Putting the 'role' back in role-playing games since 2002.
Donate to Codex
Good Old Games
  • Welcome to rpgcodex.net, a site dedicated to discussing computer based role-playing games in a free and open fashion. We're less strict than other forums, but please refer to the rules.

    "This message is awaiting moderator approval": All new users must pass through our moderation queue before they will be able to post normally. Until your account has "passed" your posts will only be visible to yourself (and moderators) until they are approved. Give us a week to get around to approving / deleting / ignoring your mundane opinion on crap before hassling us about it. Once you have passed the moderation period (think of it as a test), you will be able to post normally, just like all the other retards.

SMAC 444 - AI patch updated [p]

kyrub

Augur
Joined
Aug 13, 2009
Messages
347
New version: SMAC_444_k

Download: http://www.mediafire.com/download.php?6redidk4z42h1mo


Features (mostly bug corrections)

k1 stockpile energy bug
k2 swap base bug
k3 children creche bug
k4 AI hurrying price bug
k5 AI can build seaformers earlier
k6 more colony pods for University and Morgan
k7 reversal of distant AI cities (AI can build cities at closest distance again)
 

MetalCraze

Arcane
Joined
Jul 3, 2007
Messages
21,104
Location
Urkanistan
Is that you trying to make University and Morgan more competitive like giving them a better start?

Also now that we are at it - if you could've given Deirdre a little bit more focus on worms it would've been nice (as it should be per lore)
 

kyrub

Augur
Joined
Aug 13, 2009
Messages
347
MetalCraze said:
Is that you trying to make University and Morgan more competitive like giving them a better start?
I thought that was one of the point of the AI patch. I am trying to improve the factions without betraying their original AI. It is basically a manifold NUDGE strategy used on AIs. You try a lot of small pushes to make the AI better in certain circumstances. I used something similar on MoM. The advantage: the game feels the same as it did before, the disadvantage: you may witness a definite improvement only after long time.

Example:
In this case, Morgan and Univ AIs were programmed to slow their development after 4 + time_constant cities. This along with two other conditions (biggest producing base and biggest size base don't build colony pods) lead very often to a complete development stall, due to an imprecise check for the above mentioned conditions (any city with same POP as biggest size...). Especially Morgan, who can quickly have all cities maxed at 4 pop, suffered massively.

So, I increased the slow threshold to 5 + time_constant, I put the max_producing_check earlier to allow an early SP building, but I postponed the biggest_size_check until the player has at least 6 bases. Now, the stall seems to be completely avoided and University and Morganites create sustainable empires... It seems much better so far!
 

kyrub

Augur
Joined
Aug 13, 2009
Messages
347
Sorry. No news coming in next days.
(The development is stalled by me having a crazy bundled agenda in RL... and by me playing JA2 1.13 despite my bundled agenda, hehe. ;))


Howerer, iit seems we did hit some important spots already in 444_k. I have a report or two about visible game improvement. Just one bad side effect, now: the AI bulds too many probe teams.
It would be nice to have further AI improvement suggestions from people who know the game really well.
 

DakaSha

Arcane
Joined
Dec 4, 2010
Messages
4,792
Fuck you all for making me read this entire thread and get stoked about my AC installation
However: Is it really *that* bad with the probes? as in so bad that its gonna piss me off?

edit: and thanks for the work kyrub
 

DakaSha

Arcane
Joined
Dec 4, 2010
Messages
4,792
I let the ai play till 2350 or so and took a peak and it did look quite amazing.. I didnt see to many probes to be honest but i wasnt looking that hard.
What was cool was: All factions were pretty much balanced. even morgan (who DID have a whole island to himself) was doing great. he may have even been leader.
Guess who was 'losing'.. The hive haha. They only had sea bases which was pretty cool. Made me wonder about the 'history' of that. It looked like the spartans pushed them into the sea.
It could well have been a fluke but guess ill have to give it a try.
Kyrub i saw in another forum that you werent super interested in updating this because it doesnt get *that* much attention. I think alot of people are just waiting for you to take care of some AI flaws like movement and combat 'strategy'

either way thanks for the effort
 

desocupado

Magister
Joined
Nov 17, 2008
Messages
1,802
DakaSha said:
Kyrub i saw in another forum that you werent super interested in updating this because it doesnt get *that* much attention. I think alot of people are just waiting for you to take care of some AI flaws like movement and combat 'strategy'

either way thanks for the effort

We are thankless fuckers, but right now, I have Alpha Centauri installed in my laptop with your mod on it. Your job is very much appreciated.
 

DakaSha

Arcane
Joined
Dec 4, 2010
Messages
4,792
Ran another test with good results. At turn 200 the hive is third and the university is first. Over the course of 200 turns balance in power shifted (instead of it just being hive the whole game)

Guess ill actually try a game :P
 

kyrub

Augur
Joined
Aug 13, 2009
Messages
347
Thanks for encouragement and even more for the feedback. No worry, the patch will continue to slowly move on.

I am not bitter at the reaction to the patch. It's anybody's choice and it's normally so with everything hackers do. It pushes me into finding some time for the work when many people try to play with one of the patches and ask for more. Which did not happen so much with SMAC, so far, maybe later, why not. Every game has a life span and SMAC is by no means gone. Plus the game's AI is utterly complex - my guess is this is an adapted copy from Civ2 AI. Which means that I was trying to break through a 3rd generation AI. :P

Saying that, Dakasha, I really should have repaired that probe problem, that's easy. Next week, I guess.
 

