Great game partly ruined by extremely poor AI

Home Forums Age of Wonders 3 Discussions Great game partly ruined by extremely poor AI

This topic contains 162 replies, has 14 voices, and was last updated by  Graxx 2 months ago.

Viewing 30 posts - 91 through 120 (of 163 total)
  • Author
    Posts
  • #256423

    Motasa
    Member

    Not sure why we get different results?

    King or Emperor AI and strong starting army always yields better units for the AI on my game as well. To check this, try Classic Turns and on your first turn use the cheat “Barentz”.

    #256427

    Hiliadan
    Member

    How do you think I checked the AI units? Of course I used barentz…
    I’m talking about Normal Starting army and Knight AI (EDIT: I mistakenly wrote Emperor initially, that was incorrect /EDIT). JollyJoker said it applied to all difficulty level and all Starting armies.

    And sorry but it’s not because the AI moved during turn 1 that 300 worth of units can disappear.

    And obviously I didn’t use any mod.

    • This reply was modified 8 months, 1 week ago by  Hiliadan.

    The AoW3 Wiki, including a video library!
    Improve AoW3 with the balance mod (presentation video)
    The Battlefield.com/AoW3 the community website for AoW3

    #256428

    Motasa
    Member

    How do you think I checked the AI units?

    I’m not assuming anything 😀

    I reckoned you would, but you could just as well (and more easily) get a small map, 8 players and team yourself up with 6 of them and also see their concealed units (i.e. Frostlings, now I will always keep their Arctic Concealment in mind and Goblins with their Cave Concealment) and safe yourself a bit of time.

    I did some small testing. Loaded up a couple of maps and got quite consistent results. I present you the numbers of units a player gets (including leader, excluding first hero presenting to join).

    Weak starting army:

    Myself: 5 (mostly 3 irregulars and a cavalry)

    Lord: 6
    King: 7 (with sometimes an outlier to 8)
    Emperor: 8

    Medium or Strong starting army:

    Myself: 7 (always a full stack and a cavalry)

    Lord: 8
    King: 9 (with sometimes an outlier to 10)
    Emperor: 10

    Battle starting army:

    Myself: 13 (two full stacks and a cavalry, an odd chance of just 12)

    Lord: 14 (with sometimes an outlier to 15)
    King: 15 (with a moderate amount of AI getting 16)
    Emperor: 16 (with sometimes an outlier to 17)

    Seems quite clear the AI gets progressively a better larger starting army when you up their difficulty.

    • This reply was modified 8 months, 1 week ago by  Motasa.
    #256430

    Hiliadan
    Member

    You should calculate the gold value. The number of units is not the best indicator.

    And +1 unit is not coherent at all with the +300 gold indicated by JollyJoker.

    I rechecked my test, for some reason it was a Knight, not an Emperor. I did one more test and I got the same results as before. 7 units with a value which may be a bit higher but seems similar (I didn’t calculate the values, but the differences seem to be accounted more by randomness than a bonus, e.g. some have T3, others don’t).
    So correction:

    I checked a bit, and it’s simply that the AI gets a FLAT addition on all difficulty levels which is actually the same on each and would be something around 300 +10%

    ==> the AI gets a bonus depending on the AI level, whose value needs to be confirmed but might be something like:
    – Squire or Knight = 0 (maybe malus for Squire)
    – King to Emperor = x to 300 for Emperor (to confirm)

    Would be interesting to get the exact values and write that in the Wikia.

    According to your figures Motasa, the difference seems the same regardless of the Starting army (always +3 units for Emperor).

    • This reply was modified 8 months, 1 week ago by  Hiliadan.

    The AoW3 Wiki, including a video library!
    Improve AoW3 with the balance mod (presentation video)
    The Battlefield.com/AoW3 the community website for AoW3

    #256433

    Jolly Joker
    Member

    I checked a bit, and it’s simply that the AI gets a FLAT addition on all difficulty levels which is actually the same on each and would be something around 300 +10%

    ==> the AI gets a bonus depending on the AI level, whose value needs to be confirmed but might be something like:
    – Squire or Knight = 0 (maybe malus for Squire)
    – King to Emperor = x to 300 for Emperor (to confirm)

    Would be interesting to get the exact values and write that in the Wikia.

