2/ Import AI Scripts to maps with SC2 Editor. 3/ Using SC2 AllInOnes tools. Visit Green Tea AI Main Page for more information.
Introduction:
- GT AI is a popular AI Script that improve the difficulty of Starcraft 2 AI. It will using many strategies that keep player guessing overtime. More over, it can control units like a professional player to keep every units alive. It also be used in many custom maps like YABOT, Macro or Die, Fyn AI, Coach Ai, .... - The things that GT AI can offer more than it contain is the ability to play the Build Order player created, so giving more range of possibilities.
Current Features:
- 60 AI Strategies: from the well know 4 Warp Gate to cheese build like Cannon Rush, 6 pool... - AI Micro Control, hit and run, storm evading, improve ability using for each unit: Force field, Graviton Beam, Neutral Parasite... - You can Create more Custom Strategies to AI, include 90 slot to save (10 slot for each match: TvT, TvZ, TvR, ...)
- Practice to perfect the BO created.
- AI Chatting for entertainment. - Reminder: AI can remind player some basic mistake player do : supply block, idle worker, idle production, floating resources, energy management on Orbital Command, Queen,
Option screen: - For ingame option: Type -op to show up option menu:
* Know issue: -qr causing some bugs that make AI not follow their build order. -some issue in building placement, some times AI can't find good place to build, causing AI to stop building. - limited harassment.
* Faqs: 1/ How can I play this AI ? - You can search it on Bnet or follow instruction on the main website.. 2/ Can I play corporate with friends ? - Yes, if the map is 4 player or above, you can set team match vs AI. 3/ How hard is this AI/ What is this AI level ? - Many player see it much stronger than the original AI due to macro and micro mechanic. As for player reported the AI being from Platinum to Diamond Player depend on AI Build Order choice. 4/ Is this AI cheat ? - Depend on AI level: the first 4 level doesn't using resource cheat. + Very Easy : It will stop attacking in 10 minutes. + Easy : No map vision. + Medium, Hard : AI has full map vision. + Very Hard: Only cheat resources in later game. Limit is 1.5 times normal resources. + Insane : 1.5 times resources at the start. 5/Can I use the AI Scripts for my map ? - Yes, just include some credit line for me ^^. 6/ The AI does some weird build/ old fashioned build order - You can always creating new Build Order/ Strategies, and let AI Play Random these new build order. 7/ The AI so laggy on my computer. - Due to scipt Analysis, AI cause some performance problem on low CPU computer. You can open the Option Menu, and turn APM limit on. 8/I got this error "Cannot find 'unitdata.xml" when loading the map. - The unitdata.xml is for micro AI part in game. If missing it, the AI still can played but will missing Micro features. Check out other maps. or follow the install instruction again, remember you must set "Public" and "Unlock" when publishing on Bnet. 9/What is "Streng Analysis" in the option menu: - It the Analyzing system developed by Nexxy to compare how strong your army is compare to enemy army. - For example : 1 VoidRay > 5 Roach 10/Why is it named "Green Tea" ? - It is a popular drinks in my country, and it good to your health. I also used a lot whenprograming this AI, so can say it is an inspiration for me ^.^. 11/When i choose Practice Build Order, I don't see any BO ? - You must created the Build Order first. - To create the BO Zerg Vs Terran. You must choose Terran and let the AI choose Zerg. - Than choose the "Play/Edit Build Order" button to create your build order. - You will see the Edit screen:
Finally been looking for a bot to train against cuz the Standard AI sucks so badly Nice work!! Will try it asap and tell one of my friends that really need to train some 1v1 :3 Keep up the good work
I used to practise vs GTAI before since it used to chase my scouting drone around and I needed the multitasking practise. I haven't tried the b.net version but I definitely will now.
The video is awesome, especially the last battle (zerg fighting!)
Whoa! Amazing AI! Works brilliant. My Toss Enemy was building xpantion like a Boss! lol.. Maaany Hallucination (FF would be cool :C). Really well done!
I remember playing with your AI when I first started sc2. It seems you have worked a lot on it, it looks awesome. Nice it can cheese (but still not defend cheese, which is understandable as it's a pain to program)
Thanks for the reply on how to fix the reserved .galaxy name problem, I just edited metalopolis with this AI and uploaded it to NA under 'GTAI 0.86c Metalopolis'. not sure how long it takes to show up as I haven't uploaded anything before.
This is awesome. I LOVED GreenTeaAI in BW. The video is quite impressive (the roach micro vs the zalots... O_O) Thank you verymutch, I'm gonna try that !
I'm lazy and won't download it and go through the hassle of uploading it myself (Also ignorant as to how to do that) but I'm a huge fan of green Tea AI and I can't wait for this to show up on the north american ladder
Macro or Die was good because of the green tea ai. If it was the standard ezpz do the same shit every game blizzard AI that UMS would barely have been notable, but Green Tea made it shine.
I can't wait to get my hands on some of these maps! Thanks!
Great AI! I hadn't updated in a while but i love the improvements in micro and tactics you made.
One question though: could you give an overview of what numbers correspond to what commanders? I tried to practice against a protoss 4WG and used the numbers from the trigger libs but for some reason it never uses a 4 WG strat when i did this. Same goes for 6 pool (although the 7 pool seems to work).
I tried GTAI 0.86d (?) on Xel Naga (NA) TvP, and noted that the AI still likes to remove the rocks at my own gold expansion (thanks very much, AI).
Improved is that it removes the rocks to its own 3rd, which the old AI never bothered to do.
After I killed its natural and 3rd, and killed its army, the AI did not make a new nexus, instead spent all its remaining mineral income from its almost mined out main on re-growing its army, then ran out of minerals and was stuck. It should have seen it had poor income and re-expanded, as a new priority, or even long distance mined.
Looks very well done. A real must for my friends to become better :D
Either way, wouldn't it be possible to implement the automaton bot in this AI? Then you could have an insane AI, who doesn't cheat but just wins by insane micro.
I just played a game vs GTAI 0.86d on xel naga on NA server and the zerg AI 7 pooled, built a bunch of zerglings and sent it to its ramp and sat there and tried to transition into an expansion. But I had a good game vs the terran ai. I hope one day we'll be able to see the AI harrass effectively through drops/banshees/mutas etc
@nyc863 : Think, next time will add a re-evaluate value for resource income, if it is two low, AI will get priority for build Base first. @Nexic : thanks for uploading, did the AI micro in your map ? ( Stuff like hit and run, I can tested yet since the unitdata.galaxy file needed for AI to implement these micro )
Basically it seems to be a wonderful hit and run AI but that's all, besides we could expect better hit & run from AI with 2000 APM !
Anyways it's really a good thing you are working on a gosu computer and I hope you good luck with all the hardwork you are putting into this programation Features look very cool also =)
ps : canon rush vs terran sucks because with 1 marauder you can deny, and besides you can also lift command center =) and you can counter protoss with 1 reaper.
I'm impressed by this AI you made. I normally wipe out very-hard and insane AI super easily but I was just playing the same as I would on those difficulties versus your very-hard GTAI on metolopolis TvZ and that zergs macro was insane. The mutas came and I was like ok a few turrets to ward them off while I take a super fast third, no problem I thought...until I counter attacked while it mass muta'd my main and suddenly instead of turning back the muta's kept attacking my main and suddenly I realized he's on 3 bases too. TT
I love how its finally challenging to play a computer bot! :D Next time I'm gonna start out one base all in and then move onto macro games, can't wait to see how it holds its own versus my terrible diamond play. Well done man I'm very happy to have a worthy AI opponent to practice with!
Its good as it doesnt practice normal timings so you cant meta game it, but i cant park a ling outside its base without it dying within a certain range. It says i cant attack the AI that early, then 30 seconds later and attack would be on it way, if i manage to hold it off then i try again and it still kills it .
New update, adding 3 command to get list of commander "-lp": List of Protoss Commander "-lt": List of Terran Commander "-lz: List of Zerg Commander
When you use these command, the game speed will automatically reset to "Slower" After using select comand, ex "-se 2 11", the game speed will automatically reset to "Faster".
Hey, first of all, great work and thanks for the AI.
My question is if there is a way to see the BO of the different commanders? IMO the short descriptions are nice but it would be better to really see what the commander is going to do.
Facing the mass marine build a couple times as zerg.
First, the computer doesn't take into account spine crawlers when it decides to attack, so it was easy to thwart.
Second, the marine splitting is so good banelings are a waste of zerg resources. This is both good (good the AI is so great at micro!) but bad because baneling is the counter to mass marine until you are in the Grand Master's league. And even Grand Masters will clump marines at times.
And third, the computer seemed to ignore my ling counter attacks in its mineral lines for quite some time, choosing to just mass attack, not split off newly produced marines to clean it up.
This is getting better and better. Really love the Commander list command. Now I'll just have to wait for someone to upload it on EU . I'm trying to focus on improving my mechanics and this tool is a great help to do so. Keep up the great work ptanhkhoa, I really appreciate it.
Thanks for the AI update, but like Fyn AI it has the same major problem in my opinion which is the lack of APM cap. This is most problematic when you're practicing vs a Zerg. You can see at the big battles right at the end that the Zerg AI is losing alot of units but his food count is jumping up and down constantly between 160-200 since the AI has no cap on its APM to micro every individual unit on the field. inject larvae at every second its available and produce units all at the same time.
Nonetheless, I'm happy about the update but I'm hoping for a APM cap in the future.
It would be insanely interesting to have several people program AI for SC2 and then have each AI dedicated to a race and have the programmer's AI fight each other in some sort of tournament.
hey ptanhkhoa, your AI has been really coming along. When I try to out macro pvt, the terran really keeps up and punishes if I try to make too many probes. Your AI is totally baller!
I was wondering if you could implement stutter step. At first I thought the AI micro would be really strong, but now I find I get a lot of free shots at marine/marauder cuz they always just moving away
also as a minor thing, terran (at least for 3 rax build) will get concussive upgrade as well as weapons and armour but won't start stim until 13min in the game!
On June 23 2011 07:45 goal 888 wrote: Wats AI apm?
in pvt, i've seen the first 15min an average ~150 but during battles apm spikes to 1700, after 15min mark with maxed army, i saw apm spike as high as 2400!
