1 Defense bonus is worth 1/5 defense per 5 mana. So attack is worth 5 times as much as defense.
However, each turn rolls ~5.5% if the attack as damage, so 1 attack needs to be about 18x higher than 1 defense to be even.
Now, realms generally have much higher attack values than defense values for this reason - but not 3.6 times more.

Percent attack is worth 1/20 % for each 5 mana so 20 points equal 1% more attack strength. Assuming a typical army being worth 2000 attack, this means 20 points yield 20 ATK which is perfectly even with the non-percent ATK bonus. Above this army size, PATK is better than ATK, below it's worse.

So we'll want to have all 5 realms set in a way that ATK+PATK+DEF*3.6=same.

Currently this sum is lowest for Chaos, highest for Sorcery.

...actually, we might not want it to be the same? Some realms have better combat spells than others? Or do they... I don't think that's true anymore, Life has great combat spells (healing, buffs), Nature too (ranged summons!) and the other 3 has always been good.

But let's also consider that attack is better than defense if the amounts are otherwise equal - attacking eliminates enemies, reducing their attack strength while boosting defense does not. So I think using "3" instead of "3.6" might be more fair.

[Seravy]

So, how about this ?

   

Explanation :

Nature can summon more units effectively for a semi-decent DEF bonus.
Nature can effectively protect its big units with some defensive buffs to make them fight longer, but doesn't excel in this area. So a bit of PATK. Note that DEF doesn't really do this because it's a +X addition instead of a +X% percentage bonus.
Nature is good at dealing direct damage through summoning archers, using direct damage spells or Call Lightning, but not as good as others. So medium ATK.

Sorcery is good at summoning stuff and preventing damage to its units (high DEF)
Sorcery is good enough at protecting units to let them fight significantly longer, dealing proportionally larger damage using them (PATK) - in fact it's as good as Life. It's also good at turning enemy units against themselves (not really PATK but the closest option from the 3)
Sorcery has medicore direct damage options (low ATK).

Chaos spells deal a lot of damage. It doesn't have summons or defensive spells. So pure ATK.

Life is best at keeping units alive - healing and buffs. Those units will then do more damage as well. (DEF, PATK)

Death is great at direct damage (ATK), and can make it slightly harder for enemies to hurt their units through curses (DEF) but this is not as reliable as to contribute to the offensive capacity of its armies significantly (no PATK).


And this way all 5 realms have a unique distribution that's quite different from others.

[Nelphine]

I actually think life is much better at patk than sorcery due to healing and raise dead and invulnerability. The actually units fight much longer.

On the other side, confusion and black sleep are raw damage - they remove the enemy unit from the fight regardless of how much attack the sorcery side has. Sorcery also has loads of combat summons - raw damage and defense, regardless of how strong the original sorcery units are.

Sorcery only has patk via resist magic and magic immunity and invisibility. Significant, maybe, but I'd say that's much less than nature, who have web, entangle, web, iron skin, web, and web. (As you can see, I believe web is simply amazing for patk, but certainly not as good as life).

Other than that, yes I agree. I'd be fine with sorcery being only attack and defense. Their defense is enough higher than death that its still distinct.

[Nelphine]

One thing to note us that we are slowly raising the power of spell casting in strategic combat. Which I think is fine given how effective spells really are.