    According to your figures Motasa, the difference seems the same regardless of the Starting army (always +3 units for Emperor)

    .When I said DIFFICULTY LEVELS I didn’t mean Knight or Lord or King or Enperor, I meant STARTING ARMIES. The AI gets the same bonus, no matter the strength of the STARTING ARMY. If you reckeck what I wrote, I explicitely mentioned that the advantage would be biggest on WEAK starting armies (not on Knight or whatever).
    In any case – who’d have thought that you’d set up KNIGHT difficulty game???

    In any case it makes sense that the AI does NOT get as goo a bonus on King or Lord or Knight or Squire.

    #256434

    Motasa
    Member

    According to your figures Motasa, the difference seems the same regardless of the Starting army (always +3 units for Emperor).

    Yes, the number of units at the start of the game is quite consistent. And that is the explanation of the higher starting army value the Emperor AI gets. As @Jolly Joker mentioned earlier in this thread, I also calculated a difference of around +300, whether the game is set on weak, medium or strong starting forces (and probably battle as well). This is due to the fact it gets those three additional units; an extra T3 or T2 and an additional Cavalry or Support unit or two quickly ranks up the army value with +300.

    You on weak starting forces: your starting army is valued around 220. The Emperor AI’s forces are valued around 520-550. On medium: you 400 vs. Emperor AI 700-750. On strong: you 600 vs. Emperor AI 900. It should come as no surprise that the value on King will be approximately +200 (is 2 extra units) and on Lord +100.

    • This reply was modified 8 months, 1 week ago by  Motasa.
    • This reply was modified 8 months, 1 week ago by  Motasa.
    #256437

    Hiliadan
    Member

    who’d have thought that you’d set up KNIGHT difficulty game???

    I guess that’s the default and that was reset when I reset the settings after using the Arena or something. As you know, I don’t play SP any more, only PBEM, so I don’t really care about the AI difficulty.

    It should come as no surprise that the value on King will be approximately +200 (is 2 extra units) and on Lord +100.

    Ok, could you please try to confirm that +200 and +100 and check for Squire if it gets -100?
    Also JJ mentioned more than 300, so is it really 300 or 330 or what?

    The AoW3 Wiki, including a video library!
    Improve AoW3 with the balance mod (presentation video)
    The Battlefield.com/AoW3 the community website for AoW3

    #256439

    Jolly Joker
    Member

    You know that I’ve been asking the devs about a couple of values recently, and I’ve been experimenting a bit yesterday with those values. As it happens you can influence starting units in general with these (although within the borders of some hardcaps it seems).

    The default unit evaluation modifier is 0.9 which produced the values I reported (300-325 in one game). I’ve been toying with that (otherwise I wouldn’t have noticed), and the current setup I have uses the value 1.2. instead, and I have bonus values as low as 225 (2 Brew Brothers and an ADventurer) to 290 (an Orcish Necro).

    Of course, it’s not the change of starting forces I have in mind, but AI behaviour in general, which is really QUITE different, depending on the settings you have.
    However, this is a really difficult thing because the AI has some general routines it follows. For example, when the AI builds a settler, it wants to escort the settler (and consequently it fractures its forces, which means clearing things come to a halt (for the moment). You can mod the criteria for the AI to settle, but that’s difficult as well, because someone forgot to add that the evaluation of settling sites should be based on CLEARED sites, not on sites in general, because when you increase the conditions for sites the AI simply builds around the structures with mythical difficulty, since they have so much value, but will be cleared only in the farest future.