@Mackus: It's only availabe in the newest version GT AI 0.87, you can see the version of the GT AI at it show up at the beginning. @Eatme: Yes, only "vision", but that pretty much, since AI have to guess your army value every time it didn't scout your base, and specially when you hide unit that it couldn't see.
Note: I couldn't find the latest version of Crevasse, unfortunately.
I'd like to publish GTAI versions of Bel'Shir Beach, Crevasse, Crossfire SE, Dual Sight, Tal'Darim Altar, and Terminus SE but can't find any unlocked versions. I would also be able to import the single player version of YABOT to the GSL maps and upload them for everyone.
I'd appreciate it if someone could find out where to download unlocked copies of the GSL maps. (Also looking for iCCup Testbug).
I'm pretty impressed at how you got the AI to use those forcefields. What are the situations in which the units are commanded to lay them down? Is it random?
New Version Green Tea AI 0.87b: *Adding new strategy for Zerg AI.(Increase to 17 Commander). Type "-lz" to get list of Zerg Commander. *Focus weakness unit, (Marauder, Immortal will aim for armor unit ) *Fix Roach dancing bug.
*Note:In new version if you see this line: "Load 'unitdata.xml'. Enough Micro Control"
Then you are loading map correctly and Micro part of AI is enough.
If not, it mean the "unitdata.xml" file is missing and you have to import the map again.
Note: I couldn't find the latest version of Crevasse, unfortunately.
I'd like to publish GTAI versions of Bel'Shir Beach, Crevasse, Crossfire SE, Dual Sight, Tal'Darim Altar, and Terminus SE but can't find any unlocked versions. I would also be able to import the single player version of YABOT to the GSL maps and upload them for everyone.
I'd appreciate it if someone could find out where to download unlocked copies of the GSL maps. (Also looking for iCCup Testbug).
can you add some extra identifier to the map title (like add "lys" or something)? I tried searching GTAI, GTAI 0.87, etc but it's just turning up a load of other junk results (older GTAI versions).
I just played GTAI 0.86 on XelNaga and it played poorly. It build a gas early on at it's natural expansion, probably because I gas stole his 2nd gas.
It didn't build any zealot (not wall), and jumped straight to a stalker.
My 4 scout zerglings arrived at his base a bit after stalker finished producing. The stalker attacked the extractor I stole gas with, then my zerglings arrived. My zerglings started to attack the stalker, but he ignored them and kept attacking the extractor. I then proceed to kill a bunch of probes with just 4 zerglings because his worker micro was bad. He then build a zealot and built a twilight council even though he was (or could have been) on the verge of death. Right after my 4 ling scout attack, he must have felt he had the edge and sent out an attacking army of 2 zealots and a stalker (WTF?).
After that, I assume he tried to proxy pylon DT rush, except the probe ran out all alone and built the pylon in a completely open, common path, in front of my army of roaches, and in front of my creep.
The issue is that there seems to be absolutely NO reactive sense in the AI at all. I wouldn't call this an AI, but just a script of events it's ordered to execute, with hardly any actual feedback.
Some of it's actions seem somewhat dumb too... if I send a army of roaches to attack probes, the probes will suicide themselves on the roaches (this is when the protoss had an army)... they didn't actually check the value of the attackers to see whether they'd win or loose. That same behavior made it easy to get kills with zerglings too, since I could focus fire a probe (it wouldn't run away), and then run away before the other probes deal too much damage (this of course has limited effectiveness when its' 4 lings vs 16+ probes though).
I was expecting a much much better AI after over a year of working on it. It feels hardly any different from when it was upgraded pre-SC2's release.
@Xapti: The AI didn't react to stole gas => So the script fail after that. AI will use worker to defense if there are no army nearby. About the Proxy Pylon, it can only detect the distance to build the Pylon, but cannot detect "the safe place" that only human can sense it. I admit that AI has some weakness, but you shouldn't just take advantage of it. The army system measure in GT 0.86 has some error, the later version fix that.
Is there a guide on how to compile the maps myself?
I'm not sure where I can locate the original ladder (or any other map for that matter) files to open in the map editor so I can add the AI files from the official GTAI site.
The 0.87 maps I published were all set to private, I'm republishing them all as Public now. If you search for "GTAI0" the ones I published will be closer to the top.
On June 25 2011 19:45 ptanhkhoa wrote: @Xapti: The AI didn't react to stole gas => So the script fail after that. AI will use worker to defense if there are no army nearby. About the Proxy Pylon, it can only detect the distance to build the Pylon, but cannot detect "the safe place" that only human can sense it. I admit that AI has some weakness, but you shouldn't just take advantage of it. The army system measure in GT 0.86 has some error, the later version fix that.
I am aware that it's important to not take advantage of AI.
I play against AI like I play against humans. The fact that the GT AI cannot cope with a gas steal and blizzard's can is a problem; same thing with workers — if there's no nearby army AND the probes' strength is less than the attacker's strength, they should run instead of fight. If there's an army nearby, then I guess it's a matter of choice whether to fight or not.
As far as I know, the gas steal did not make the script fail at all (by my definition of fail). It just made the build order programmed into the script less viable. I guess it messed up the stalker, but that's maybe something in the programming to prioritize proxied buildings or something? In situations like that, the AI should perhaps change it's build (although it doesn't have to). The gas was not being mined by probes as far as I know (I should maybe double check, I might be wrong), so I don't think it messed things up in that sense.
With regards to the proxy pylon... I recommend it should not be programed to execute a proxy pylon, if it cannot figure out safe area to put them (or if it's safe to attempt a proxy in the first place, which is more complicated) AND doesn't have an army to defend it. For instance, if it was a 4 warp-gate proxy pylon, it would be fine (and not need to be hidden) because there will be an army to protect it. However when the AI has little to no army, it should NEVER attempt a proxy pylon (as long as it can't figure out what places are safe to put them, and how to get to those spots safely without being spotted)
I don't mind AI's having weaknesses, such as not noticing un-advantageous attack positions (a major problem in custom brood war AIs that used BWAPI), or not being able to cope with wall-ins (which can be handled better in SC2, but still I don't necessarily expect it to be handled yet) or other things, but if the AI's weakness is that It plays completely blind (I know it has SOME feedback, fallback measures, and other stuff, more than I'd think, even I suppose) or close to it, that is not acceptable. An AI should be adaptive, and predicting— • Opponent's zerg? get some zealots to start with early on, not a stalker • Scouted (or maphack spotted) a roach warren (and/or more-so roaches)? cut zealot production, buff stalker production, build robotics facility or stargate. • Got gas stolen? switch to a low gas build, or if running a low gas build, ensure one doesn't switch to a high gas build. Two good builds being fast expand or 4 gate timing attack.
I am saying these things for 2 reasons: 1. I really want AI to be good in SC2 2. I high high expectations for Green Tea AI considering SOME (I know terrain obstacles is hard to deal with from what I understand) of the great features of SC2 editor/AI. Another reason for the expectations, is because within the first 1- 2 months of the AI being made... it had really excellent improvements and big changes. Extrapolating that period of time to the present, the development just seems terribly slower and/or less pronounced.
I hate how both toss and terran AI tend to go Battlecruisers and Carriers as soon as they know you got bad/worse AA. That is so wrong. Terran will ignore making tanks and will make BCs vs Zerg?!?! Toss likes to make Carriers vs Terran that goes mech but does not make Vikings early. But does not stop making carriers once you do get plenty of vikings.
• Select commander doesn't seem to be working right. the chat text is cleared, but there's no other feedback acknowledging that the AI switched, nor does their build seem to change.
• Restart doesn't seem to work either, although I don't know what it's supposed to do (difference between quick restart and restart?)
• Wall-zone is implemented really poorly. If you're going to put effort into making locations, you might as well tell the AI the coordinates to build their buildings so that they can wall on their own. Otherwise, AT MINIMUM make a "force field" blocker that only lets their units move in and out, not the opponents'. Even just a instantly move units to location would work better, or even just force order units to move back. The location of the wall zone is also poor. Protosses cannot wall ON a ramp (unless they built like 3 zealots), they can only wall above one, so it's stupid that units will die as soon as they move on the ramp, as opposed to going or square past the ramp.
So, about what level is this AI considered to be similar to on medium difficulty? ( in terms of ladder rank; gold, diamond, plat, etc.) Because I cannot beat this thing on medium or even easy The micro is even more insane than I've seen any pros do even on easy difficulty. I don't play much 1v1 as it is and I'm in the gold league, but I'm just curious as to what level the AI is considered to be playing on.
I've published more maps to NA. Search for "87b" to easily find them all. GTAI 0.87b Mapname, publisher: Lysergic.
• GSL Bel'Shir Beach • GSL Crevasse • GSL Crossfire SE • GSL Dual Sight • GSL Terminus SE • GSL Xel'Naga Fortress • iCCup Testbug • Shattered Temple (MLG version) • Tal'darim Altar LE (removed rocks from third expo)
These will be updated to GTAI 0.87b and uploaded soon: • Backwater Gulch • Metalopolis (MLG version) • Shakuras Plateau • Typhon Peaks • Xel'Naga Caverns
I could also add a couple 2v2 maps if there are any requests.
Suggestions to OP: •Half the time you face the Zerg AI, it does some sort of 1 base rush/cheese - there's not enough macro builds. I think you should remove the Drone Rush and 7 Pool Commanders. (Also, the frequency of BCs and Carriers are annoying too.)
•Select Commander isn't working as far as I know. I think you should make the game restart when you select a commander, so the ai will be able to follow its new build correctly.
•Very Easy AI doesn't seem to do anything - I only tried a couple times but they just made a few workers and nothing else.
•Restart does nothing, Quick Restart works though - it would be nice if you could implement a game clock reset with it.
On June 27 2011 09:42 Dibond wrote: So, about what level is this AI considered to be similar to on medium difficulty? ( in terms of ladder rank; gold, diamond, plat, etc.) Because I cannot beat this thing on medium or even easy The micro is even more insane than I've seen any pros do even on easy difficulty. I don't play much 1v1 as it is and I'm in the gold league, but I'm just curious as to what level the AI is considered to be playing on.