Malakal

Arcane
Glory to Ukraine
Joined
Nov 14, 2009
Messages
10,266
Location
Poland
Kyrub how about simply adding more weight to good buildings and good units for AI. From my, albeit limited, experience with AI modding sometimes this helps greatly without moving into deeper areas. All civ games, AC too, suffer from building inflation where dozens of buildings are not cost effective and AI is wasting time and resources on them.

Less useless building, more good units and formers should help very much. Because I suppose that combat tactics will be rather hard to change and improve...
 

DakaSha

Arcane
Joined
Dec 4, 2010
Messages
4,792
he often seems to find bugs in the AI coding that, when squashed, greatly improve the strategic capabilities. Not that what you are saying isnt a good idea

Goes to show that time and money are the worst thing in the world :P
 

DakaSha

Arcane
Joined
Dec 4, 2010
Messages
4,792
Did another test with AI till turn 250:

Hive is defeated. I think Spartans are leader. Looks good overall.
One thing that could be improved on with the AI is its base placement.
Right next to the spartan continent there is an island containing the borehole cluster for instance which is the only place they didnt settle.
Ive seen this with other specials as well
 

kyrub

Augur
Joined
Aug 13, 2009
Messages
347
Malakal said:
how about simply adding more weight to good buildings and good units for AI.
SMAC AI doesn't work with simple weighs + RND (like MoO or MoM). It uses opportunity cases, mecanisms that are switched on and off, depending on circumstances, on faction etc. I am fascinated by the attempt, although it is far from perfect.

Sure, you can use rough force to increase the odds of X building. I promoted the most obvious, Recycling tanks. Later on, I saw a faction giving up a defense of an empty city (size 3) against worms. Instead of defenders, the city was building - you know what. I stiffled the internal mechanisms, inadvertently, had to reverse them a bit,

Which buildings need more AI attention, in your opinion?

Right next to the spartan continent there is an island containing the borehole cluster for instance which is the only place they didnt settle. Ive seen this with other specials as well
I can positively assure you the SMAC AI loves landmarks. They are mad about Monsoon Jungle (and rightly so), they pursue Garland Crater. I've seen Morgan sending his second Colony Pod 12 tiles away to settle it.

Borehole cluster, however, was added in some later patch, am I right? Probably that. The other possibility is the ship settling problem, all Civ AIs seem to share...
 

DakaSha

Arcane
Joined
Dec 4, 2010
Messages
4,792
I trust you on that. Maybe borehole cluster doesnt count because it is nothing but a terraformed tile
 
Joined
Dec 30, 2010
Messages
30
In this case, Morgan and Univ AIs were programmed to slow their development after 4 + time_constant cities. This along with two other conditions (biggest producing base and biggest size base don't build colony pods) lead very often to a complete development stall, due to an imprecise check for the above mentioned conditions (any city with same POP as biggest size...). Especially Morgan, who can quickly have all cities maxed at 4 pop, suffered massively.

Shit, in my latest unpatched game I witnessed Morgans with five cities, 150-200 turns into the game, with huge swaths of unclaimed land nearby,which was the exact reason why I went searching for fan-made patches such as this one.

Thank you Kyrub for all your hard work, I'll be sure to give it a spin.
 

kyrub

Augur
Joined
Aug 13, 2009
Messages
347
I finally found some time and ported a few fixes (from my previous work on SMAC) to SMAX. The Smax Unofficial patch plus includes latest Scient's Unofficial patch for SMAX (see HERE and it adds several bug fixes on top (see release notes below).

Download: Smax unofficial patch PLUS

The highlight is the fixing of the infamous energy stockpile bug and of the children creche (and brood pit) combat bonus bug. I decided to include the better terraforming algorithm from smac patch as well. The old terraforming is not exactly a bug per se, but the AI behaves so stupidly, that I simply felt obliged to do the hard work of transmitting a whole part of the code in SMAX as well.
- Enjoy!


release notes

- the unit limit, changed by scient, is back (crashed the game)
- the automatic energy stockpiling bug is fixed
- the swapbase bug, which made the AI swap bases in a ridiculously bad deals, is gone
- AI basic terraforming (and automated basic terraforming) completely rebuilt, should be more competent
- the children creche, brood pit combat bonuses now work like they should
 

kyrub

Augur
Joined
Aug 13, 2009
Messages
347
MetalCraze said:
Were there any more improvements for SMAC?
No and I don't see an update in a reasonable time horziont. Sorry. The game AI has silly complexity and my tools are on the limit of their usefulness. I cannot even display the crucial AI routines in disassembler. And the compiler used for SMAC is a pain in the *** as well, giving enigmatic code. I could spent months on SMAC to get about 20% of the effect of my MoM patch.

As I've said elsewhere, we need a good AI programmer / hacker capable of inserting (dll injections) whole new AI routines on addresses I would give him. This would push SMAC AI efforts on a whole new level.


does it work fine on GOG version?
GOG has SMAC, so no. You better use the SMAC_444 version, that's a better patch (see my signature). I have now ported a few parts of it to SMAX, that's all.
 

Pelvis Knot

Cipher
Joined
Nov 19, 2010
Messages
885
I played a game a few days ago, took Peacekeepers, thinker difficulty. Found empty Believers capital. The Morganites did ok, but they fell to Hive which first steamrolled Gaia (very weak faction now imo, saw them getting raped in two other games), then Morgan, half of university and finally stopped at my borders. On the other border I had Spartans. After taking a few of their cities, bullshit started. They sent 5 probes each turn to one of the conquered cities, first destroying buildings and then pointlessly inciting drones to revolt. This continued for 70 turns, when I won the diplomatic victory.

tl;dr: if you ever continue working on this patch I hope you make probe building a lesser priority for ai
 

As an Amazon Associate, rpgcodex.net earns from qualifying purchases.
Back
Top Bottom