    Anyway. You do NOT want to alter the unit production value multiplicator without a lot of experimentation, because with that one you change a lot more than just starting values for everyone, but also strategic AI behaviour. And the default value is 0.9, as I said. Now, with all these values, if you set one, you cannot expect that there always IS a fitting value, what with the +-10% for Dwarves/Goblins, the Necro bonus and of course the limits with a view on HOW MANY units a starting or bonus force should consist of, so that means, there MUST logically be a spread value which I would expect to be +- 10%. With my current value I have values between 225 and 275 which would support that the base value would be 300+- 10%. So the default setting should be something like 333 +-10% which is 300-365, but I have to see a T3T2T2 bonus yet.

    • This reply was modified 8 months, 1 week ago by  Jolly Joker.
    #256443

    Hiliadan
    Member

    The default unit evaluation modifier is 0.9

    This looks very much like the modifier that calculate the gold value of defenders on site and quests (which is also 0.9). If I were you, I would not touch that, as the consequences could be very big. Or at least, check you still get the same defenders in site/quest.

    So do you plan to check the exact values of AI bonus for the different difficulty levels (Knight, Emperor, etc.)?

    The AoW3 Wiki, including a video library!
    Improve AoW3 with the balance mod (presentation video)
    The Battlefield.com/AoW3 the community website for AoW3

    #256444

    Jolly Joker
    Member

    The unit evaluation multiplier modifies EVERY unit value in the game, including defender strength. Example:
    Ancient Ruins: the default defender Strength is WEAK, which is 175 production points. Ancient Ruins have a default Modifier of 1.2 and an ideal number of 3. The global multiplier (0.9) is 10/9, so on normal defender strength, the multiplier is 4/3. *175 PP is 233, equaling, say, an Ogre and 2 Boars. On Strong Defenders, you get +1 unit and a multiplier of 1.5, so we are at a multiplier 2, making it 350 points.

    If you change the global multiplier (I did it in more than one way), say to 1.2, general strength DECREASES. For example, Strong Defenders in ancient Ruins gave me 1 Ogre and 4 Boars.
    Of course, lowering the multiplier increases defender strength…

    Since I’m not really interested in lower difficulty levels, there are no EXACT values (because it may be impossible to hit the EXACT value) and there is no way to check, but only to GUESS, I won’t.

    #256445

    Motasa
    Member

    Also JJ mentioned more than 300, so is it really 300 or 330 or what?

    It varies. In my previous post I mostly compared me playing the Human race to the AI playing the Human race. And also did some comparisons between other similar races (i.e. same prices for the different unit types); the differences are similar in that the Emperor AI gets roughly +300 more. But if you would compare the Human race to say the Tigrans, who have more expensive units, the difference may be greater. But overall there is a large variation, for example: my Tigran strong starting army has a value of 650, the Emperor AI playing Tigran gets values of 920, 820, 855, 855, 945, 940, 895. Average = 890. So in this case it isn’t +300, there is a lot of deviation. On Lord, with one extra unit, the AI may get an archer, but may just as well get an T3. So the average would presumably be around +100 on Lord and the median may well be +90 (since the likelihood of the AI getting a Support or Cavalry unit seems higher).

    #256446

    Hiliadan
    Member

    I doubt the bonus is expressed in number of units or race dependent. It is much more likely to be expressed in gold value, as all the rest in AoW3. The variation you describe are just a +/- 8% or something like that around the medium value.

    I guess I’ll have to do the test myself to input something into the Wikia…

    EDIT: by the way, I think I read somewhere the gold you don’t get into your unit is compensated in the closest gold stash. Maybe the AI also gets compensated somehow for the gold it doesn’t get through its units because it can’t reach exactly “900” (or “890”) through a combination of units.

    • This reply was modified 8 months, 1 week ago by  Hiliadan.

    The AoW3 Wiki, including a video library!
    Improve AoW3 with the balance mod (presentation video)
    The Battlefield.com/AoW3 the community website for AoW3

    #256448

    Motasa
    Member

    I guess I’ll have to do the test myself to input something into the Wikia…

    Good luck with that. It is painstakingly bothersome to restart a map and count the values of all the units times 8.