Diamond if you play against it properly. As in not being annoying with scout/gas steals and doing early pushes against it and not abusing the ai in general. It macros quite well but if you defeat its push and counterattack it will lose, so dont counterattack . I strongly recommend not attacking the ai for the first 10 minutes or so of the game so it can build up its army and economy. It makes for a much better challenge.
this AI is tough! i am low diamond and have problems beating it on medium.
but i just want to point one problem i have noticed, the ai will ignore probes and focus down my nexus which is not what a real player would do. Especially if my warpgates are not disabled and my army is not all dead.
Something extremely strange happened today on Shakuras Plateau, PvT. I pressure with a probe, 2 zealots, and a stalker, but when I get halfway up the AI's ramp two of my zealots INSTANTLY evaporate at the same time as if his marines started doing 100 damage. He had a bunker up, so yes, I would have lost that engagement, but it was the rate at which my zealots died that seemed odd.
The select commander work for me, when you see the line: " [Blue] Select Commander xx " then you're doing right way. " [Blue] Select Random ", that you must do something wrong. it can work only with the first 10 seconds.
Yeah I was thinking the Select Commander might not of been working for me because I wasn't doing it early enough. Having to do it within the first 10 game seconds (7 real seconds) isn't enough time to do all this:
-Make a Drone -Split workers -Overlord scout -Control Group Hatchery/Overlord -Set rally points -Read the list of Commanders "-lp" -Select Commander "-se x y"
That's why I suggested the OP to implement a map restart when you select a Commander. This would make the map more like YABOT (you can choose a new build for the AI and restart the map). I think the game clock should reset too.
It'll give you enough time to choose a build. It'll let you restart the map whenever (if you're about to win or lose, if you mess up, or if you don't like the Commander picked, etc).
On June 27 2011 10:32 lysergic wrote: •Half the time you face the Zerg AI, it does some sort of 1 base rush/cheese - there's not enough macro builds. I think you should remove the Drone Rush and 7 Pool Commanders. (Also, the frequency of BCs and Carriers are annoying too.)
Regardless of zerg... I would say that ALL the races have too many commanders. Commanders like protoss cannon rush is TERRIBLE because AIs do not know how to cannon rush. The list of random commanders which are chosen automatically should be made much more lean.
If people want to choose a specific commander that it doesn't normally choose from, that's fine, but many of these commanders should NEVER execute automatically. At the minimum they should execute much less than other commanders. More is not always better. Quantity ≠ quality.
One of the biggest issues with commanders though is with regards to not accounting for the opponent's race & build-order. commanders should be selected based off the opponents race (Don't cannon rush a terran, although like i said before, they shouldn't cannon rush at all). For an "AI" to actually be an AI, it needs to have intelligence. Intelligence consists of the knowledge of countering different things, and the ability to see such things (feedback).
I just played another game against it and it moves out with 2 phoenix, a sentry (or two) and 2 zealot.
As if that wasn't bad enough, it also didn't even used graviton beam except for once later on, which was on my LARVA.
As if that wasn't bad enough, it seems like most builds the AI runs it gets a robotics facility for some stupid reason... (apparently it transitions to immortal later which I guess is fine, but robo bay was a bit early to get) it doesn't even use the robo bay I find. Both this phoenix attack, as well as "2 gate expand" get a robo bay and don't even use it. It's especially strange with the "2 gate expand" build which goes 2 gate robo, then might attack or just sit on it's ass, THEN expand at like the 8 minute mark (really late for a 2 gate expand).
The 2 gate expand build needs to be modified, and the weird-ass phoenix build with robo should probably be removed (assuming there isn't a build that gets voidray+phoenix instead already, otherwise switch it to that).
Overall this is the problem with most of the commanders... It's just quantity over quality. There may be a bunch of commanders but they run trashy builds and/or are not executed well (I saw one game where toss got 3 bases and intentionally decided not to mine gas, preferring mostly mass zealot (which obviously got stomped).
yea terran is OK but zerg ai is terrible. Half the time it does a 1 base all-in with no transition. sometimes itll make 2 spawning pools and an inbase hatch...
@Xapti: The programmer take a good jobs in making AI that far better than the original Blizzard AI. If you find that you are a real strategist, why don't you make and AI by yourself if you thinking of making AI is easy. I understand it take him much time to develop this excellent program, but he need time to do other thing. Instead of complaining , you should support him. At least by action, there is a 9 combination in SC2 Z/P/T vs Z/P/T, so for each combination, let say 6 strategies, then 54 strategies is required. Do you have time to making this list by yourself and send to him ? No, I assume.
@Ptanhkhoa: Thanks for the great AI, but could you set the choosing time a little longer, for me 10s is too short. 15s is good enough.
On June 28 2011 04:41 lysergic wrote: Yeah I was thinking the Select Commander might not of been working for me because I wasn't doing it early enough. Having to do it within the first 10 game seconds (7 real seconds) isn't enough time to do all this:
-Make a Drone -Split workers -Overlord scout -Control Group Hatchery/Overlord -Set rally points -Read the list of Commanders "-lp" -Select Commander "-se x y"
That's why I suggested the OP to implement a map restart when you select a Commander. This would make the map more like YABOT (you can choose a new build for the AI and restart the map). I think the game clock should reset too.
It'll give you enough time to choose a build. It'll let you restart the map whenever (if you're about to win or lose, if you mess up, or if you don't like the Commander picked, etc).
I can do this. but it will mess up with the GT opening code. So I can increase the time to 15 seconds. I think that a reasonable time before the first drone is build, otherwise it will messed up with some build order. Also, when you time "-lp" at the begining, the gamespeed will reset to "Slower" give you more free time.
I update to GT AI 0.88 version which give more detail descriptions, also adding "-nc" (no cheese) command that cross out the cheese strategy from Commander list. Thanks for uploading map on the US Server ^_^.
@Xapti: Sorry, my time to develop on this project is very limited ^^. Your idea is very good, but thinking that instead of developing the strategy by my self, i will make the AI that can actually learn your BO xD, so the more you play, the AI will be better.
@ thisisSSK : Zerg AI has balance strategy between Rush and Expand, so do some more match, or in the GT AI 0.88 version, you can type "-nc" to cross out the "cheese" strategy (6pool, 7pool,... ) , so i will use more standard "2 base" opening.
@Ptanhkhoa: Try new GT AI 0.88, just increase the time to 15 seconds. Also you should using with "-lz", "-lp", "-lt" command to set the game speed to "Slower" and give you more time. After typing "-select x y" the game will be "Faster" again.
On June 28 2011 14:00 alonestar_ak wrote: @Xapti: The programmer take a good jobs in making AI that far better than the original Blizzard AI. If you find that you are a real strategist, why don't you make and AI by yourself if you thinking of making AI is easy.
First of all, if I knew 1 year ago that in 1 years time, starcraft 2 would have only one (and a half) melee AI (as far as I know at least), made by mostly just 1 person with quite limited time, I would have actually dedicated a lot of time to learn about it, if not work on it myself. However, I did not expect this future, and actions are best done according to predictions and plans. The major lack of SC2 AI developers has utterly surprised me. I thought many Starcraft 1 people would move to it, as well as a whole bunch of other people who want to innovate, help the SC community, learn things, program, get famous, etc.
Secondly, Criticism is one of the the best forms of support, as that is how it can be improved. I specifically posted in a previous post that I care about this AI and want it to get better (among other things).
Lastly, Blizzard's Very Hard AI is better than some of these commanders (Blizzard AI is also programmed based off no strong metagame, and before a bunch of nerfs and buffs, such as reaper and roach). One of the biggest issues is that the AI has too many bad build orders, or just completely bad builds/strategies (cannon rush).
I wonder if you can implement MMM kiting? Most of the time the Terran army just splits up individually as the zealots chase marauders down, reducing dps time and thus gets annihilated. If the terran bioball would just clump up and stim+run the zealots will never be able to catch up.
On June 29 2011 22:17 LookNaph wrote: I wonder if you can implement MMM kiting? Most of the time the Terran army just splits up individually as the zealots chase marauders down, reducing dps time and thus gets annihilated. If the terran bioball would just clump up and stim+run the zealots will never be able to catch up.
While I have thought of the same thing, I think it might be rather difficult to implement.
The easier thing to deal with that you somewhat bring up though, is changing shooting time before moving for stutter-stepping. Marines have an extremely fast move delay after attacking, which makes them very effective at stutter-stepping, but the AI doesn't seem to take most advantage of it (although I can't specifically remember for certain).
Oh, and a serous problem/bug I noticed, is at least sometimes the AI (at least terran) will not rebuild their expansion if it got killed. In fact after taking their 4th position expo around or after the 3rd died, they never took any additional base, not the 3rd, not the gold, or any other, even when in a situation to do so, and in a situation with low money. Another time I saw protoss build 2 nexus at 3rd expansion for some reason... I'm thinking it might have been intended for the gold, but it was on the wrong side of the cliffs .
I find the AI is very bad with managing their money that way... they just try to expand a lot while pumping units, never being more defensive or careful if they are about to run out of money, granted I don't think this sort of thing is much of a priority, mostly just rebuilding expos is important.
Another small issue is that AI attacks neural parasited units a bit. While that might not be an issue in itself, when it wears off, they will continue attacking it, even chasing it. Not sure if it eventually stops, or if it kills the unit entirely.
On June 28 2011 14:00 alonestar_ak wrote: @Xapti: The programmer take a good jobs in making AI that far better than the original Blizzard AI. If you find that you are a real strategist, why don't you make and AI by yourself if you thinking of making AI is easy. I understand it take him much time to develop this excellent program, but he need time to do other thing. Instead of complaining , you should support him. At least by action, there is a 9 combination in SC2 Z/P/T vs Z/P/T, so for each combination, let say 6 strategies, then 54 strategies is required. Do you have time to making this list by yourself and send to him ? No, I assume.
His post is filled with more constructive critisicm than yours. If the OP wants feedback, this post would be extremely useful.
To the OP, Good work but I am not that impressed so far. I will write up a more descriptive post when I get home.
Awesome work on this, had some fun messing around with a few new builds though I'll confess I'm far too slow to manage a decent commander selection at the beginning. I find some AI builds provide for far better games than others for whatever reasons, like with some openings the AI follows up properly and others it doesn't or something ^^
It's got some sick burrowed Roach micro lol, watching a replay and it's doing 1.3k apm during a big battle heh.
