|
Thanks both, it’s just about figuring out a framework for it all I think.
|
#
?
Mar 7, 2023 09:43
|
|
|
|
| # ? Jun 13, 2024 05:51 |
|
|
Played a little of Planetfall again and it is infuriating how they sacrifice units to get kills. I just watched a full health enemy unit take 4 melee overwatches (stagger immune heavy unit) just to move to a spot where it could kill one of my low health units. I mean their chance of winning the overall fight may have been higher if it just stayed and hit the guy that was in melee range of him at the start of his turn. I really hope they tone the hard on the AI has for this down in Age of Wonders 4.
|
|
#
?
Mar 7, 2023 22:45
|
|
|
Hryme posted:Played a little of Planetfall again and it is infuriating how they sacrifice units to get kills. I just watched a full health enemy unit take 4 melee overwatches (stagger immune heavy unit) just to move to a spot where it could kill one of my low health units. I mean their chance of winning the overall fight may have been higher if it just stayed and hit the guy that was in melee range of him at the start of his turn. I really hope they tone the hard on the AI has for this down in Age of Wonders 4.
|
|
#
?
Mar 8, 2023 00:53
|
|
|
Hryme posted:Played a little of Planetfall again and it is infuriating how they sacrifice units to get kills. I just watched a full health enemy unit take 4 melee overwatches (stagger immune heavy unit) just to move to a spot where it could kill one of my low health units. I mean their chance of winning the overall fight may have been higher if it just stayed and hit the guy that was in melee range of him at the start of his turn. I really hope they tone the hard on the AI has for this down in Age of Wonders 4.
|
|
#
?
Mar 8, 2023 01:09
|
|
|
also broadly the ai is much better at scoring kills than it is at staying alive
|
|
#
?
Mar 8, 2023 03:28
|
|
|
Yes it is a good tactic for slowing down an enemy force in a war. But this is guards in a map site. In fights against other empires doing this makes more sense. I must admit I don't enjoy spite being programmed into a combat AI.
|
|
#
?
Mar 8, 2023 03:44
|
|
|
I do think that the fact that it is impossible to retreat from defensive combat skews how one tries to write a challenging tactical layer AI. OTOH I can't think of any good way to allow for retreat from defensive combat that wouldn't be super annoying for the player and/or easy for the player to abuse.
|
|
#
?
Mar 8, 2023 05:07
|
|
|
The spite play is logical, I can't fault them. Chip damage will be healed up within a few turns but a dead unit is dead forever. play more defensively if it's truly aggravating you. Otherwise accept that losses are inevitable in war.
|
|
#
?
Mar 8, 2023 05:46
|
|
|
Yeah, it sucks but any further and we'd be getting more complaints that the AI is too dumb or whatever. Thankfully a good solution is making it so the negative outcome (losing units) doesn't sting as much, and one way to do that is making it so replacing them isn't as painful, like say, not having to commit a city's entire production to training a new unit :v
|
|
#
?
Mar 8, 2023 06:05
|
|
|
Reminds me of the drop in campaign battles thing Total War tried for a bit, where some rando pubby would get the entire enemy army killed to murder your general. Really thats where stacking CC effects outside of stagger really pays off. Blinds, slows, morale penalties, even just reducing their resistance so that the overwatch attacks hurt more.
|
|
#
?
Mar 8, 2023 06:20
|
|
|
Beefeater1980 posted:Re planetfall, I’ve started playing it again properly now and have just got into the Amazon campaign. Well, I don't recommend rushing with T1 units. Not as Amazon anyway. Huntresses are generally kind of fragile and need to get relatively close. Lancers are T2, you start with them researched and just need to build the skirmisher barracks. Arborian Sentinals are also T2 but you need to research them first, and build a specialist barracks. So if you're rushing, consider those. If you're teching up for something else, you're going to want some placeholder T2 units to keep you going anyway. Turn by turn: Move my scouts around, paying attention to any hexes threatened by enemies (you can click on them and see where they can move to within 1/2 turns). Now is a good time to consider if I want to do any strategic operations - I may be able to do some extra scouting with a unit summon or monitor, or I might want to save my operation for army attacks/buffs/economic stuff. Move my armies around, keeping the above in mind. Don't let armies get surprise ganked. City management: Priority 1 - don't lose cities. Militia/towers are the most cost effective way to defend. Know where the nearby enemies are - consider building a scout if coverage is bad. Prioritize expansion either via colonizers, armies, or citizen growth (new sectors). If I'm close to being able to do one of these, go all in working to it via citizen management/saving up. Colonizers will eventually get too expensive however. Make sure to keep a reserve of energy for operations. Rush the pioneer rewards for things like builder. Last, build whatever cheap economic upgrades I might want. Research: Social research you can usually just wing it. Military research: Early on I just grab all the cheap stuff that gives me useful tactical ops or mods. Anything with stagger / resist stagger is good. Then, pick a single unit to specialize in, something durable and fighty, either a T2 or early T3. For Celestial Amazons that means Lancer, Arborian Sentinal, Bombardon, or Lightbringer. Once researched and producible (the lancer starts researched), go in depth researching mods for just that unit until you have a set of good ones you can use. Meanwhile keep building up the armies (economy permitting), and throw some support units in if needed. 1 "Army" can be 3-4 stacks due to the adjacent hex rule. Try to get an advantage, and attack as soon as you have it. If you can eliminate someone and take only one city, that's worth a few losses. Once you have a good lineup of 3 mods for your special unit, go for some powerful late T3 units or T4s.
