So, how is the new patch K AI?

[Lone Wolf]

So, let's start talking about the new AI update.

I'm playing only one WM game so far, but from what I see, economically the new AI is vastly better then the pathetic horror we had before the merge.

From the flavour/roleplay side of things the new AI is better too, managing to escape the Kilmorphworld phenomena.

Military-wise, the situation is a bit more strange. In some ways, the AI is vastly better (at using heroes, for example), however it occasinally can produce weird results (in my WM game, half or so of Jonas' vast army are adepts/mages, for example, making him weaker then he looks).

So, compared to pathetic pre-J AI and to raw, unfinished, J, the new AI is vastly better.

However, compared to Turin's AI + some of my tweak, the improvement, as I've founded, isn't that noticeable. Sometimes the slight military weirdness makes itself noticeable. The mililtary magic improvements aren't that huge - previous AI spectred, pitbeasted, RoF'ed and Maelstromed me too, however, the peaceful magic use is a huge improvement.

So many of AI routines moved to dll isn't good for XML/Python tweakers like me. On the other hand, the majority of FfH players aren't XML tweakers.

So, my verdict so far - Vast improvement over what was before for the majority of players, a small improvement for me personally.

 

[Angrenost]

I can't comment on the military side or the AI spellcasting yet but they appear to be quite slow in expansion and culture building. I'm having a crazy emperor game using Loki atm and my brother's saying the same. AI seems to handle economics well.

 

[Lone Wolf]

Some occasional weirdness in expansion is also possible. Also, it seems that Kael didn't merge Sephi's XML tweaks since 5.0, since the -25 buggy threshold is still here.

 

[Sephi]

you should play a few games with the new AI before you give feedback. Also why do you start a thread about the FFH AI and then only talk about modmods? <snip>

Which AI routines in FFH were moved to the DLL that weren't there before? I don't know a single one. In fact there are several new XML tags to modify the AI.

 

[Lone Wolf]

Which AI routines in FFH were moved to the DLL that weren't there before?

FavouriteReligion tag, for instance.

you should play a few games with the new AI before you give feedback.

That's why I said so far.

 

[Sephi]

FavouriteReligion tag, for instance.

that tag still exists. Any other example you got?

 

[Lone Wolf]

that tag still exists. Any other example you got?

It exists right now, but these XML files are clearly the mess-up of Kael merging your dll and python, but not your XML. You can notice that no new tags of your AI were introduced in patch k, despite Python in k referencing them.

 

[Tlalynet]

Disciples of Archeron are cool, but they don't cast spells to defend the city if its under seige. I think its part of the whole stockpiling warriors then sending them out code, it doesn't activate the spell casting when they are on defense.

I attacked them early with cats and wood golems, when the disciple of Archeron appeared I thought I was toast, but it just sat there for three turns while I approached the city and broke down its defense, and died when a wood golem moped it up.

The economy appears better and a lot of things look really great, I really appreciate the patch K AI. I'm not going to vote on how much better it is till I've had a few full games though. I thought I should mention the Archeron thing as soon as I saw it.

*Edit
Also Eionian founded both Order and Empyrean and proceeded to convert as often as he was allowed too. It would have only been funny, but he kept built Valen and kept converting...

 

[aimlessgun]

Another 'so far' post here. Turn 685 on a marathon game, havn't really noticed any differences, except I do see a couple AI ships, which is nice. They have utterly failed to colonize nearby (not separated by ocean) islands despite being very constricted on the mainland though, which is not so nice. I remember in patch h (I think it was h) the computer aggressively colonizing. Difficulty seems the same, but obv. judging that based on 3/4th of one game is completely inconclusive.

 

[DaveGold]

Looks promising on 200 turns of one game. AI building early monuments, catapults, ships, avriety of units. Barbarians are acting less like tourists.

 

[Lone Wolf]

Did Kael merge Sephi's unitinfos XML, where heroes are told to act properly (unitAI_hero)? Looks like not, based on unchanged leaderheadinfos. That's bad, comrades.

 

[autopsy-turvey]

what are the different options for AI and various patches? i've only ever played standard unmodded ffh2

 

[esvath]

I vote a little better so far because :

The civ AIs managed to establish their home base well. In turn 200 (normal game speed, lake map) AIs already have 3-5 cities. The most poor one was Illians, since Auric started in the middle of jungle. The most prosper one was Clan, since Jonas started in plains/grassland without any adjancent AIs. The others were in between, including my Varn Gosam.

