AMAI 2.50c Strategy

This thread should be dedicated for ppl who like to modify the AMAI strategy and talk about it.

The default AMAI 2.50c strategies is really not in favour for Night Elve's race because of lack of wood and often their base got busted in an early attack, especially when the orc tower rushed them.
By default almost every tier 1 strategy in AMAI for NE is like this :

call BuildUnit(6, ARCHER, 45)
call BuildUnit(1, TREE_AGES, 50)
call BuildUnit(1, hero[1], 80)

that make the NE slow to start and also weak in early game. I'm afraid for AMAI simple edition user found that AMAI NE player is a very poor and no good as opponent as well as an ally.

I have tuned up the NE strategies to solve the wood problem and make them alot stronger race from the beginning of the game as they supposed to be.
I think all the NE strategies have the code for tier 1 like this :

call BuildUnit(15, WISP, 50)
call BuildUnit(12, ARCHER, 50)
call BuildUnit(6, HUNTRESS, 50)
call BuildUnit(4, BALLISTA, 50)
call BuildUnit(1, TREE_AGES, 40)
call BuildUnit(1, hero[1], 80)

code to make amai build 15 wisps is to make sure they have enough wood to go to tier 2. And then they train archers and huntresses after they summon a hero simultaneously. This tier 1 code would make amai NE player have a strong army to defend their base from early attack and even make a great counter attack.

for example I made one stategy, I call it the double druid strategy :
Tier 1 code (same as above)
Tier 2 code:
call BuildUnit(10, DRUID_TALON, 45)
call BuildUnit(5, DRUID_CLAW, 45)
call BuildUnit(6, DRYAD, 45)
call BuildUnit(1, TREE_ETERNITY, 80)
call BuildUnit(1, hero[1], 80)
call BuildUnit(1, hero[2], 60)
call BuildUpgr(1, UPG_DRUID_CLAW, 55)
call BuildUpgr(1, UPG_ABOLISH, 55)
call BuildUpgr(1, UPG_DRUID_TALON, 65)
call BuildUpgr(1, UPG_STR_WILD, 65)
call BuildUpgr(1, UPG_HIDES, 65)
Tier 3 code :
call BuildUpgr(1, UPG_MARK_CLAW,80)
call BuildUnit(6, DRYAD, 45)
call BuildUnit(20, DRUID_CLAW, 75)
call BuildUnit(20, DRUID_TALON, 65)
call BuildUnit(1, hero[1], 80)
call BuildUnit(1, hero[2], 60)
call BuildUpgr(2, UPG_DRUID_CLAW, 80)
call BuildUpgr(2, UPG_DRUID_TALON, 65)
call BuildUpgr(2, UPG_STR_WILD, 65)
call BuildUpgr(2, UPG_HIDES, 65)

I tested my strat by let amai players play against each other (computer only) with 3 NE vs 3 orc. With standard AMAI strategy it's almost impossible for the NE to win but with my NE strategy NE amai player could beat the orc quite easily.
Many archers in tier 1 make the orc have to think twice before they use mass wyvern strategy and they tend to use mass grunts and catapults which is quite easy for NE bears to eat them. And then the orc tried to mass taurens and the NE counter them with the talon to blow em sky high and lower their armor and let the bears kill em one by one.
heh.. I love to watch the orcis horde dies like flies..

Well, for my A.I I just let the algorithm do the building... but I do script everything from start to tier2.

I found, in many cases, that AMAI is inefficient in its start. See my A.I for inspiration :p (they've been tuned to be as efficient as possible down to seconds)

The best thing of amai is it's engine and it fast and quite easy to customize. In a game actually doesn't really need many strategies maybe just one or 2 or 3 good strategy will cover to counter the enemy threat. But still is fun to see the massively same units fighting in a game which is not possible in standard AI.

I'll take alook of your AI but where to find it..lol j.k

http://www.wc3campaigns.net/showthre...919#post901919

Customizability is why i fell in love with AMAI as its much easier to implement into a custom mod or personalise for yourself. We used to have the amai strategy database at wc3sear.ch but it died and no one actually made AMAI strategies apart from one i did :) And amai2.50 just extended its customizability further and enabled the ai vs amai which was is just an amazing feat if you had ever looked through the 2.43 code... YUCK!!!