    But if your at it, why not have a look at the AoW_GameLogic.rpk pack in Age of Wonders III\Content\Title\Packs. Are these spawn markers the things I guess they are? It looks like here is the information on the starting army value for the human player. Still have to find the multiplier for the AI difficulty settings.

    #256449

    Jolly Joker
    Member

    The standard Cav isn’t included in the starting armies, but a hard-coded thing. No matter what you get, 1 Cav will be included. Also, MAXIMUM number of units is set: 3 on weak starting armies, 5 on medium, 5 on strond and 6 on battle. Also the values are set:
    160 for weak, 350 for medium, 640 for Strong and an additional stack of 6 with 700 on battle.
    Unit value is multiplied by the global Unit Value multiplier, though.
    The Ghoul factor is 1.3 on weak and medium and 1.1 on Strong and Battle.

    I don’t see any reason to assume that the starting armies for the AI would work different than the Extra army thing. There will be an Extra Army Lord, Extra ARmy King and Extra Army Emperor, that will have a maximum number and a value. Since we have seen 4 units on Emperor, this will be max number on Emperor, and I don’t see any real reason not to assume a value of 300 (and 200 and 100, respectively, for the others).

    #256455

    Hiliadan
    Member

    Also, MAXIMUM number of units is set: 3 on weak starting armies, 5 on medium, 5 on strond and 6 on battle. Also the values are set:
    160 for weak, 350 for medium, 640 for Strong and an additional stack of 6 with 700 on battle.
    Unit value is multiplied by the global Unit Value multiplier, though.
    The Ghoul factor is 1.3 on weak and medium and 1.1 on Strong and Battle.

    Is this from the mod tools?

    AoW_GameLogic.rpk pack in Age of Wonders III\Content\Title\Packs

    Did you check that?

    I got 1220 gold + 130 mana starting army on Battle. Mana counts as x2 in sites and Quest’s defenders, I assume it’s the same here, so that’s a value of 1220+130*2 = 1480.
    If I understand correctly your figure, the maximum should be (640+700)/0.9 = 1489, which seems to match.

    My results so far:

    Gold 1380 1635* 1410 1350 1525 1520 1368
    Mana 230 175* 240 200 220 170 180
    *: Necro

    By applying (gold+mana*2)*0.9 (and /1.1 for Ghoul units, but not for units like Death Bringer, Reanimator or Cadaver), I get the following values:
    1656 1649* 1701 1575 1769 1674 1555
    Average = 1654
    If we compare that to 640+700 = 1340, it’s a bonus of about 300 indeed. (314)

    I’m going to check for the other difficulties.

    • This reply was modified 8 months, 1 week ago by  Hiliadan.

    The AoW3 Wiki, including a video library!
    Improve AoW3 with the balance mod (presentation video)
    The Battlefield.com/AoW3 the community website for AoW3

    #256458

    Jolly Joker
    Member

    Is this from the mod tools?

    Yes.

    #256461

    Hiliadan
    Member

    Results for King:
    Gold: 1370 1450 1400 1515* 1560 1440 1325.5
    Mana: 240 200 210 145* 170 210 198
    *: Necro

    After calculating the Total value with multiplier ((gold+mana*2)*0.9); divided for Ghoul units by 1.1 if Undead for Battle and Strong starts (1.3 for Weak and Normal), I get an average of 1610.
    That’s 270 more than the value for humans. So looks like we’re not at +200, but I guess we’d need more valued to be sure. (also I may have some incorrect values because I’m using cost values from v1.4 of AoW3)

    I’ll check Lord and Squire later. I updated the Wiki with the current data on bonus starting army here: http://age-of-wonders-3.wikia.com/wiki/AI
    I’m going to create another page on settings and write there what we know about starting armies.

    By the way, I corrected the post above (Emperor) and applied the /1.1 only to Ghoul and not to Undead but not Ghoul units.

