Posted on 07/25/11 @ 06:38 AM (updated 09/17/11
Still fairly weak, but now it flushes!
Tested Game Settings:
But it was made for any game settings.
It IS Non-cheating, but it contains the older version's cheating file in case you want a real challenge and know how to load it.
Dont worry, it doesn't cheat on moderate, if you look at the file you'll know what I mean.
1.2- Also shortly after I uploaded 'UnfairestEel AI'
I discovered that it was bugged on Real-world maps. That is now fixed and I suspect it was also bugged for Map sizes greater than Tiny, That's also fixed.
1.5- dramatically stronger than 1.2. also now comes with a cheat detector (only on Hardest difficulty) so it cheats when it detects certain cheats or when you message it the number '42' in the hardest difficulty it will cheat! for those people who like a challenge.
1.6- improved from 1.5, inlcudes a unit ID list and group ID list from Berry66 at http://forums.aiscripters.com/
1.7- Flushing test! It's like a whole new script altogether!
1.8- Fixed some obvious problems with 1.7.
1.9- Easily beats previous versions thanks to small changes with the help of advice from campeador
2.1- Smarter flushing selection of units.
2.2- Compatible with UserPatch 1.1
I hope that I can improve it in the future, feedback would be appreciated!
In the last 2 tests I did, it's scout failed to find the enemy until they were advancing into castle so it couldn't successfully flush. I beleive that this was just bad luck but if you encounter similar problems with UnfairestEel: Please let me know!
|Pages:  2 » Last » |
|Author||Comments & Reviews ( All | Comments Only | Reviews Only )|
First of all your ai cheats. When people download an Ai they would like to know its specifications (what civ, does it cheat, which maps, death match or random map etc.).
I recommend you post this information in the description. For instance, your ai plays all civs, but it prefers aztec or byzantine. Your Ai adds 100 resources as if playing the standard Ai on hardest(i think it actually adds more).
Despite your ai cheating, its still(as youve mentioned in your description) relatively weak.
I versed it(aztec) against my Ai Brititron and Brititron did away with your ai very easily.
Few tips: Build more lumbercamps. You had only one the entire game if im not mistaken (i stopped checking around 45 min)
Economical management needs heavy finetuning, its not terrible, it just has alot of room for improvement.
late Castle and early Imperial age researches. When my ai attacked around 40-45 min, it only had longswordsman and jaguar warriors. It also researched plenty of monastery upgrades but created only a few monks. Maybe save that gold more military upgrades? Hopefully you see what i mean. Get the more important upgrades that are gonna be used more.
Every attempt at scripting should be acknowledged, so good work, keep it up.
Thanks for the advice cakemaphoneige.
I will look into lumber camps especially.
About the cheating...
From Easiest to hardest(in order):
On Easiest it doesn't cheat, in fact it gets majorely handicapped with certain SNs and the difficulty parameters.
On Standard it still gets handicapped a fair amount but not a great deal.
Moderate was it's testing difficulty, it doesn't cheat and it's the AIs strongest non-cheating difficulty.
I must admit I did go overboard on the cheating on Hard, should've probably been the cheating on hardest only. Anyway, it get's an amount at the start of the game and then every time it advances. Once it reaches imperial age it gets 500 of every resource every 20 minutes.
It doesn't really have any.
It was made to play byzantine and then adapted to all civs but it's actually strongest as Aztecs(so far). Can play Nomad and water maps. Untested on lower Population-limits but it is made to play on any game settings.
But it's especially weak on Deathmatch. This is because it doesn't really use all these starting resources (it was originally made to not use resources early, I tried to change this but it's still not that great) so in deathmatch it can get blown out of the match early.
By the way, does any one know what
sn-target-evaluation-distance means? I thought that giving it a negative value might make it attack near opponents but I'm not sure.
PS. Please let me know if you would rather it if in my next update the AI doesn't cheat but doesn't scale for difficulty either.
I just thought that it was a good idea.
[Edited on 07/28/11 @ 05:28 AM]
On standard everything takes longer to build, research or train. It can't attack villagers. No big deal? No, it pretty much cripples the AI.
Moderate difficulty handicap is not a big deal; on hard it has no advantages and no extra resources. On hardest it gets 500 each res at the start of each age. This why AIwars are hold at hard difficulty.