There are many things need to improve, but first, i modify the AI Micro a little bit: make all unit move to the same direction, using vector system instead of spiting unit in all direction. Also adding Order Build Recording, so you can make AI do better Opening. After implement the build order, i think of making AI doing general AI scripts that can counter every race, but it will take more time and tester for developing
On July 01 2011 08:24 ptanhkhoa wrote: There are many things need to improve, but first, i modify the AI Micro a little bit: make all unit move to the same direction, using vector system instead of spiting unit in all direction. Also adding Order Build Recording, so you can make AI do better Opening. After implement the build order, i think of making AI doing general AI scripts that can counter every race, but it will take more time and tester for developing
Excellent! This AI is getting more and more advanced and better too. Couple flaws I'd still like to point out: -The Build Recording system is awesome, but it also means that you have to play other races and then perfectly execute the build you want the cpu to learn. This sounds like an incredible hassle to me, so I suggest some kind of way to edit/import BO's like YABOT. Honestly I like this GTAI much better than YABOT because its so much more advanced ^_^
-This is just a minor critique: the micro of the computer might be TOO good. I don't know if theres a fix for this but if so, it might be better to actually lower the micro level because it becomes too inhuman and unrealistic.
-I haven't tried it on the new version but from what I remember the -qr and especially the -re were broken on some maps
Thats it! Again, I commend you on your hard work and dedication Please keep at it!
2) like lysergic already said there are to many strange commanders in the pool. i think its better to pick a random commander out of a smaller pool and let the other in for direct selection.
Tbh, this isn't very useful for practice purposes. The fact that the AI has an income cheat ends up encouraging you to over produce units. I've picked up this bad habit back when I used to practice against the regular insane AI.
@AndAgain: Ofcourse, that is your choice for more challenge you play at harder difficulty. But from Medium below, AI have no minerals cheating. @ thisisSSK: Well, i will think of adding a manually adding BO, beside playing and Recording. @ skeldark: yeah, when I think I get up the data recording from pro tester, I will replace the old BO with the new professional BO.
Oh by the way: as the AI gets better, especially when the worse builds are removed and/or improved, the difficulty settings should be re-adjusted.
As it is now, there's only 3 settings for non-cheating income, but as the AI gets harder, people will almost never want cheating income AI. Even now, I think it makes sense for hard to be the best non-income-cheating AI right now, because it beats many players (I think diamonds?). Blizzard's AI goes to very hard without cheating even, and this is on par or better in some cases than that.
Hence, I think only insane should have cheating income, which should be like 10% to start, and 1 additional percent every minute later. If extra-hard modes are desired, then just have them accessible in-game from chat commands.
Frankly, to build a real AI, none of the difficulty levels should have income cheats OR map hack.
If you get rid of a lot of the junk builds and just focus on some core solid builds the AI should be able to macro flawlessly. Add in a micro-bot for some sick micro and that will already be quite a challenge even without counting strategy & tactics. Basically the AI should be mechanically superior to any human but strategically and tactically inferior. Adding in the strategy and tactics decisions when the AI has incomplete information (no maphack) would be the final stage.
IMO there needs to be a smoother difficulty curve. Right now the jump from medium to hard to very hard to insane is too much due to the ridiculous income cheats and map hacks. It's far better in the long run to do away with such gimmicks and actually make the AI react to scouting information. But that's for the strategic/tactical programming. As far as a basic method of differentiating difficulty, its easy enough to program in some common mistakes that players of lower levels have. For instance:
-Every time a worker finishes building, set a random number of seconds between say 0 and 20 before it starts the next worker. You can do a similar thing for production buildings. This approximates the Bronze League mistake of forgetting to make workers and/or constant production out of production buildings. Add the delay into more things like MULES, Injects, Chronoboosts, etc. and you basically have bronze league macro.
-Add a condition where the AI doesn't build units/workers when its fighting. Another common mistake.
-Add in occasional supply blocks.
-Add in floating minerals/gas and not enough production buildings.
etc.
Once you've added these mistakes for the AI, the difference in difficulties can be managed simply by toggling these conditions on or off or tweaking the numbers. For instance, the Very Easy AI would have all of these conditions on. The Easy AI might have the worker/production delay between 0 and 10 seconds instead of 0 and 20 and perhaps the supply block condition toggled off. The Medium would have constant worker production, not get supply blocked, and enough buildings, but forget to build stuff during battles. So on and so forth.
I'm still iffy about APM limits since part of the fun of AI's is to use human's superior strategy and tactics to overcome the AI's huge mechanical advantage. But either way, the ideal result would be that the difficulties should roughly be:
Very Easy -> Bronze Easy -> Silver Medium -> Gold Hard -> Platinum Very Hard -> Diamond Insane -> Master Super Insane -> Grandmaster Bonjwa -> Bonjwa
The last two can be toggled on by in-game commands. =D
It would be useful if you can list out the current BO which build is bad, and which need some modification. For now I'm actually clueless, and haven't much time for testing ^^.
What does "Wallzone" with a timer mean? It appears vs Terran but I have no idea what it does and can't find anything on this subject. I would suspect something with Terran making a walloff but that doesn't seem to happen when playing the AI.
thanx for the AI maps, I used it in macro or die map, and will download asap for regular map practice. I think i found a bug considering .82 version( i think) in macro or die, dont know if it was fixed: my protoss opponent had his natural blocked by a semi-completed bunker, which he ignored and tried to get his natural(late) in a different location.
There are so many version on EU out now. Can you make one "official" upload for each map and update it, so we don't have to search for the newest version?
thx
@Sevenofnines if they try to make a good ai i think its hard for the guys to motivate them self to put work in it so its act bad ^^
hahaha this is awesome. So much better than Blizzard Ai. I have been beating the zerg and protoss but get tottally owned by the terran. The marine splitting is crazy good, but it is nice to play something that doesnt just A-move.
* Do you code in GUI or Galaxy? * How much do you use the built-in AI commands and how much do you actually do yourself? * Do you use the Issue Order or Issue AI order for the orders?
-Add in floating minerals/gas and not enough production buildings.
just no. thats a terrible idea.
terrible idea now; though when the AI improves its strategies and tactics (especially in harassment, flanks and surrounds), these would then become incredibly useful things to implement.
So what level should the Great Tea Ai on medium be considered? As plat I'm getting stomped by GTAI on medium... is it comparable to diamond level? (assuming you don't exploit weaknesses in the AI).
Great AI, although it has some weird builds. The '-re' command is not working for me in the 0.9 version, and when I do '-qr', it fast restarts but with the gamepeed on low.
On July 02 2011 13:40 skeldark wrote: @Xapti blizzard ai use income-cheat!
Only insane uses income cheat. When I mentioned Blizzard AI, I mentioned the very hard difficulty, which is the hardest difficulty before AI uses income cheat. what I was saying is that green tea should only have one income cheating setting (insane) just like blizzard's (or additional income from game chat commands if desired).
On July 01 2011 18:34 Sevenofnines wrote: Frankly, to build a real AI, none of the difficulty levels should have income cheats OR map hack.
If you get rid of a lot of the junk builds and just focus on some core solid builds the AI should be able to macro flawlessly. Add in a micro-bot for some sick micro and that will already be quite a challenge even without counting strategy & tactics. Basically the AI should be mechanically superior to any human but strategically and tactically inferior. Adding in the strategy and tactics decisions when the AI has incomplete information (no maphack) would be the final stage.
IMO there needs to be a smoother difficulty curve. Right now the jump from medium to hard to very hard to insane is too much due to the ridiculous income cheats and map hacks. It's far better in the long run to do away with such gimmicks and actually make the AI react to scouting information. But that's for the strategic/tactical programming. As far as a basic method of differentiating difficulty, its easy enough to program in some common mistakes that players of lower levels have. For instance:
-Every time a worker finishes building, set a random number of seconds between say 0 and 20 before it starts the next worker. You can do a similar thing for production buildings. This approximates the Bronze League mistake of forgetting to make workers and/or constant production out of production buildings. Add the delay into more things like MULES, Injects, Chronoboosts, etc. and you basically have bronze league macro.
-Add a condition where the AI doesn't build units/workers when its fighting. Another common mistake.
-Add in occasional supply blocks.
-Add in floating minerals/gas and not enough production buildings.
etc.
Once you've added these mistakes for the AI, the difference in difficulties can be managed simply by toggling these conditions on or off or tweaking the numbers. For instance, the Very Easy AI would have all of these conditions on. The Easy AI might have the worker/production delay between 0 and 10 seconds instead of 0 and 20 and perhaps the supply block condition toggled off. The Medium would have constant worker production, not get supply blocked, and enough buildings, but forget to build stuff during battles. So on and so forth.
I'm still iffy about APM limits since part of the fun of AI's is to use human's superior strategy and tactics to overcome the AI's huge mechanical advantage. But either way, the ideal result would be that the difficulties should roughly be:
Very Easy -> Bronze Easy -> Silver Medium -> Gold Hard -> Platinum Very Hard -> Diamond Insane -> Master Super Insane -> Grandmaster Bonjwa -> Bonjwa
The last two can be toggled on by in-game commands. =D
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ this a thousand times. I loved playing the BW AI because ( as far as i know) it doesn't use cheats, it was just infinitely mechanically superior to me. In some daily Day9 said that the standard is the way to get a maximum army value, and any interesting tricks took away from that standard army value. so program the standard into the AI for each race, and simply limit its APM to perform that standard depending on the difficulty.
On July 03 2011 08:53 cekkmt wrote: this a thousand times. I loved playing the BW AI because ( as far as i know) it doesn't use cheats, it was just infinitely mechanically superior to me. In some daily Day9 said that the standard is the way to get a maximum army value, and any interesting tricks took away from that standard army value. so program the standard into the AI for each race, and simply limit its APM to perform that standard depending on the difficulty.
That's not true, in BW AI, I can easily beat 2, or even 3 AI at the same time. Also, I have no problem in beating 2 Blizzard Very Hard AI in Starcraft 2. But GT AI is much harder to deal with even on 1 vs 1 on Medium difficulty if it choose the right build.