    The AoW3 Wiki, including a video library!
    Improve AoW3 with the balance mod (presentation video)
    The Battlefield.com/AoW3 the community website for AoW3

    #256462

    Hiliadan
    Member

    I’m going to create another page on settings and write there what we know about starting armies.

    http://age-of-wonders-3.wikia.com/wiki/Settings

    @Jolly Joker: did you see anything about how many additional units the AI is allowed? If there is a cap on maximum number of units, it seems the AI breaks it. You mentioned that but didn’t seem to have spotted any value for the extra number of units the AI is allowed, right?

    The AoW3 Wiki, including a video library!
    Improve AoW3 with the balance mod (presentation video)
    The Battlefield.com/AoW3 the community website for AoW3

    #256463
    #256466

    Jolly Joker
    Member

    That’s easy to check, you just need to reduce the general multiplier. Say, you pick 0.3, then you get units worth 900 – or less, due to slots.
    I did that, and it seems that Emperor and King offer a max of 4 units, Lord a max of 2, and Kbight none.

    #264438

    Hiliadan
    Member

    We’ve worked on many stuff on the balance mod and are getting closer to the point when we’ll be able to work on an AI patch. So I’d like to list the stuff we can actually edit:

    Strategic AI
    – maximum number of summons per spell (e.g. Banshees limited to 4) – Max AI sustained at once
    —- In particular: Banshee and Dread Reaper are currently limited at 0. The max of different class scout units should be reviewed compared to other summons.
    – maximum proportion of one given unit (currently 20%)
    – change research priorities (lower CP upgrade priority and increase Empire Upgrade priority for Warlord and Dreadnought)
    – change building priorities (including for units: BuildFrequency)
    – change hero upgrade priorities
    – remove Dispel for AI (to avoid Dispel spams)
    – upgrade the AI tactical (defense, resistance) bonus for Strong and Very Strong Defenders (require a mod by strength though)
    – delay research of the Disjunction spell a bit and its cost on every spell to prevent the AI spamming it like crazy, yet increase its effectiveness a little to counter the ease of reinforcing spells
    – boost buildings’ bonus because the AI builds most of them

    Tactical AI
    – reduce obstacle damage (e.g. of Fireball to avoid the AI using it on a Wooden Wall instead of units)
    Die to Unit Score Penalty (value the AI adds to target evaluation scores if the actor will die attacking a target unit)
    – priority of some spells like: Call Ancestral Spirit, Summon Siege Engine which do not seem to be used

    AI boosts
    – hero upgrade with XP boosts for AI heroes
    – Avatar skills “Free for AI” and “Research Type” – never (cannot be obtained by players)
    – sneaky healing
    – starting army boosts
    – income boosts

    Could someone please list the parameters that can be modified for tactical abilities (including attacks)?

    • This reply was modified 2 months, 1 week ago by  Hiliadan.
    • This reply was modified 2 months, 1 week ago by  Hiliadan.

    The AoW3 Wiki, including a video library!
    Improve AoW3 with the balance mod (presentation video)
    The Battlefield.com/AoW3 the community website for AoW3

    #264480

    Jolly Joker
    Member

    I’m working on an AI mod as well, and since I now have the feeling that on the strategic level at least things have been tweaked as much as moddable, what I can pretty safely say, that in my mind an EXHAUSTIVE AI-mod will not work for each and every setting, game set-up and mod-support. I’ve stopped publishing my mods, since they make sense only in connection with other mods I use and play, and the AI mod is no exception. The more specific a mod becomes, the higher the likelihood of failure when basic settings are changed.

    If you want to make an AI-mod, my advice for you would be to determine a setting first (map size, number of players and amount of sites for “general density and game length”, settling or not, defender strength, starting conditions, max hero level, if any, and of course mods to be used anyway and no matter what). Then you can make changes and compare results.

    A very central problem is that the AI plays on different levels depending on certain parameters. Give it an underground start, and it already plays with a handicap, since it is not “happy” in the UG and wants out. 🙂 It also seems the AI is not playing with some classes better than with others, and an AI mod may have to consider that and add “UG help” and “class help” in addition to general things.