Target evaluation number affect targeting of units. To attack your next neigbor you need to use Town Size Attack: [URL=http://aok.heavengames.com/university/other/town-size-attack/]Link[/URL]
I guess its up 2 you if you want it to cheat or not.
If your just scripting for fun, do whatever you like.
If your scripting to make an Ai to compete with all the best Ai's designed for Random maps who dont cheat, and then possibly enter the AOKH tournament ladder, i recommend you disable the cheats and get to work!.
Make 2 versions if you like. One that cheats to a point that it challenges a human player.
Another non-cheating version to compete with other Ai's that dont cheat.
Oh and i also recommend you script it to work specifically on hard difficulty setting. 90% of people who download it will test it on this setting, plus it allows the Ai to use its full potential.
If you havent been to it yet, check this site out:
Thanks for your comments, guys. I'm going to look into those links now!
I may not get to work on my AI for a while now. :(
By the way, I made update 1.1 but it kept losing to the original so I'm going to have another look at it and move into update 1.2.
UnfairestEel v1.2 has been released!!! UnfairestEel v1.2 is Non-cheating and better than ever! If you want proof, download it and test it against the old one (if you downloaded the old version)
Comments on how to improve it would be appreciated! It's thoses comment that got it to be as good as it is.
I have started on Version 1.5 of "UnfairestEel"!!!
I have this idea, I think that it will ask you on Hardest difficulty if you want it to cheat. Also, on hardest difficulty I think I will make it so that it detects certain cheats and Cheating AIs and if it detects cheats or if you message '42' within the first 90 seconds of the game it will begin cheating itself!
Let me know what you think about this idea...
Unfortunatly, I can't seem to think of ideas to make "UnfairestEel better ):
[Edited on 08/13/11 @ 01:07 AM]
You still have much work to be done :P
I versed it against wartron to see how its coming along and to test wartron against it for there upcoming game in the ladder.
In this game unfairestEel was random => mayans.
Your feudal time is good, pretty quick, but very risky. However, your castle time was very slow. Heres a few hints/tips for you:
1.It built a mine in dark age, at around 6 minutes, big no no as that wood could be used for farming and is just a plain waste IMO.
- As soon as the fuedal button is pressed, build your mine then.
2. This is more of a hint, i figured it out myself when doing the same thing. When building a stone mine in feudal, put minumum-dropsite-distance stone < 4 (again, experiment to what you prefer) so that in the odd case when your gold mine is built perfectly inbetween gold and stone, you dont need to waste wood on a second mine because theres allready a mine next to both.
3. You only had 26 vills when you pressed the feudal button, i recomend at least 28 but the best would be 30(again, experiment).
4. Town centers. You didnt build any in castle age!. As soon as you hit castle, drop a second one straight away!!!, then you want either 1 or 2 more straight after that. Start with trying to get 2 more town centers built within 5 minutes of advancing to castle then if you want add a 4th after that. You want more town centers as soon as possible. Will make booming much faster and more efficient.
5. Towers. Keep using them for now if you really want too, but eventually u may come to realise that they are a bit of a waste, so just keep your eye on that strategy, you might be able to do better things with that stone and the vilagers mining it.
6. Your whole strategy doesnt have much flow to it. Its not bad, in fact its pretty good for someone who is new to scripting. But keep your eye on it and compare it to other ai's strat and try and figure out more efficient ways to do things. I dont wanna say which is the best way because there isnt one and in fact, i encourage you to try different things to everyone else because i love uniqueness. So just remember that.
I have only tested 1 civ with your new version and i must say your making improvement, so good stuff.
Ill test some others when i can and try and give you some more ideas if you like.
Also in the game i tested, both ai's were having a good fight and unfairesteel was keeping up with wartron quite nicely :P.
But wartron hit a bug and didnt research Two handed swordsman or champion. So i ended the game to fix wartron before much else happened.
My ai wartron is a gothic infantry training ai so i suggest you get some rules in to defend against such an ai. I havent released wartron yet but i will in the next few days, so do some testing before our match in the ai ladder.
I versed them against each other again. This time unfairesteel was chinese. I watched from wartrons perspective this time. Unfairesteel put up a good fight, but when wartron attacked fully upgraded and a age ahead of unfairesteel, it wasnt long before he overtook unfairesteel and destroyed his city.
