Main Download Information
- Thread starter 45°38'N-13°47'E
- Start date
Afforess
The White Wizard
Yep, I made the change in Patch G, after receiving comments about it. With Great Commanders, great generals can no longer attach to units, and can only lead them in the battlefield.
Wow. I didn't even manage to download Patch I
Will you be moving more of the game options to the RoM BUG settings, Afforess? Less to click each game is always better for me!
I try to. Some settings shouldn't be alterable once a game has started, so they won't be moved.
In somewhat related news, I've improved MP support for the RoM Bug Settings in 1.74 development branch, so that 9 more options are selectable for all players instead of just the host. The remaining options will be only selectable by the host, as they reflect game-wide selections that should not be tampered by other players.
Hashishim
Prince
i found something maybe its part of the new dawn game dynamics etc. It first time i playing it properly, using my civ pack aswell, with realistic diplomacy, fixed borders, and realistic culture spread, aswell as ruthless and aggressive ai.
What i noticed was that it takes very very long for the ai borders to expand :/ and difficulty level is deity. Also they dont give much of a challenge compared to before. They take ages to expand, and form other cities,
Any idea why? is it possible to make the AI more faster in terms of expansion? and im playing my game speed is marathon. (on snail it was worse)
In the game due to this i became the strongest, so they also not attacking me is it possible to make them more bolder.
On a further note: Game in terms of ctd's and end turn times, its actually quite fast
So thanks for that.
What i noticed was that it takes very very long for the ai borders to expand :/ and difficulty level is deity. Also they dont give much of a challenge compared to before. They take ages to expand, and form other cities,
Any idea why? is it possible to make the AI more faster in terms of expansion? and im playing my game speed is marathon. (on snail it was worse)
In the game due to this i became the strongest, so they also not attacking me is it possible to make them more bolder.
On a further note: Game in terms of ctd's and end turn times, its actually quite fast
So thanks for that.
- Joined
- Mar 31, 2008
- Messages
- 14,926
i found something maybe its part of the new dawn game dynamics etc. It first time i playing it properly, using my civ pack aswell, with realistic diplomacy, fixed borders, and realistic culture spread, aswell as ruthless and aggressive ai.
What i noticed was that it takes very very long for the ai borders to expand :/ and difficulty level is deity. Also they dont give much of a challenge compared to before. They take ages to expand, and form other cities,
Any idea why? is it possible to make the AI more faster in terms of expansion? and im playing my game speed is marathon. (on snail it was worse)
In the game due to this i became the strongest, so they also not attacking me is it possible to make them more bolder.
On a further note: Game in terms of ctd's and end turn times, its actually quite fast
I never had that problem. The AI seems to like expanding faster than me most times.
It's probably because of Realistic Culture Spread. It takes more to spread over hills and through forests.
Killtech
Discutator
i found something maybe its part of the new dawn game dynamics etc. It first time i playing it properly, using my civ pack aswell, with realistic diplomacy, fixed borders, and realistic culture spread, aswell as ruthless and aggressive ai.
What i noticed was that it takes very very long for the ai borders to expand :/ and difficulty level is deity. Also they dont give much of a challenge compared to before. They take ages to expand, and form other cities,
Any idea why? is it possible to make the AI more faster in terms of expansion? and im playing my game speed is marathon. (on snail it was worse)
In the game due to this i became the strongest, so they also not attacking me is it possible to make them more bolder.
On a further note: Game in terms of ctd's and end turn times, its actually quite fast
i suppose you don't use revolutions. the problem there is that if the AI expanded faster it killed itself because of revolutions and rebel civs breaking apart. anyway as it is now the AI considers expansion too early (it builds settlers sometimes in a size one capital... which is quite a suicide) instead of developing its cities to a decent level first and expanding afterwards with maybe some more dedication.
Hashishim
Prince
I am using revolutions on
Killtech
Discutator
well, then it's strange. all earlier games i watched on autoplay and from observations on AI civs in normal games the AI expanded much more but lost most of the cities again (creating new rebel civs). so even when you started with 15 civs the map was full with 40 after a short while. it's not that bad now.
but i still see AI picking some very strange building choices at the start - which btw increase rebelliousness. i'll see if i can fix it with my mod by changing the flavour settings of different buildings. so the big question to Afforess is:
[TAB]what effects does Flavour and AIWeight tags have on the AI building choice with your current function relative to the building stats? is there something else i could do through xml to influence AI decisions anyway?
but i still see AI picking some very strange building choices at the start - which btw increase rebelliousness. i'll see if i can fix it with my mod by changing the flavour settings of different buildings. so the big question to Afforess is:
[TAB]what effects does Flavour and AIWeight tags have on the AI building choice with your current function relative to the building stats? is there something else i could do through xml to influence AI decisions anyway?
Afforess
The White Wizard
but i still see AI picking some very strange building choices at the start - which btw increase rebelliousness. i'll see if i can fix it with my mod by changing the flavour settings of different buildings. so the big question to Afforess is:
You can change the value all you want, but buildings are categorized into 12 or so types, so if the AI isn't interested in a happy building, even with a weight of MAX_INT, it will ignore it.
Generally speaking, the AI is pretty decent at knowing what it needs, but sometimes chokes when it tries to find a building that matches its needs. That can only be fixed in the SDK.