The barbarian AIs also do much better in harassing me (and AFAIK other AIs except Clan). They sent goblins, lizardmen, warriors and skeletons in waves I have not seen before.

The problems (?) that made me not yet voting that the AIs is vastly better now are :

about religion founding :
Looking at GameUtils.py, I know that AIs using Leaderheadinfos.xml to determined which religion can not be researched. Any negative value there means that AIs won't research the corresponding religion. And this is okay for me.
Looking at same GameUtils.py, I saw that AIs choose which religion will be researched using favoritereligion tag in Leaderheadinfos.xml. The problem is, no leader has a favorite religion. All favoritereligion tags in Leaderheadinfos.xml value NONE. Thus, my question is, how did an AI choose which religion he/she will research among two/three/seven valid religions?
In my game, Capria founded RoK, even when she perfectly able to research Order. Jonas founded OO while able to research Veil. While maintain some degree of flexibility, I think religion-founding should be more lore-appropriate (except when dungeon popped a disciple, of course).

about animals :
the second question is about animals. Since barbs spawned and harrased me since turn 50 (more/less) and escalated all through the game, I recognised a very low-presence of animals in my game. I only scouted several wolves and lions. Even bears are rare. While no more bear-spams is a good thing but the rarity of all animals is rather unexpected. Can we tweaked the spawn-rate of animals? More preferable, make animals last until end of game with strength adjustment like barbs, or even spawn better animals at later games?

That is my impression so far. I am really glad with this AIs improvement.

 

[JonQ]

The AI is doing pretty well.
Have tested the new patch in 3 games, one I managed to get a TOM win with amurites, the others the map filled up so soon I hardly had time to tech bronze working and find copper.
On emperor they mostly have a huge lead since they seem to spam settlers like crazy, giving them less barbarian problems.
The first game I probably managed to get a decent empire since the barbarian dragon city created a huge unsettled area between me and the other civs.

I noticed alot of religion switching, the evil civs seem to always go for OO in most previous patches, but in my latest game the sheim actually adopted AV (after founding OO ofc)

Had a big stack of pyre zombies take a city of mine, sent in my valiant stack of 1 catapult and 2 swordsmen only to have the zombies head back home. Had time to send of a fireball from a hastily constructed gibbon gothia before they where out of range.

They could have easily destroyed my civ, perhaps the AI decided not giving him KOE wasnt reason enough for genocide.

The AI seems to be using spells, but not seeing any particular strategy.

 

[Lone Wolf]

Looking at same GameUtils.py, I saw that AIs choose which religion will be researched using favoritereligion tag in Leaderheadinfos.xml. The problem is, no leader has a favorite religion. All favoritereligion tags in Leaderheadinfos.xml value NONE. Thus, my question is, how did an AI choose which religion he/she will research among two/three/seven valid religions?

The "favourite religion"'s in the *.dll, or supposed to be, at least. The FavRel tag is obsolete and is a result of Kael not merging Sephi's XML.

In my game, Capria founded RoK, even when she perfectly able to research Order. Jonas founded OO while able to research Veil.

Was that early? Civs are supposed to found early religions for happiness and culture and be lore-appropriate later. But like I said, maybe the incomplete merge causes AI screwups.

 

[esvath]

OIC. I am downloading Wildmana now. It might be insightful to compare the structure of these two

By the way, Lone Wolf, may I steal your avatar for a spell icon in my personal modmod?

 

[Lone Wolf]

By the way, Lone Wolf, may I steal your avatar for a spell icon in my personal modmod?

I'll be only honored by that.

 

[Valkrionn]

The "favourite religion"'s in the *.dll, or supposed to be, at least. The FavRel tag is obsolete and is a result of Kael not merging Sephi's XML.

Not really, the Favourite Religion tag in the xml is used in the DLL for processing. That's how all DLL code works; Why would Sephi have hardcoded the religions into the DLL when it's actually EASIER to make a new tag and enter it via xml?

 

[Lone Wolf]

Not really, the Favourite Religion tag in the xml is used in the DLL for processing.

Yeah, it's still there. Strange, didn't look hard enough, apparently .

But since it's the dll that influences favrel now, doesn't that mean that the new tag doesn't do anything now? Or am I misunderstanding the new religoin adoption mechanics?

 

[Tlalynet]

Or is the Fav religion all thing in the XML causing the religion flipping glitch that three people have reported now?