I must commend you though on how well the chinese looked. They had a nice array of different units and although wartron won, unfairesteel did pretty well. If it had of been in the same age as wartron it may have put up more a fight. But my ai wartron is really good and can beat alot of ai's out there, so dont feel bad about it losing once.
Like i said above, get wartron when i release him and start doing some tests. Then when our game comes around maybe it will be really close ;).
If you cant wait till i release him, go and find some gothic ai's over at aiscripters and test against them, or even farmertron, my ai's are built around its script.
[Edited on 08/27/11 @ 03:09 AM]
It's a good AI and definately has potential. I've tested it against a few other AIs and it, whilst putting up a fight, has loads of room for improvement... A few things I'd look into if I were you are:
In the games I played, your AI made a barracks, market, blacksmith, archery range and 2 stables before it started advancing. While this can be quite beneficial for dealing with flushes, you should not need the market until castle age unless you have any allies. When you don't train troops until you're advancing too, I wouldn't suggest this. Try with barracks + archery range + stable/blacksmith (blacksmiths cost less wood).
As cake mentioned, your mining camps need sorting out if you haven't done so already, and you may want to work out a way to see if you can get your first lumber camp close to your base every time...
The AI tends to get to the imperial age very slowly. Try messing about with escrows and some other things. Remember, you don't need to research everything in the castle age with your escrows. I'm not sure which things you do use escrows for or if you even use them at all but I'd advise you to look into this :)
Thanks a lot guys!
I only with that I had read these comments before starting 1.7! I will take on what you said about mining camps, towers and villagers in the next update...
In the meantime... 1.7 flushes so give me feedback on that strat.
Also, in the meantime, I'm starting a new script. Just for arabia.
Nice flush, UnfairestEel! You still have a lot of space to improvement, but I agree Alevo, this has potential.
I’ve tested your Huns against Sassanide Empire. Your Dark Age is very bad, with a 14:55 and an unacceptable idle TC. Reasons:
- Farms, once you have built a mill and a lumberjack you have to start to build farms the faster the better until you reach a minimal number (which depends about your strat, think in the food gatherers that you’ll want in feudal transition and the farms that you can afford), and also keeping a minimal number of idle farms (which will be greater in later ages, except in moments that you need wood for critical buildings).
- Other buildings. A second lumber camp is in general a good idea, except for FC strats in which you mine gold instead of selling stone in the market, when you can build the gold camp instead. But a 2nd lumber camp, a gold mine and a stone mine are excessive, except if you have a lot of farms and can do it with a good feudal time and without idle TC (I’m not able to do it in RM and Low Resources). And the barracks can wait until feudal transition except if you are drushing (I don’t recommend it for AIs), because it delays your feudal time too much, and anyway, you are not going to train militias them until feudal transition or even feudal age.
- Militias. Only train them in feudal transition if you are sure that you can afford man-at-arms upgrade once in feudal. They are useless without that upgrade. Anyway, it’s possible that it was your idea in my test but you didn’t do it because you changed to spears once detected the enemy stable.
Your later ages are much better, anyway here you have extra tips:
a) Your economy is too unbalanced in the feudal age. Too much gold and too little food and wood. I imagine that the reason is that you designed your gatherer percentages for m@a and archers, but you trained spears/skirms to counter Sassanide army. Possible solutions:
- Market, you have built it anyway. The only two reasons to do it are trade with an ally and sell/buy resources.
- Dynamic gatherer percentages. You can see that I define in my Constants.per file four constants for four unusual SNs which are added to your gatherer percentages. I use them to make my eco more balanceable in late feudal and later ages. Lade uses a different system, which must be better but is too complicated and requires even many more rules, so I don’t recommend you it.
b) If you are attacking an enemy and have built a siege workshop, train rams, they can help you to take down buildings much faster.
Anyway, good job. Defeat Sassanide Empire is not a bad thing.
Nice flush, it causes a fair bit of damage to Wartron.
All you need to do now is capatilise on having an advantage when your flush is successful.
Have different goals for different strategies to make your ai more diverse. When the flush goal is activated set a whole different build plan for this goal so you can get the best eco running for that particular strategy.
In the game i just did it never reached castle age, so thats definitely a good thing to start working on first. Try and get a castle time of less than 30 minutes, while keeping up the unit training/researching etc.
Do some research on the horde/is machine and observe there flushes and how they operate after they begin there flushing strategies.
Anyways, your ai is improving all the time so nice work.
|Pages:  2 » Last » |