On July 03 2011 08:53 cekkmt wrote: this a thousand times. I loved playing the BW AI because ( as far as i know) it doesn't use cheats, it was just infinitely mechanically superior to me. In some daily Day9 said that the standard is the way to get a maximum army value, and any interesting tricks took away from that standard army value. so program the standard into the AI for each race, and simply limit its APM to perform that standard depending on the difficulty.
That's not true, in BW AI, I can easily beat 2, or even 3 AI at the same time. Also, I have no problem in beating 2 Blizzard Very Hard AI in Starcraft 2. But GT AI is much harder to deal with even on 1 vs 1 on Medium difficulty if it choose the right build.
How can you say it's not true that the BW AI was a finite-yet-significantly superior to him (since I'm sure you understand that he used a hyperbole)? He was not saying that SC2 AI sucks and BW didn't.
What he seems to be saying is that he sucks (at both SC1 and SC2), and wants much easier versions of GTAI. Even though I'm aware that you probably do not specifically desire an easier mode for GTAI, I don't think you wish to disallow such a thing to occur for them.
that said,
On July 03 2011 08:53 cekkmt wrote: In some daily Day9 said that the standard is the way to get a maximum army value, and any interesting tricks took away from that standard army value. so program the standard into the AI for each race
Is it possible to have it updated on the NA server? I love using 0.86, but I would like the up-to-date version. Thank you for this great tool! Love it.
On July 02 2011 18:58 Siretu wrote: * Do you code in GUI or Galaxy? * How much do you use the built-in AI commands and how much do you actually do yourself? * Do you use the Issue Order or Issue AI order for the orders?
- I coded in galaxy file. - The built-in AI commands is not enough for creating AI, so I creating my own function and variables that can be more useful. You can find these on the GT AI core folder. - Most of action in game used by the Casting system ( AICast function ),
These new features included in the GT AI 0.92 version:
_ Creating your own strategy and add in to AI. Also you can upload and share these new strategies with your friends. => The AI basically is not strong enough, but as you improved it, the AI will be a lot better with your Strategies.
On July 01 2011 18:34 Sevenofnines wrote: Frankly, to build a real AI, none of the difficulty levels should have income cheats OR map hack.
If you get rid of a lot of the junk builds and just focus on some core solid builds the AI should be able to macro flawlessly. Add in a micro-bot for some sick micro and that will already be quite a challenge even without counting strategy & tactics. Basically the AI should be mechanically superior to any human but strategically and tactically inferior. Adding in the strategy and tactics decisions when the AI has incomplete information (no maphack) would be the final stage.
IMO there needs to be a smoother difficulty curve. Right now the jump from medium to hard to very hard to insane is too much due to the ridiculous income cheats and map hacks. It's far better in the long run to do away with such gimmicks and actually make the AI react to scouting information. But that's for the strategic/tactical programming. As far as a basic method of differentiating difficulty, its easy enough to program in some common mistakes that players of lower levels have. For instance:
-Every time a worker finishes building, set a random number of seconds between say 0 and 20 before it starts the next worker. You can do a similar thing for production buildings. This approximates the Bronze League mistake of forgetting to make workers and/or constant production out of production buildings. Add the delay into more things like MULES, Injects, Chronoboosts, etc. and you basically have bronze league macro.
-Add a condition where the AI doesn't build units/workers when its fighting. Another common mistake.
-Add in occasional supply blocks.
-Add in floating minerals/gas and not enough production buildings.
etc.
Once you've added these mistakes for the AI, the difference in difficulties can be managed simply by toggling these conditions on or off or tweaking the numbers. For instance, the Very Easy AI would have all of these conditions on. The Easy AI might have the worker/production delay between 0 and 10 seconds instead of 0 and 20 and perhaps the supply block condition toggled off. The Medium would have constant worker production, not get supply blocked, and enough buildings, but forget to build stuff during battles. So on and so forth.
Pretty much everything here I agree with.
Easy and Medium should both not have vision, and income cheats should be independent of difficulty setting (eg. make it something that can be toggled: "-ic 20" for a 20% income cheat).
I will be testing out this AI later after work. With my internet being in terrible shape it should give me some nice practice.
PS. What is the song that plays in the micro video? It sounds like a song from an anime series, but I would like to know what it's name and composer are
Hey man I am a big fan of this a.i and most people I've shown have been blown away! I agree with some statements about no map hacks and money cheats. I am wondering though, I downloaded the recent version I think 0.89 on crevasse? and about 10 .87 and every time the game starts in the text on the left one of the lines says. Cannot find unitdata.xml (a.i Micro disabled) I'm not sure why it does this, the a.i is still sick. But roaches don't micro or or marines don't split like on the video. Is there something extra I am supposed to download?
@chip789: Because these map missing the Unitdata.xml file. You can watch this video to see proper method to import the AI Script into map:
Then you can test it by pressing Ctrl + F9 in SC2 Editor or uploaded on Bnet by yourself.
I will thinking of modify the difficulty, but you can see that on the Hard level or above, AI don't get any bonus at the beginning (except Insane AI), So it still use the standard opening and you won't feel much different compare with Medium AI until 10 minutes after.
Alright so I was messing with this today and found that the AI macro's well, but doesn't respond well to typical muta harrass.
Basically it allowed me to take out all of it's production without ever moving marines to defend. I almost thought it was just gonna go counter attack since in the middle of my harrass he moved out toward my base, but then it just ran back. Seemed very indecisive.
I don't know if you can program the ai to react correctly to Muta harrass though, when not in combat or moving seige tanks should seige at tthe front of base and marines should go chase off the mutas. I basically won the game off of doing typical muta harrass and that's it. :O
(PS. Would still like to know what the first song in the micro video is)
Thanks ptanhkhoa though I haven't even touched the map editor yet....I've encountered this on every single GT AI map iv downloaded on bnet on north american servers...does that mean NA doesn't have a legit working Green Tea map? Cause I've searched hundred times for every GT AI map and all of them do the UnitData.xml error, I guess since I do care for this a.i I will follow your instructions and load them onto bnet. Perhaps someone with more experience could also assist?
On July 08 2011 14:22 chip789 wrote: Thanks ptanhkhoa though I haven't even touched the map editor yet....I've encountered this on every single GT AI map iv downloaded on bnet on north american servers...does that mean NA doesn't have a legit working Green Tea map? Cause I've searched hundred times for every GT AI map and all of them do the UnitData.xml error, I guess since I do care for this a.i I will follow your instructions and load them onto bnet. Perhaps someone with more experience could also assist?
I actually didn't get that error in the Shattered Temple .9 one I downloaded that Antman put up.
@Atlas_950: It seems that map missing some defend function (some of beta version). I just tested Terran AI vs Mutalisk harass and i see it used Marine to defense base.
The song in the Micro video is remix version of original classical music: Beethoven's Virus and Pachelbel's Canon.
I noticed while playing against the 0.9 version AI; the AI attempted to repair its hellions after a battle, but they set off across the map to attack again with the SCVs in hot pursuit!
Could you set it so that units being repaired are immobile unless being attacked?
@darkness: the uploader is Nexxy. One's of my scripting partner. @iHirO: I see, I will add the function to stop the units being repaired from moving in the next update.
A fun match of Terran vs Protoss Green Tea AI in Medium difficulty.
On Na Server, what is the title and author of the most up to date Green tea Ai Map? Latest and best working one I've found is 0.92 on Crevasse. Still does UntData.xml error though.
I think the terran 1 rax expand build should be thrown out, or at minimum modified. AI is just not properly build to be able to execute the build.
It built only 1 marine, build command center before bunker, and build a gas after barracks, all of which are terribly stupid things to do (although the late bunker is fine IF the opponent has no attacking units)
What's just as crazy, is the next game I played, terran seems to do the same build, except because I had some zerglings outside of the main, he suidiced his marine and a bunch of scvs. The weird part is that the AI then said "Atlanta: Want to play again, please?" or something like that. This is weird for 3 reasons: 1. It gave up because it couldn't fast expand, even though I didn't enter his main and threaten it in any way (aside from denying expansion). 2. I haven't seen a commander reveal their name for ages... (in fact never in these later versions) 3. Atlanta is a mixed-ground unit commander, not a fast expand commander... why would it run 1 rax gas expand?
Another time I played terran, it just stopped producing most stuff, if not everything, after a certain period of time (I looked at the end-game score screen that there was just an ass-load more unspent resources than me, when normally I would have like 5 times more unspent than them. I don't know what caused the problem though.
Lastly, the AI doesn't seem to respond well against expansion blocking (pylon, depot, creep, burrowed), although it's hard to say. Definitely has huge problems deal with gas steal. Sometimes it seems to act strange on it's own though regardless of (without) blocking stuff.
On July 10 2011 05:07 chip789 wrote: On Na Server, what is the title and author of the most up to date Green tea Ai Map? Latest and best working one I've found is 0.92 on Crevasse. Still does UntData.xml error though.
If you don't want build order recording feature, I think 0.87b is a good version to use- the uploaded 87b maps by lysergic don't have the micro problem. Otherwise, you should download the green tea AI yourself, follow the instruction to apply it to any map, and play via the map editor, or even upload it to battle.net.
Useful command to test with AI ( Type in chat box ) -rc : Record your BO -le : Save BO to slot. -re : Restart Map -qr : Quick map restart -ffa : AI use strategy on FFA map -aichat : AI will sometimes chat with you -aishutup : AI will remain silent -se x y : AI player x will select commander y; -do x y : AI player x will choose custom BO y.
Best A.I available people...this is awesome. I've shown it to gold/plat players and they just go "wow...how does it know how to do all that!" and then they get stomped. It's seriously a plat to masters training program. It's awesome! Thanks ptanhkho.
I'm research on add something called Coach into AI Scipts: So besides AI challenge, it can also help player by : _ Show Standard Build Order Step. _ Remind you when you forget to build worker, or train unit. _ Recommend unit build based on information you scouted.
I noticed that in Yabot which is GreenTea .84 I believe. The way the AI does the expansions are a little weird. They don’t expand to their natural they have a tendency to expand right there in the base. I don’t know how many times I ran into the protoss base and there was 2 or 3 nexuses sitting in the base. Is this something that has been corrected? Is it specific to Yabot?