    #264481

    Hiliadan
    Member

    Could you please be more specific? Why is the AI unhappy with UG? Do you have a changelog?

    The AoW3 Wiki, including a video library!
    Improve AoW3 with the balance mod (presentation video)
    The Battlefield.com/AoW3 the community website for AoW3

    #264484

    Jolly Joker
    Member

    With “Unhappy” I mean that even the UG races Goblins and Dwarves strive to settle OVERground, at the expense of the exploration and exploiting of the UG (where they would actually profit from their good speed). AIs won’t build UG empires, although they are much easier to come by and to maintain and guard for those races.

    For you, and everyone else who wants to mod the AI behavior, it’s important to have a clear idea about what you want to achieve and for which purpose. If you do play human-only games, only the TACTICAL AI is important, since you will face it against independents, clearing sites. If you just see THAT aspect of the game, it’s essential to give killing/converting units highest priority, because independents can’t do more than disrupt players (and killing player units is a god way to do that).
    If you ad AI-Players, though, things change a bit, because you also have to protect an AI with long-term goals in mind.
    That’s also what the strategic part is about.

    My own AI mod, as it is now, is more or less tailor-made for my preferred settings and the mods I play with, and I’d say that it might not work as intended in a different environment.

    I think that is more or less true with all mods. I have no idea what settings the PBEM community is playing with, but obviously a change in site frequency will change the general availability of resources (more sites = more plunder) which will completely change the balance (and thereby the validity of a balance mod).

    As it is, most AI settings are somewhat questionable. For example, you can give a max number for spells (inclusing summons) for the AI. Now, say you look at Sorcerer AI which has a lot of summons available, the longer the game lasts. If you just look at Wisps, Phantoms and Node Serpents (leaving the rest out of the equation) – should you limit ANY of them? And how important is map size? On a small and even medium map you’d want a significant number of Wisps and Phantoms, while on bigger maps Serpents (and Horrors) are more useful. Also, density of mana-giving sites plays a role here – plus, how much mana should the AI use for spell-casting.

    That’s why I say, when you want to improve the Player-AIs, you gave to concentrate on a specific setting and study AI behaviour over the course of a couple of games by reviewing the situation with the barentz cheat every few turns.

    • This reply was modified 3 months ago by  Jolly Joker.
    #264486

    Hiliadan
    Member

    For you, and everyone else who wants to mod the AI behavior, it’s important to have a clear idea about what you want to achieve and for which purpose

    Ok JJ, we got that, you already said it in the previous post. There is no need to try to guess what we want to do, just describe what you did yourself… If you don’t want to disclose what you did, fine, but please don’t post walls of text for nothing…

    BTW, you did not notice but the mod we’re working on is not for the PBEM community, it’s PBEM & SP.

    The AoW3 Wiki, including a video library!
    Improve AoW3 with the balance mod (presentation video)
    The Battlefield.com/AoW3 the community website for AoW3

    #264488

    ddnation22
    Member

    Hey good people! I registered an account just to hop in on this thread and say a few things about this topic.

    First and foremost, to Hiliadan and everyone else behind the PBem mod, a huge shoutout and thank you. That mod really re-energized my ability to enjoy this game – I don’t have the time to commit to any kind of Pbem or MP games in general, so I’m left to SP exclusively. Unfortunately, though I’m far from a pro or anything, I’m apparently good enough that the AI, even a good 4 or so Emperors teamed against me were an absolute cakewalk.

    Lets not beat around the bush – the AI needs to cheat heavily in the game to keep up with a human player’s efficiency (particularly in leveling heros, prioritizing research, and scouting IMO). But it needs to do so in a subtle way that doesn’t come across as “fake difficulty”, and the game already does a good job of this IMO.