|
|
#
?
Mar 8, 2023 07:05
|
|
|
This reminds me, any word if AOW4 retains Planetfall's garrison system or will that be too much when coupled with the new production/draft split?
|
|
#
?
Mar 8, 2023 07:10
|
|
|
toasterwarrior posted:This reminds me, any word if AOW4 retains Planetfall's garrison system or will that be too much when coupled with the new production/draft split? It looks like there won't be any buildings that specifically add garrison-only units. It's replaced by a siege system where a settlement has a defensive siege value that the attacking force has to whittle through over a number of turns before they can attack and take it over. Some units have traits that reduce the besieging time ("Siege Breaker"). Settlements can continue to build units while under siege too, I believe, so you may be able to build up a force before the battle, which spawns you into a map where walls are already broken I'm pretty sure. There are things sort of analogous to the towers in Planetfall too, it seems. It seems like a kind of interesting system. No more sniping cities with a single unit of grim crows or whatever, which is a good thing.
|
|
#
?
Mar 8, 2023 08:51
|
|
|
Ceramic Shot posted:It seems like a kind of interesting system. No more sniping cities with a single unit of grim crows or whatever, which is a good thing. Yeah, gently caress that. Hope this system works out, the Planetfall change was very welcome.
|
|
#
?
Mar 8, 2023 08:55
|
|
|
toasterwarrior posted:Yeah, gently caress that. Hope this system works out, the Planetfall change was very welcome. It looks promising to me! I think the dev diary that goes over siege stuff is coming up within a couple weeks. So happy this is coming out in less than two months. The eXplorminate YouTuber guy said from his 50 hours with the game that it's his favorite among the AoW series and maybe his favorite fantasy 4x ever or something to that effect. If you're a fan of this genre I think it's ok to be more than cautiously optimistic. I just hope they make their sales goals or whatever it takes to get more expansions post-2023. Is anyone here thinking of making a thread specific to AoW4?
|
|
#
?
Mar 8, 2023 11:55
|
|
|
I'd be wary of the Explorminate Youtuber guy to be honest. I've listened to his podcast for years and he's really enthusiastic about the genre but time and time again he falls for the hype and retrospectively admits he wasn't objective enough, I got the same vibes from his preview this time. That being said it does genuinely look really promising. I think the biggest risk is that all that customisation could lead to a certain lack of personality or cohesiveness to the game. Stellaris really suffers from it imo and I think it also held Humankind back.
|
|
#
?
Mar 8, 2023 13:17
|
|
|
I don't think cohesiveness or personality will be a problem from what I've seen. I can't speak for Humankind but the reason Stellaris suffers here is a combination of over a hundred graphical portraits with dozens of traits that are can be very minor and extremely variable in terms of impact on another player. When all that is randomly mashed together by an algorithm it becomes difficult for a player to parse any coherent identity from another encountered race, and the most distinguishing bits might not even impact the player at all. If in AoW4 as I understand, form traits are fixed unless the player or map designer specifically (and probably purposefully) changes them, and the societies seem visually distinct enough that between those 2, other encountered factions will already start with a distinctiveness that the player can easily recognize and form ideas around. The cultural traits are the closest at risk to the above though they seem kinda fine and honestly if the above form+society (and Leader) distinction is already working then some muddiness in the extra culture traits wouldn't really break things. I think the actual critique that Explorminate guy gives that rings plausble to me is that due to the adjusted structure of a game/faction development, the early game between different races/leaders could feel a little samey over time.
|
|
#
?
Mar 8, 2023 15:28
|
|
|
Still, sounds like the mid and end games are really engaging.
|
|
#
?