One thing about the current initial strategies for all races, is that they were made to get AMAI upgrading to the next tier as fast as possible and still train units while upgrading as strategies in 2.43 were too slow. Obviously you seem to find them tiering too fast now :) Cheers for the strategy ideas as it is the hardest thing to balance with many people having different ideas of a good effective strategy for warcraft (hence reasoning behind having multiple strategies in first place)

Hence why Warcraft III is I think one of the best (if not THE best) RTSes out there. There's so many possibilities that can lead to a win.

With that initial strategy in fact it doesn't slow the NE from upgrading to the next tier (yup this version upgrading tier pretty fast and I love it ) instead of having a strong army and enough resource to comply the strategy use in the 2nd and 3 rd tier.

I haven't found good initial strat for other race to match that NE initial strat If anyone have a one I wouldn't hesitate to challenge it hehehe....

Btw for Strategy Master in strategy editor extract and insert button doesn't seem to work please check that out... thx

Actually that's one thing why I don't like AMAI. You can't really make good, efficient builds when you have to consider compatibility and modularity.

In short, I want total control (if that's possible.. dam Blizzard and their native functions).

@shadowstrike:

Open AMAIStrategyManager.pl with notepad and modify line

if($] == 5.008004 or $] == 5.008006) {

to

if($] == 5.008004 or $] == 5.008006 or $] == 5.008007) {

So you add for whatever version of perl currently is. 5.008007 = 5.8.7
I think latest version is 5.8.8 i can't remember, but that should fix the buttons not working temporarily.

@SM
Ok thanks it works..

NE still has expansion problem especially in playing normal mode, Does AMAI have a routine check whether amai player has an expansion or not ? It shouldn't be triggered by timer only.

If we want AMAI playing in optimal performance in normal mode game, amai should always check whether they have expo or not and keep looking for one because when their resource kinda running low they tend to idleing in middle of the game which is suck! cus I want amai to be more aggresive with action per minute way over human player could have done.

Right now I think amai still optimized in insane mode, with little bit change strategy and settings they play a extremely fast and vicious full action game.

@Shadowstrike
It is only timer based with a chance for them to ignore the timer and do a fast expansion. If they keep taking expansions it will have a negative effect on AMAI due to the way the build code works. I havn't messed around too much with the build code as its the most complex part of AMAI and messing around too much causes weird things to happen.

Nope now amai does make a very good and efficient builds, it's only depends on strategy and build sequence script which can be easily to modify. Now with new function with a little bit control of placing buildings I could make NE no longer stuck in their base

btw I tried your AI, It seems not quite different then the standard ai which is creating small amount of every type of unit a race has and the buildings which is suck cos it's not efficient, there's too many unused buildings and poor NE with that crappy builds they always stuck in their own base .

@shadowstrike: Lol what u got against nightelfs getting stuck in their own base. Its part of the decoration :P

Would you be able to upload your work with nightelf ai here?

Replay? Actually there's some "learning" variable but it works so slowly sometimes. As for now it's reactive instead of proactive. And of course they'll build everything if you also build everything in 1v1, that is. Team games and FFA games are a totally different beast and I haven't done them yet.

No I'm not saying that the build sequence isn't easy to modify. Some things can potentially break the core AMAI system.

I'm trying to port my blademaster harassment to AMAI btw, but I haven't found solutions to some problems :
1. It should start right after the first green creeping is done, not tied in any way to time.
2. It requires the Voodoo Lounge to be built after first wave (don't think this is a major problem since it only needs lounge after first wave of harassing, which is plenty of time).
3. The grunts should keep creeping on their own without relying on the BM.
4. It requires that the Blademaster is not counted anyway towards the owning player's "Military strength" or something. This is to prevent the grunts in #3 suiciding to creeps that they can't handle by themselves.
5. I don't know which functions are already available in AMAI core and which isn't. Variables, also. A list would be nice, along with short description of what they do.

@ Strategy master

yes sure I can do that :
here with my vicious killer NE's strategy.. hehehe

@zycat
sry I didn't save the replay, will do next time I play it.