    I installed the following mod: https://steamcommunity.com/sharedfiles/filedetails/?id=786153595&searchtext=AI , and I’ve found it a good “baseline” for what should be improved about the AI. More production, better scouting (increased vision of units), more casting points, etc. I think with a few more tweaks, the AI could actually be a force to be reckoned with. The main things I would improve about the AI , if I knew even the first thing about modding this game, would be:

    1. Leaders cast Dispel less / not at all, supports use Dispel / Steal Enchantment more often.

    2. Heroes for AI get the “fast learner” buff, and perhaps get little bits of XP/turn for free if possible, so that their heroes get that raw power in extra levels to compensate for their overall more “aimless” use of heroes than the player. I don’t want to have to nerf heroes super hard in all my games, just because the AIs heroes fall so far behind in levels.

    3. The AI prioritizes researching and summoning higher level summons. The MOST frustrating thing for me when it comes to the AI is when they clear out treasure sites that offer the spells, “Summon kobold”, and “Summon dire penguin”. The AI just WONT stop making these crappy summons, even as late as turn 50+! The Theocrat AI also loves to do this with the Cherubs. Surprisingly, the Sorcerer and Archdruid AIs seem MILES better at this than the other classes. I wonder why?

    4. If it could somehow be coded that the AI necessitates forming “power stacks”, or some kind of “honor guard” thing, where it forms stacks of its most powerful units and has them accompany heroes/leaders – and, most importantly, DOESN’T split them up on the overworld map. This alone could boost the AI’s gameplay by a ton, if possible.

    5. The AI plays some classes really subpar, for example it just CANNOT play the Necromancer class. If I had to rank the classes in terms of how the AI can handle them, I’d go with Sorcerer > AD > Dread > Theo > Rogue > Warlord > Necro . People complain about waves and waves of Manticore spam for the AI, but I never see it in my games. The Warlord AI is ALWAYS lowest score, and seems to always fall cumulatively behind (besides the aforementioned craptacular Necro performance). I haven’t a clue what to do about this, or really how the AI scripts for the various classes are coded in any way.

    6. This isn’t AI related, but I’d also like a mod that could remove some of the most egregious RNG in the game. Some RNG is great, even Lucky is generally fine by me. The things that really irritate me are Critical success / failure disjuncts (just an utterly stupid mechanic), and some of the Cosmic Happenings, namely Planetary Alignment and Enter the Void. I love to play with the happenings, but those two really take the fun out of the game for me.

    Just a few of my many thoughts on this. Let me know what you think! You guys are awesome, thanks again for breathing life into this amazing game. The thing about AOW3 is, and I really believe this, though the base game has tons of flaws, it can easily be modded into one of the greatest strategy games of all time. And I’m all about tinkering with my game until I reach that!

    #264489

    Jolly Joker
    Member

    I have no problem to disclose anything, but it’s not as straightforward as it may seem.

    For example, you and also ddnation22 talk about giving heroes fast learner (or something like that). Why only heroes? Heroes are their own balancing problem, and I’ll quite probably half the “basic reaistance unit” from 20% to 10%, which would halve the effectiveness of all immunity abilities and artifacts. Heroes are generally too powerful units.

    In my balancing mod, it’s a lot more difficult to level up herpes whereas you can get medals for units faster. Facing an equally leveled Arch Druid is one thing; facing an equally leveled AD with a slew of Champion T4 monsters is quite another.

    Strategically spoken, settling is the most important thing, because on one hand humans gain via settling spam, while the AI is extremely hampered in this regard. The AI needs to keep 13 hexes distance between city centers, but need only 4 units for each town in order to found a new one. It’s obvious that this is really, really suicidal. For one thing, 13 hexes distance will blow up the empire, for another 4 units per town will leave towns too sparsely guarded. Keep in mind that with the additional bonusses the AI actually isn’t needing those towns as dearly as humans.

    Conversely, humans need to keep only 5 hexes distance (which is due to Fortresses).