On July 12 2011 06:41 p1cKLes wrote: I noticed that in Yabot which is GreenTea .84 I believe. The way the AI does the expansions are a little weird. They don’t expand to their natural they have a tendency to expand right there in the base. I don’t know how many times I ran into the protoss base and there was 2 or 3 nexuses sitting in the base. Is this something that has been corrected? Is it specific to Yabot?
I think it occurs in some maps more than others, as well as might occur if you run a form of map reset (as opposed to ending the game, and loading the map over again)
On July 12 2011 06:41 p1cKLes wrote: I noticed that in Yabot which is GreenTea .84 I believe. The way the AI does the expansions are a little weird. They don’t expand to their natural they have a tendency to expand right there in the base. I don’t know how many times I ran into the protoss base and there was 2 or 3 nexuses sitting in the base. Is this something that has been corrected? Is it specific to Yabot?
The Yabot Build Order structure is different from GT, since it can not distinct between expand and base location. But GT AI Record function is different, when you build another Nexus/Command Center/ Hatchery, AI know that you are mean to expand. Also it can recognize and divide the defense building like Photon cannon, Bunker, Spine in each base instead on concentrate on the Start Location.
I've been using GTAI a little lately to help refine some builds, and I think its really great, but I just wanted to point out something I noticed.
The Terran AI is really good, like super good, even when he does something stupid like battlecruisers i still occasionally lose to the sick micro and sheer numbers of units. I'm in diamond for reference.
The protoss AI is solid, almost never lose but its at least challenging.
The Zerg AI is laughably bad, and in many cases broken. I was playing against the "Charybdis" bot who made some lings and hydras off one base and then just sat there. Even at the 15 minute mark he was still on one base. Obviously zerg is harder to program than the others due to larva mechanics, but I've never been challenged thoroughly by it.
ANOTHER THING: the "wall" countermeasures in place are really dumb, generally. Sometimes I try to gas steal and it tells me I can't build in their base. Gas steals are perfectly legitimate things to add to build orders, why can't i practice that? Also a lot of the invisible walls that kill your units are stupidly placed. The big one thats a problem is on Tal'Darim altar, the ramp outside of the natural leading to the middle is apparently a wall that i'm not allowed to cross, even though I've never seen anybody actually build a wall there before (you can just go around it through the third).
@Morphling: The AI just have some basic build included, but it main feature is you can creating strategy by yourself. So the AI will be better if you put some effort and adding more pro build into AI.
You can adding it by 2 ways: * Do a normal game and record your build.
Or using GT Encoder tools to creating BO code and copy into GT Banks File.
You can see that any strategy will be obsolete with time go on. As for me, i can not changing AI strategy every time SC2 update. So with this tool, you can easily adding the new strategy and AI can keep up with the modern build.
@ ptanhkho: Could you give me some more detail how to public the map since i'm not experience with SC2 Editor. I would like to be your official uploaded on NA Server too. Otherwise, it will be many GT AI Maps on Bnet that make us confuse.
On July 12 2011 23:22 ptanhkhoa wrote: You can see that any strategy will be obsolete with time go on. As for me, i can not changing AI strategy every time SC2 update. So with this tool, you can easily adding the new strategy and AI can keep up with the modern build.
In my opinion an AI should be able to run builds on it's own, with just a bit of guidance... AI just needs to know basic openers like forge fast expand, 2 rax, an everything else should be almost entirely dependent on what it scouts, as opposed to what it was told to do. Having multiple long-term build orders programmed into an AI is just a guaranteed way to never attain good success rates, because it's like playing rock paper scissors with your eyes closed.
That said, I'm aware that it takes much more work to program "real" intelligence than to just get a machine to roughly mimic things, but I think it's important, and/or worth it overall.
On July 12 2011 16:39 Morphling_ wrote: The Zerg AI is laughably bad, and in many cases broken. I was playing against the "Charybdis" bot who made some lings and hydras off one base and then just sat there. Even at the 15 minute mark he was still on one base. Obviously zerg is harder to program than the others due to larva mechanics, but I've never been challenged thoroughly by it.
That's how it is with Blizzard's AI I find even. While Blizzard's zerg isn't terrible, TvZ and PvZ are really easy vs AI Zerg, while ZvT and ZvP are A fair bit more difficult if the player is zerg.
On July 13 2011 15:19 alonestar_ak wrote:
@ ptanhkho: Could you give me some more detail how to public the map since i'm not experience with SC2 Editor. I would like to be your official uploaded on NA Server too. Otherwise, it will be many GT AI Maps on Bnet that make us confuse.
In my opinion an AI should be able to run builds on it's own, with just a bit of guidance... AI just needs to know basic openers like forge fast expand, 2 rax, an everything else should be almost entirely dependent on what it scouts, as opposed to what it was told to do. Having multiple long-term build orders programmed into an AI is just a guaranteed way to never attain good success rates, because it's like playing rock paper scissors with your eyes closed.
@Xapti: yes, that what currently GT AI does, you just need to guidance AI the opening step, about 12, 15 steps. Then the rest will automatically build the counter ( Banshee vs Immortal, Marauder vs Roach, etc.. ). Tested with the new GT AI you will see, first record your build, than after done your build, AI will automatically do a follow up strategy that counter enemy strategy. So basically the Custom Build will do better than the core AI, since it has dynamic strategy after the Opening, and AI know which strategy should used when fighting Protoss / Terran/ Zerg etc.
The important is some update change the whole strategy, like building requirement, disable, enable on some unit ( like the changing of building Supply first before build Barracks, remove the Infestor Ability). That would be very hard for AI to know and respond with these change.
On July 13 2011 15:19 alonestar_ak wrote:
@ ptanhkho: Could you give me some more detail how to public the map since i'm not experience with SC2 Editor. I would like to be your official uploaded on NA Server too. Otherwise, it will be many GT AI Maps on Bnet that make us confuse.
I uploaded the Green Tea AI 0.96 Map Packs that contain some 1 vs 1 Ladder Maps and included the AI Scripts in each map. So you just need to upload these on Bnet.
So, people on EU can play the map on their server and it has the AI built into it... but I'm from NA, so do I have to download the scripts and import them?
On July 15 2011 05:39 ClysmiC wrote: So, people on EU can play the map on their server and it has the AI built into it... but I'm from NA, so do I have to download the scripts and import them?
_ The Map Packages above contain maps that has AI build into it. So you can just download and no need to import the AI Scripts.
_ Also update the new version that change the strategy core. After the first strategy failure, instead switching to random commander, it will switch to general purpose commander that can counter to any strategy. But it need more testing and feedback.
A big issue I've noticed that I haven't yet addressed, is that when the AI is confronted with a superior army, it doesn't tend to back off if the opponent is near any base, and stream-in "reinforcements" constantly to suicide, even as far as warping-in units right in front of the enemy's army without any support.
This isn't a terribly huge issue, but it makes it so that once a player's won a fight at their base, it's completely over for the AI, even if the AI has a huge economic and production-capability lead.
The AI should have some sort of pacifist/defensive/turtle mode which it enters if the opponent arrives at a base with a superior-valued army, letting most of it's structures fall while building up units. It's not too likely to recover many games into victory, but I think it will help.
Another thing is dealing with destructible rocks. It is not too difficult to detect that rocks are blocking the closest access to an expansion, and destroy them if they are. It's somewhat difficult, but most of the work has been done already by others, if not yourself (choke&expansion detection).
3 really simple features I think many people would want: • Commander reveal command (obviously people will use this at their own discretion) • Select random commander after a reset, either via optional manual randomize commander command (kinda pointless IMO), or automatically after every reset (recommended) • Acknowledgment of all commands, such as nocheese(-nc)
Speaking of reset... it seems like it still doesn't work at all (restart), or that well (quick restart). Quick restart doesn't: • Remove creep • Reveal the map without units (resets fog of war) • Make AI behave properly, maybe because it relies on the game timer or something. It won't perform timing attacks that it's supposed to.
I'm also still left in the dark as to what the difference between quick restart and [normal] restart is, or is suppose to be.
_ A lot of Micro improvement : + Viking now only land ground when there are no air unit near by. + Hellion will stop running when being repaired. + Stalker will not hit and run again Bunker and focus on Armor unit. + When Infestor close to Die, it will generate the Infested Terran before dying.
_ Adding and improve some commander: + Changed Hera to faster 4 Warpgate build. + Adding 2 new Macro Style Terran Commander. + Fixing Zerg Commander
_ Improve the upgrade timing : + Adding more upgrade building ( EngineerBay, Armory, Forge ). + Upgrade timing faster. Expect +3 attacks, +3 defense units in late game. ( But AI will stop upgrade when low on unit number )
For GTAI 0.99, how exactly does it cheat? I know it gets instant upgrades sometimes such as blink/charge/marine shield etc. but does it cheat in any other way? I know it doesn't resource cheat.
I find myself playing against easy Ai all the time because that one does not have map vision, does not see my unit compositions (I play Zerg) and does not build Battlecruisers/Carriers all the time to exploit my "seeming" lack of antiair (I play ling/infestor style). Or when Terran decided to build mass banshee the whole game.
Is there a way to turn off map vision manually? Or finally forbid them to build units nobody builds in a serious game?!!?!
GT AI 0.99A: _ New select screen: using "-se" command, the selection screen will appear. Game Paused until close the selection screen. So take your time xD.
Terran Commander Screen.
Protoss Commander Screen
Zerg Commander Screen
Some option moved into GT Option panel. Open by "-op" command.
Now, I would just kill if it was possible to choose different difficulty settings after we start the game: 1. How smart is the AI 2. How much extra resources it gets 3. Whole map vision ON or OFF
Is this 0.99A version uploaded to EU server? How do I find it? entering GTAI into search comes will lots of useless results uploaded by different people whose names are not ptanhkhoa
New version GT 0.99B will reduce AI APM, and increase CPU Performance, as well as the Start Menu Screen. So you won't need to type any thing, so won't interrupt your opening.
The Start Menu will popup at the start but to view AI vs AI as observer mode, click view as one player and you will see the GT Start Menu:
Changelog: GT 0.99B: _ Drop FPS bug fixed. _ Instead of typing "-se". The Game will let you select 3 Game Mode at the beginning: + Play Random Commander + Play Selected Commander. + Play Last-Selected Commander
_ Option setting will be saved for next time loading. But you must be player 1 (ie Red Player) _ Show harvest rate now can be turn on off and in window mode. _ Mule now drop at the Mineral Position _ Creep Tumor will change direction when it get stuck (but if it too many object, it still can't get past).