Mar 8, 2023 17:36
|
|
|
Autsj posted:I don't think cohesiveness or personality will be a problem from what I've seen. I can't speak for Humankind but the reason Stellaris suffers here is a combination of over a hundred graphical portraits with dozens of traits that are can be very minor and extremely variable in terms of impact on another player. When all that is randomly mashed together by an algorithm it becomes difficult for a player to parse any coherent identity from another encountered race, and the most distinguishing bits might not even impact the player at all. yeah that stuff rubs me the wrong way. When traits have only a minor impact on how you play a faction, they have effectively nil impact on how you interact with another faction with those traits. In Stellaris even some of the major traits dont really change up the game much. Years ago I took a stab at a Stellaris mod to make traits a lot more impactful but got distracted by another idea to make a defense resource that autobuilt platforms and never finished either of them. The traits shown off so far for AOW4 seem encouraging though. I think something like Bulwark or the one that lets you evade most damage from opportunity attacks are going to be *very* noticeable when playing against them, so long as the AI can take them into account. As will things like Mount traits or arctic spec.
|
|
#
?
Mar 8, 2023 17:51
|
|
|
I bet the AI can take advantage of them. The AI in Age of Wonders has historically been remarkably good about that sort of stuff. Gerblyn can no doubt give way better insights about how it works exactly, but the AI in Planetfall is able to adapt to and make use of all sorts of unit mod combinations, even ones that don’t just alter stats but add entirely new abilities. As the recent discussion about being willing to sacrifice units to secure kills in a losing battle shows, it’s good enough at it to be annoying! I really like that the AI does that though, it’s just one more way the game links the Tactical and Strategic layers.
|
|
#
?
Mar 8, 2023 18:49
|
|
|
It can be used against the AI, too. If you have a monster of a combat hero in planetfall, one that the AI would very much like to snipe, put the hero in a separate stack. With the right positioning, you can get the enemy to turn their flank to your main force for an easy win when you'd otherwise suffer attrition.
|
|
#
?
Mar 8, 2023 19:28
|
|
|
I also like the spiteful AI, because it lets me be spiteful in return. Oh I was just gonna sue for peace but you killed the vorpal sniper from my starting stack, so now I have to stay at war until every last one of your cities has been razed or purged, no offense.
|
|
#
?
Mar 8, 2023 19:34
|
|
|
at least pf's spite ai doesn't have the intense murderboner for specifically heroes that aow3's did. it just highly values killing blows rather than highly valuing "damage to heroes specifically" i think it also prefers to attack wounded units. to my understanding the ai doesn't look at its forces as a cohesive whole, it basically just cycles through the units one by one and works out what action that unit can perform right now that will result in the highest damage. so in theory it won't focus fire until a unit's health is low but it tends to focus fire as an emergent behavior of "that unit has the lowest defensive stats, hit it" -> "that unit with the lowest defensive stats is also missing a lot of hp, definitely hit that one"
|
|
#
?
Mar 8, 2023 19:47
|
|
|
ninjewtsu posted:at least pf's spite ai doesn't have the intense murderboner for specifically heroes that aow3's did. it just highly values killing blows rather than highly valuing "damage to heroes specifically" Makes sense given autoresolve’s tendency to throw Cavalry into the fray but not being able to get them out again
|
|
#
?
Mar 8, 2023 20:43
|
|
|
if you use only cavalry it works out!* *don't fight pikes
|
|
#
?
Mar 8, 2023 21:23
|
|
|
Yeah I find if you mod a support unit to be too highly (but not decisively) damage focused, it does a weird dance where it still overprioritizes buffs / premature heals, but will charge in deep on the odd turn it can get a shot in with nothing better to do, moving completely out of position and dying for it.
|
|
#
?
Mar 9, 2023 06:37
|
|
|
ninjewtsu posted:to my understanding the ai doesn't look at its forces as a cohesive whole, it basically just cycles through the units one by one and works out what action that unit can perform right now that will result in the highest damage. so in theory it won't focus fire until a unit's health is low but it tends to focus fire as an emergent behavior of "that unit has the lowest defensive stats, hit it" -> "that unit with the lowest defensive stats is also missing a lot of hp, definitely hit that one" Not highest damage, best score, which includes things like: "Damaging a unit with an AoE attack has a higher score than damaging a unit without one" "Taking damage to get to an attack location reduces score" "Killing a unit with 5 damage is worth more than doing 20 damage to a unit that survives" "Attacking from cover is better than attacking from the open" But yeah, it has no forethought and only minimal ability to coordinate actions between units.
|
|
#
?