    I put minimum distance for both to 11, eliminating the use of Fortresses for humans, but since the AI don’t use them, it’s just leveling the playfield. I also doubled units necessary to 8. This has another advantage insofar, that with 8 units per town, the AI has enough guards and escorts and can still keep exploration forces (with 4 that doesn’t work).

    A major field of work is the importance of specific researches (which includes Dispel).

    #264490

    ddnation22
    Member

    It would be FANTASTIC if you could or someone else could figure out about the specific researches Jolly Joker. That could singlehandedly fix a lot of the reason why the AI plays certain classes so poorly imo. If the AI could simply put its advanced economy, production and research towards a template of reasonable “optimal” decisions, I reckon that alone could fix much of the reason why the AI plays Necro/Warlord so poorly.

    Also, a script that makes the AI more likely to try conversion/ghoul abilities would be nice. The AI seems to make little to no attempt at it. It’s why, despite using most of the other class abilities fairly well and in general having a decent overall strategy, the Theocrat and Rogue AI don’t perform as well as the AD/Sorc on average.

    And it’s one of the main reasons why the AI is bad at Necromancer, it really doesn’t do any of the actual Necromancery stuff. It hardly even summons Dread Reapers in my experience, it just kind of spams crappy Ghouled units from cities without rhyme or reason.

    Also, your attempts at nerfing forts and buffing the AI’s city-settling script would be great as well, Joker. What I have right now, instead of that, is a mod that removes the cap of cities the AI will found. The original AI stops settling cities when it’s a certain number of cities ahead of the player – removing that is a nice buff to the core essence of how the AI is gonna challenge the player, through attrition and sheer production / city founding.

    • This reply was modified 3 months ago by  ddnation22.
    #264493

    Jolly Joker
    Member

    I repeat that the AI doesn’t need towns as much as humans, economically. Having too many towns will only lead to the AI creating points of vulnerability. Instead you have to make sure the AI has the means to actually defend against an onslaught, and if you have 8 units, you are a much more difficult target when you have one town instead of two.
    In effect the AI is doing you a favor when it spreads too fast.

    Speaking in mod terms, going through all the settings of how much emphasis the AI is supposed to put on certain techs and builds and spells, the problem is, that this amounts to handing the AI a specific strategy. If you influence an AI to go for specific things, you make it predictable AND you will either be too general (which is the case vanilla) because you have to serve too many masters, or you will have to make a different mod for different essential base settings.
    Necro is a special case.

    #264495

    ddnation22
    Member

    That is true. I don’t really want to tinker that much with the base AI in terms of research and spells, my MAIN problem with it is how it summons stupid, irrelevant crap even late into the game.

    Predictability is an issue, however… I already find the AI to be fairly predictable as it is. The classes that it plays the best (sorcerer/AD/Dread), IMO, it plays extremely predictably.

    If, for example, the AI was scripted to do something like: “I want to build supports, so I’ll build them with a Magic Academy in this city with a Lost Library I cleared”. And so on, for example with High Elf longbowman and the Focus Chamber. At some point, yea, it’s predictable. But it’s also a pretty rock-solid decision to make all things considered. That + prioritizing keeping strong units in stacks with heroes, and being more mindful about defending cities, are all quite “predictable” things, but still, they would help the AI perform better IMO.

    The city founding thing, and your take on it Joker, is quite eye-opening for me. A lot of the things I thought would make the AI stronger seem to actually weaken it, all things considered. Playing a map with a low amount of independent cities and less city founding FAVORS the AI, now that I think about it. “Spreading itself too thin” and failing to form power-stacks are two of its biggest flaws.

    EDIT: And I’d also like to mention that I’m looking for a mod that essentially gives the AI free defenders in their cities. Implementing that in an elegant way can help a bit with the AI’s tendency to not guard cities well.

    • This reply was modified 3 months ago by  ddnation22.
    • This reply was modified 3 months ago by  ddnation22.
    • This reply was modified 3 months ago by  ddnation22.
Viewing 30 posts - 91 through 120 (of 163 total)

You must be logged in to reply to this topic.