_ Also Reduce difficulty of the Very Hard and Insane AI.
_ Very Hard : Bonus will increase from zero to 1.5 times after 10 minutes. _ Insane AI : AI get 1.5 Bonus at the start. (ie Blizzard Insane AI very hard AI )
It's a different army calculation system of Nexxy, which let AI know is its Army weaker or stronger than enemy (ie. 1 VoidRay >> 8 Roach ). But it consume more CPU than the current AVS ( Army Value System ).
On July 28 2011 02:31 LookNaph wrote: Are there any plans to limit the AI's APM (to, say 1000) and still do equally well in order to put less stress on the CPU?
^ I think you must put unit into group instead of individual moving, it will reduce the AI APM.
In a TvZ on XelNaga on version 1.0 I found a bug. I placed a marine at the 3 and 9 o'clock expansions to scout for when the computer would take it's 4th base. A drone was sent out and just kept twitching back and forth at the bottom of the ramp to the 3 o'clock expansion. Because of that the AI never took it's 4th expansion and stayed on 3 bases the entire game.
Can you pls be more specific on how exactly to find the newest version? And maybe the uploader name (EU-ladder)? I played a couple of games on 1.01 Version maps (they have french description from "Helmat") but the strategy-choosing-system seems broken on a couple of those maps
-qr sometimes bugs out after the second -qr or even earlier and the ai will just build workers and the minimum of production buildings. This has happened in past versions but thought I should mention it just in case it wasnt already known.
Hi, I noticed the 1.01 version hangs if you set the AIs to fight each other. Can you make the commander selection dialog appear even when there are no humans playing?
+ Change commander select type: Now observer can select the commander as well but you must view as one of players.
+ Reduce difficulty for Hard => Insane: All upgrade now required for AI and they progress like medium AI.
+ Fixing the BO record problem for Zerg Race ( Thanks to pi-sa )
+ Improve Zerg AI: => Modified counter unit of Zerg Race. => Zerg know when is the right time to massing drone and when it needed to stop product Drone for Defense or Build Army.
+ Edit/ Creating BO: => Not only you can record your BO, you can also edit the BO you just recorded for better timing. => You can also creating the new BO from the start and let AI do step by step. There are also Play and Pause button so you can paused AI to adding step to BO or let AI continue the BO.
+ Adding two random mode: => VS Random BO : Let AI pick up a random BO you created. => VS Random All General and BO : AI will pick up Randomly among the current AI Generals and your BOs.
New updates GT AI 1.03: _ Creep Tumor will stop spreading if there are enemy army nearby. _ High Templar stop using Hallucination on High Templar and Phoenix. _ Fixing some small bugs and improve AI performance.
ZvT the toss doesn't seem to wall off, played against the 4 gate one and since it is so easy to scout it is also ridiculous easy to stop. Then it tried to expand to the gold when it had no map control, seemed fairly wierd.
On July 16 2011 15:28 Xapti wrote: 3 really simple features I think many people would want: • Commander reveal command (obviously people will use this at their own discretion) • Select random commander after a reset, either via optional manual randomize commander command (kinda pointless IMO), or automatically after every reset (recommended) • Acknowledgment of all commands, such as nocheese(-nc)
Speaking of reset... it seems like it still doesn't work at all (restart), or that well (quick restart). Quick restart doesn't: • Remove creep • Reveal the map without units (resets fog of war) • Make AI behave properly, maybe because it relies on the game timer or something. It won't perform timing attacks that it's supposed to.
I'm also still left in the dark as to what the difference between quick restart and [normal] restart is, or is suppose to be.
Still don't seem to be addressed. Also, what even happened to the no cheese command? I didn't see it in the new system.
Another little tidbit: Playing protoss recently, I noticed they do this a lot but never really realized how big of a problem it is: AI wastes their guardian shields quite easily. He had only like 3 sentries when I attacked (and a small force of other units), but activates guardian shield for ALL THREE of them. As opposed to having to activate a new guardian shield if one (or some) units aren't in the main guardian shield (which is what I resume is being done now), it should just try to keep the units inside the current field instead (unless the army is too big, then more shields make sense. This army was only like 8-9 units though!)
Playing around with it one afternoon, I noticed vs Zerg the AI of any race would focus fire the cocoons next to a hatchery, instead of the hatchery or the drones. Pretty puzzled why do this, because cocoons have probably the largest armor in the game and it takes forever to kill one. That makes the offensive AI vs Zerg a lot weaker than it could be, as long as the zerg keeps making units.
On August 13 2011 14:38 tarodotoxin wrote: is there any option to take away AI mineral cheating? (great maps btw)
Yes, that depends on the difficulty setting:
+ Very Easy AI: AI won't attack in the first 10 minutes. + Easy : GT AI ( normal vision like player) + Medium : GT AI ( full vision ) + Hard : Better upgrade than Medium AI. + Very Hard : Same harvest rate at the player at the beginning, then increase from 1 to 1.5 times in 15 minutes. + Insane : 1.5 bonus at the start. (ie Blizzard Insane AI harvest rate ).
On Hard there's no mineral cheating for the AI. On Easy there's no maphacking for the AI.
Changed to GTBOP to filter the old map version. It new version now will started with GTBOP.
Changelog in GTBOP AI 1.07: - AI won't attack Larva or Egg any more unless there are nothing else to attack. - GT Reminder don't show at the start, so you can choose the commander at usually. - Two new command in observer mode: + "-r" : Quick random match. + "-s" : Quick commander select. - New features: Build Order Practice: + Let player practice to perfect their Build Order they have just created. + This mode will appear the "AI Attack" button, that can let you practice build order in peaceful, or let the AI attack at usually.
Is it possible to let the AI send out the probe before it has the right amount of minerals so that it builds build the pylon immediatly when it is possible?
When you ran out of ideas to improve the AI, check out the "1000 tips" thread. I think there are plenty of improvements like upgrade orders (for protoss it should do armor first when the AI goes for zealot heavy compositions; moreover, when it has two forges it should upgrade shields lvl 1 instead of armor lvl 3 when it is on 2/2, so it can upgrade to shield lvl 2 and armor lvl 3 at the same time after that.)
BTW, I did notice some problems (which are probably a relatively easy fix. IDK):
- T or P never seem to wall off, even when the box is checked. - Some builds just seem to not exist. For example, I tried the 4gate and the 7rr general. The 4gate DID make a 4gate, but then he just 1based with it and never pushed out. Meanwhile, the 7rr didn't build a pool in 20 minutes.
These might just be minor glitches for me only, but I felt I should share. This AI is still brilliant though =D
@GoldTalon : - About 4 gate and 7rr general, it depend on your base location, if AI cannot find the right placement, it will get stuck. Try different map, or restart the match.
New version 1.09. You can search the official version of GT AI on EU Bnet : Key word: "GTAI Public Release" Publisher : Nex 144
* added a new Army Value without Structure Defence values * complete new Roach AI, I stopped Roach-burrow-dancing * changed some Medivac behaviors * added first steps for Anti-Drop-Detection-System * fixed a TriggerError caused by Pylons in very very long games * removed some bad manners right after the first attack (HAHAHA I WIN) * for AI vs AI tester : new command at the first 10 sec : "-go" * Toss should build only 2 Forges now * new Pylon and Structure placement for Protoss to remove the Stalker and Zealot trap * Performance changes in AVS * Enemyarmy value calculation in AVS has changed a little * reduced own-AVS Strength for Protoss to avoid early attacks * changes in AVS to detect ArmyValue without Structure Defence values * limit AVS calc to 1 per sec * Overworked the 4-Gate Commander Witness IV to become more deadly * some minor performance changes * some minor AI micro changes
On August 23 2011 05:31 BECOKA wrote: Looking forward to 1.09, how can we get it uploaded to the NA servers?
this is turning out to be a fantastic practice partner.
You can try "GT AI" and look for maps by IcyTournesol. I'll keep them constantly updated without name changes. Humdinger also have a bunch uploaded, but they're currently outdated last time I checked.
On August 23 2011 05:31 BECOKA wrote: Looking forward to 1.09, how can we get it uploaded to the NA servers?
this is turning out to be a fantastic practice partner.
You can try "GT AI" and look for maps by IcyTournesol. I'll keep them constantly updated without name changes. Humdinger also have a bunch uploaded, but they're currently outdated last time I checked.
How do I know if an IcyTournesol map has 1.09 or an older version? Or a newer version later for that matter?
I'll download them in hopes they are all 1.09 for now.
On August 23 2011 05:31 BECOKA wrote: Looking forward to 1.09, how can we get it uploaded to the NA servers?
this is turning out to be a fantastic practice partner.
You can try "GT AI" and look for maps by IcyTournesol. I'll keep them constantly updated without name changes. Humdinger also have a bunch uploaded, but they're currently outdated last time I checked.
How do I know if an IcyTournesol map has 1.09 or an older version? Or a newer version later for that matter?
I'll download them in hopes they are all 1.09 for now.
Note: looking at them and the titles all have v0.8 what is that related to?
Edited: Silly me, it says when you start the map what version of gtai it is.
new Note: it would be much easier if the title of the map said what version of GTAI was used.
On August 23 2011 05:31 BECOKA wrote: Looking forward to 1.09, how can we get it uploaded to the NA servers?
this is turning out to be a fantastic practice partner.
You can try "GT AI" and look for maps by IcyTournesol. I'll keep them constantly updated without name changes. Humdinger also have a bunch uploaded, but they're currently outdated last time I checked.
Thank you so much for the updated maps on bnet =)))
after 400 smart ass posts on TL.Net i think i'll actually contribute something and try to place the 1.09 versions of GTAI of some major maps onto the North American Server.
I'm starting with MLG Shattered Temple
alright MLG Shattered Temple with the 1.09 Green Tea AI is now operational on the North American Server people.