Killtech
Discutator
the question is a different one. if the AI is interested in a science building, is it sufficient to make the science flavour of building A higher then of building B to let the AI pick building A? and what happens if i leave out all flavours, will that stop the AI to build this building before other more important ones?
or to be more concrete: what can i do to keep the AI from building bandit hideouts in its early cities?
and what does AIweight tag do? can it be a negative value to convince AI to not build a building until a specific situation where it requires a building category so much that the buildings flavour in this category is sufficiently high so it will chose it.
this is somewhat relevant to me as i added the concept of negative effects on some buildings that give huge bonuses (like maintenance for factories). thus it is not wise to build a factory in a remote size 1 city any more. it is also relevant for building upgrades since you could build a forge first and upgrade it to the foundry then (with a 15% production boost from the forge) or you could build the foundry directly.
i would like to experiment on the values a bit but i need to know how they work before i do so i have an idea how to use them and how to interpret the results.
--
and lol, why is my [ TAB] quoted with smilies??
Afforess
The White Wizard
Probably, but you will need to test it.
Okay, if you look at CvCityAI::AI_buildingValueThreshold(...), the AI has free promotions values with buildings that provide experience to units. So the AI is trying to build a building like a barracks, but sees only the bandits hideout, and settles for that. Since it is not worried about happiness, the minus value from the revolution sentiment (which is categorized in happiness) is low.
The real kicker here is that buildings were never meant to give negative effects. If you look at BTS, very few buildings have negative effects. There is no cap in the AI code, so it does value the negative effect, but it doesn't scale well... A city may be interested in fixing its experience problem, but won't know to not break it's happiness or revolution issues along the way.
(Sidenote: for quick reference, here are all the flags for how cities breakdown buildings into catagories)
Spoiler :
For the AI to properly evaluate buildings with negative modifiers, you would have to rewrite a large portion of AI_buildingValueThreshold() so that the AI knows when they shouldn't sacrifice a happy face or two and when it's alright, or when 15% more revolutions is bad (border, newer cities) and when its fine (capital, larger more established ones).
That way, the AI would realize the bandits hut was not the building it was looking for, and try to look for a better one.
CFC sucks at formatting.
Killtech
Discutator
i will do.
ah, it's the free promotion it wants. hmm, so the best solution would be to teach AI to have only some developed cities that concentrate on military production while the others do never build any military (neither buildings nor units)! thus the bandit hideout should only be build in the capital which usually is one centre for military production through the whole game and in few other developed cities that are designated for that purpose.
...the less elegant solution is just the remove the promotion from the hideout i guess.
yes, i see. but i do plan my reworked buildings in such a way that it is still optimal to build all buildings but some should be delayed until a city reached a certain development level so the bonus is worth the trade-off. so the general strategy remains mostly the same: build something and care about its negative effects when they become a problem. however a factory should not a too high priority in a small town as it really costs more then it helps.
the idea behind this is that earlier i build factories in remote size 1 cities and hurried them with caravans + gold as it made perfect sense. the AI never understood this. but now this isn't worth it any more which should help the AI. however i need to make sure it prefers earlier production buildings for small towns and not go directly for heavy industry facilities.
ah thx. are they ordered according to priority?
lol, what? a tab is censored? i guess it's a bug isn't it?
If it thinks it is bad, it changes it to
.
Afforess
The White Wizard
The AI doesn't know how to specialize cities. Changing that would be a long and hard task, to be sure. What can be improved is some of the weighting though. The promotion values are just a static value, so the building could be giving an uber-promotion, or a crappy one, and the AI doesn't know the difference. I can probably fix that at least, which should fix the over-valuing of the bandits hut.
If you see more very questionable behavior, do mention it. The AI is *trying* to do the right thing, but often fails in the execution. The real reason City AI improved between 1.64 & 1.73 is because I fixed a problems with valuing buildings. The AI would look for a food building, but not see what it was looking for. More test cases will help me make the AI less questionable.
Edit:
I rewrote promotion logic, which I will need to test out some, but hopefully will be added in the next patch. It's no longer a static +50 to value and varies based on your units and the promotion
Edit 2:
My tests indicate the AI is no longer rushing to build the bandits hideout anymore, and it's value is more in line with reality. Let's move future AI discussions to the new thread I opened for it.
Any chance of making this optional? I prefer the way it was prior to patch G. I like the idea behind Great Commanders but not enough to give up the ability to attach GG to units.
Also thank you for creating such an awesome mod, I can't even imagine how much work must have gone into this.
os79
Deity
Better solution is to make two buttons. Default for attaching and another for Great Commander to be turned on for that GG. If you don't turn on that GC button, you get NO benefits of GC, but if you do push that button, your chance for attaching GG is lost forever. That would be perfect solution.
Otherwise, I don't care, not like the poster above me.
Otherwise, I don't care, not like the poster above me
.
Afforess
The White Wizard
Okay, I can do that.
Blackflag421
Warlord
- Joined
- Nov 3, 2009
- Messages
- 198
This.
Afforess
The White Wizard
I've compromised. There is a new action button with great generals that enables their great commander abilities. Toggle it, and they become a great commander. But you can't switch back, and Great Commanders can't be settled as military instructors, hurry buildings, or join troops; you will need to decide beforehand what you want.