Mar 9, 2023 11:08
|
|
|
Actual real life example! Posters here (and people from a User Research test we just did) just complained that the AI would suicidally run through multiple attacks of opportunity to get a kill. What's happening is something like: Damage From moving = 20 Damage done to the target = 10 Bonus for killing the target = 50 Score = 10 + 50 - 20 = 40 This score is probably better than hitting the guy next to you: Damage to the target = 30 Damage taken from retaliation = 10 Score = 30 - 10 = 20 So I'm adding code now that says: 1. Damage taken from moving down a path is now worth twice as much negative score 2. If you're not in spite mode, and the damage from moving is above a threshold (20? I dunno, it's all just by feel really!) then just subtract 100 from the score instead Now, the original calculation becomes: Score = 10 + 50 - (20 x 2) = 20 So the AI will probably do the other attack instead (also score 20) since the target is closer.
|
|
#
?
Mar 9, 2023 11:18
|
|
|
My favorite AI moment was tossing a stack of musketmen at a dungeon (I think, whatever the place with an enemy manticore rider is), and hitting autoresolve, being ok with losing units or even the entire stack, so long as they softened it up. Instead, the AI cleared it out with zero losses. Turns out musketmen have a decent melee attack and enough bulk that they can cover for another musketman while it reloads, and the replay made it look like a synchronized dance. Dunno if it was just luck or musketmen happen to be a really good unit that synergizes well with AI priorities, but since then I've always been using monotype stacks for battles I want to autoresolve and mixed stacks for manual battles, and it works a lot better than letting the AI ever try its hand at a stack with disparate units.
|
|
#
?
Mar 9, 2023 11:56
|
|
|
Muskets are the absolute best and AoW4 won’t be finished until we get them back. They’re Tier 1 melee infantry with guns, what’s not to love?
Triskelli fucked around with this message at 12:16 on Mar 9, 2023 |
|
#
?
Mar 9, 2023 12:07
|
|
|
Yeah, musketeers were just really strong. Their initial attack did so much damage that needing to reload and whatever other issues they had became a bit irrelevant, since the opening volley could pretty much wipe out anything. We had to nerf them 3 or 4 times because the Goblin Musketeer Doomtrain was just unstoppable.
|
|
#
?
Mar 9, 2023 12:20
|
|
|
The biggest misstep with AOW3 Musketeers was not giving their models bayonets, to make their surprisingly decent melee capability more clear
|
|
#
?
Mar 9, 2023 12:22
|
|
|
Musketeers are amazing, yeah. One of my absolute favorite units. It's not just that they're strong, they're really fun to use. Something about positioning correctly for a killer volley without having to worry (too much) about retaliation just tickles my brain in tactical battles. Also loved setting up enfilades with cannons.
|
|
#
?
Mar 9, 2023 14:31
|
|
|
ninjewtsu posted:at least pf's spite ai doesn't have the intense murderboner for specifically heroes that aow3's did. it just highly values killing blows rather than highly valuing "damage to heroes specifically" Zulily Zoetrope posted:I also like the spiteful AI, because it lets me be spiteful in return.
|
|
#
?
Mar 9, 2023 15:48
|
|
|
Triskelli posted:Muskets are the absolute best and AoW4 won’t be finished until we get them back. 
|
|
#
?
Mar 9, 2023 15:55
|
|
|
Age of Wonders: Napoleonic Wars, please
|
|
#
?
Mar 9, 2023 16:05
|
|
|
Dev diary about cities that just dropped: https://forum.paradoxplaza.com/forum/developer-diary/dev-diary-8-cities-and-expansion.1572753/ Seems there are some militia buildings after all!
|
|
#
?
Mar 9, 2023 16:21
|
|
|
Calling the recruitment resource draft does give me the amusing imagery that all your cities have a bunch of dragons and such just chilling as normal citizens, and the effort just goes into persuading them to sign up for that sweet sweet signing bonus. I am somewhat curious about the new siege system - does that imply that you're going to have to spend a few turns waiting outside any city you conquer, or is it just a way to get some free wall breaches or some equivalent once you do start the battle?
|
|
#
?
Mar 9, 2023 18:42
|
|
|
Fitzy Fitz posted:Age of Wonders: Napoleonic Wars, please A fantasy 4x where technology developed over time Ala civilization would be sick. You could even do the dominion thing where magic gets less prevalent /powerful as time goes on.
|
|
#
?
Mar 9, 2023 18:47
|
|
|
|
| # ? Jun 13, 2024 05:51 |
|
|
the fact there's not an Eberron 4x by this point is a crime.
|
|
#
?
Mar 9, 2023 18:59
|
|