* Fixed a very critical bug in AI-building functions * changed some AVS calc again, to be more precise * new Chrono system for Protoss .. up to 10times better performance * new Commanderconfig. Each commander has now its own config. Like Chrono priority lists autoupgrading etc. * some performance things in the deep core * new anti-drop detection system + avoidance * AI should fight now better with its own def-tower * again changed some Pylon placement functions * some minor changes
Everytime i play as protoss against T there is some sort of massive lag as the game progress (FPS lag) it keeps dropping the fps until it reaches 0 fps. Just happens when i play against Hard AI Terran, started happening on v1.0+. It's not a problem of computer peformance as the sc2.exe is just cionsuming 30-40% of overall RAM and CPU Usage when that happens.
Just wondering if anyone is having a similar problem or it's just something on my end?
P.S.: I can play against P or Z fine and no fps lag ever happens just happens when playing against T. (Already tried to change the APM limit and nothing happens)
On August 31 2011 23:25 ptanhkhoa wrote: Strange, since AVS is caculation for all race, if it happen, it will occur in every race. I will have a check on that, thx for report.
With all the other 2 races (playing as protoss) it doesnt happen any kind of lag at all, and i don't turn APM limit. Not even if zerg goes extremely heavy on zergling. As terran the lag usually starts to become unbearable when the AI is taking its third expo. Not sure if have anything to do with the 3rd expo or the fact that everytime it lags the Terran has Thors or Banshee+BC (I choose Random strat and since i go heavy on chargelots on early game the AI usually transition into air more often than not against me).
It doesnt need to happen an engagement at all, when im around 120-130+ food usually the games drops down to 1 fps or 0, when the engagement happens it becomes worst. i already capped the APM to 200, doesnt change a thing. I tried the ai on more than one different PC but it always end up the same against terran AI.
I wouldn't stress out too much tho, since looks like im the only one who reported that problem it could be some problem on my end, some program i have on all pcs or something similar. Thanks for the help either way.
On August 13 2011 20:30 ptanhkhoa wrote: - AI won't attack Larva or Egg any more unless there are nothing else to attack.
Sweet, that's great, thank you!
Is there a way to make the GTAI replays parsable with sc2gears? The replay can be played from SC2, but not analyzed with sc2gears as any other replay. I don't know if the problem is in the way sc2gears parses replays, or the way GTAI produces replays.
love the AI, but is it possible to add more conventional builds you see on ladder, like for example a 6/7 gate timing attack? i find a lot of those default commanders are very gimmicky builds, like for example i just played a 1 base stargate build where the protoss banked 1.5k minerals and pressured me once in a 20 minute game.
I'm really amazed at all the clever stuff your AI can do, well done I am struggling a bit with the UI so I haven't been able to do too much with it, so I'll check back after a few updates.
I'm finding something a little whoozy with the AI. I'm using version 1.13 and I play Terran mostly against Protoss and Zerg on Hard 1v1. Since I like the AI using different strategies each time, I set random generals to play them.
What I've found out recently is that despite having the generals set on random, when up against Protoss, I always end up playing against a 1 basing no gas AI that just masses up zealots, even when it's a 4 player sized map like Shakuras Plateau, and they all charge initially for my bunker, but when I bring SCVs out to repair the bunker, the zealots suddenly chase only the SCVs (and can never catch them because of acceleration speed difference). So basically I just run my SCVs in a circle forever until all the zealots are dead. Then it does absolutely nothing but massing zealots in its single base until I run up there and destroy it.
Against Zerg, it's always a commander that fast expands with no gas and just masses up zerglings off of 2 base and runs them into my bunker depot wall-off to their deaths.
What's worse is that the AI can't seem to do anything else, like the AI transitioning into another build (which they do on earlier GTAI versions) after a specific build order around 6 minutes.
Does the AI APM limit have something to do with the AI's units being left unattended at times? Or is there another issue, like say, with regard to randomizing generals?
anybody have the best way to search for the newest versions on the NA ladder?
I've been having success with "GTAI_107" but it looks like we've updated well past that version.
Great AI, really fun to practice builds against. Only see wonky behavior every now and again where its like the AI just decides to not build anything but harvesters. Usually happens after a few "-qr" restarts.
Edit: Oops. Just saw "IcyTournesol" post. As long as those versions stay updated, they seem easy enough to find.
On September 13 2011 23:57 ptanhkhoa wrote: @Monkeyballs25: Which problem did you have with the UI ? I can help u .
I didn't understand the manual build order option at all. And it didn't copy my builds properly either. And there doesn't seem to be a way to get the start menu back up once the game begins, even after a reset.
On September 14 2011 00:31 KanoCoke wrote: I'm finding something a little whoozy with the AI. I'm using version 1.13 and I play Terran mostly against Protoss and Zerg on Hard 1v1. Since I like the AI using different strategies each time, I set random generals to play them.
What I've found out recently is that despite having the generals set on random, when up against Protoss, I always end up playing against a 1 basing no gas AI that just masses up zealots, even when it's a 4 player sized map like Shakuras Plateau, and they all charge initially for my bunker, but when I bring SCVs out to repair the bunker, the zealots suddenly chase only the SCVs (and can never catch them because of acceleration speed difference). So basically I just run my SCVs in a circle forever until all the zealots are dead. Then it does absolutely nothing but massing zealots in its single base until I run up there and destroy it.
Against Zerg, it's always a commander that fast expands with no gas and just masses up zerglings off of 2 base and runs them into my bunker depot wall-off to their deaths.
What's worse is that the AI can't seem to do anything else, like the AI transitioning into another build (which they do on earlier GTAI versions) after a specific build order around 6 minutes.
Does the AI APM limit have something to do with the AI's units being left unattended at times? Or is there another issue, like say, with regard to randomizing generals?
Yeah, in my opinion the AI itself still needs much improvement, as opposed to just adding or fixing features (maybe with the exception of reset, which I don't know why it's so hard, not to say that it isn't).
Sure, people shouldn't be trying to exploit the AI, but that doesn't mean that the AI should get away with being stupid.
The AI doesn't work with the latest patch. Since all the match-ups I've been running into are TvT nowadays, I wanted more practice against mech/sky/MMM builds, but I guess there won't be any of that any time soon. :<
I only wish Zerg AI would actually attack once in a while, when I push marine tank into it. It just sits in the main base while I kill its 2nd and 3rd base.
I used to play macro or die and that had GTAI 0.86 I think, it seems a lot better than the newest version which I just tried, the comp just tried to cannon me and after i broke through with tanks it kinf of gave up...
Is there anyway to micro only and I don't have to macro? I mean just jump straight in the battle. I just wanna test the new composition only and my micro skill.
On September 27 2011 14:43 Tatari wrote: The AI is really CPU intensive in big battles, and my game ends up lagging. It's never happened before this patch. Is there any way to fix this?
If the lag has never happened to you before, then the patch 1.4 is most likely the culprit here. The performance issue is listed in one of the known issues by Blizzard - http://us.battle.net/sc2/en/forum/topic/2267599524.
EDIT: I suppose you can try limiting the AI's APM in the option menu.
I like the concept and principle of this AI, however I have a problem with it, one small fault that keeps me from getting it.
I don't like the fact that your AI gets map vision and extra resources at higher difficulty levels it kind of detracts from the already impressive accuracy it has to mimicking real players because of the diversity of strategy it uses.
It would be nicer if at higher levels of difficulty the AI uses either harder to execute strategies and/or executes them better while also increasing the APM it does and also learns finer micro, like constantly kitting zealots with marine/marauder till the zealots are dead then kill stalkers, or if it does efficient drops or poking with vikings etc.
I would love if you could develop and AI that works like that, doesn't cheat at higher levels but just plays better overall, would feel like Insane is GM/Master, Very hard is Master/Diamond, hard is Diamond/Plat.
You have done it. You have successfully found a place where ladder fearing noobs can retreat and practice just as well, without losing any ladder points! i thank you for this
On October 02 2011 08:05 firehand101 wrote: You have done it. You have successfully found a place where ladder fearing noobs can retreat and practice just as well, without losing any ladder points! i thank you for this
:D. I realize you're joking...but joking aside, playing ladder leads to dealing with really unpredictable strategies (at lead it does in lowly diamond). For me, practice is most effective when I can try the same thing over and over again. GT AI allows me to do this so I can then ladder and deal with the unpredictable while having my chops down.
Thanks for the map, very well done. I use the method 1 - download and play from EU custom game I have a question -
1. Is this updated for version 1.4 - Xelnaga/Shattered Temple / Metal? 2. Is this only vs Terran I can play? 3. I want the AI to learn a BO that I face vs terran - marine opening into BF harass into marine tank timing - how do I get the AI to do this?
On October 02 2011 07:59 Destructicon wrote: I like the concept and principle of this AI, however I have a problem with it, one small fault that keeps me from getting it.
I don't like the fact that your AI gets map vision and extra resources at higher difficulty levels it kind of detracts from the already impressive accuracy it has to mimicking real players because of the diversity of strategy it uses.
It would be nicer if at higher levels of difficulty the AI uses either harder to execute strategies and/or executes them better while also increasing the APM it does and also learns finer micro, like constantly kitting zealots with marine/marauder till the zealots are dead then kill stalkers, or if it does efficient drops or poking with vikings etc.
I would love if you could develop and AI that works like that, doesn't cheat at higher levels but just plays better overall, would feel like Insane is GM/Master, Very hard is Master/Diamond, hard is Diamond/Plat.
Great idea. I find it obnoxious that the AI sees all my shenanigans from the get-go and hard counters the crap out of me in Hard~Insane on the mid to late-game scenario, mainly because it didn't even have to do any scouting at all, while I struggle all around just to get info. There should be an option in the options tab for you to enable or disable the AI's map hack.
The Zerg AI also uses too much APM maneuvering each individual overlord to safety when an attack comes to its' main that uses up so much memory due to it going at 1000 APM, which is not only physically impossible for a human player to do, but is also impractical. Toning down the AI's max APM via the options tab also suddenly makes the AI dumber for some reason, often times with the AI not being able to do its' own build orders because it needs at least 600 APM minimum before their actions register. Not to mention, blocking an expand via a probe or SCV results in the AI one or two-basing for a good minute.
All in all, the AI still kicks my butt, but sometimes only due to the map being fully revealed to it, instead of a really good army composition or constant effective harasses.
On October 02 2011 07:59 Destructicon wrote: I like the concept and principle of this AI, however I have a problem with it, one small fault that keeps me from getting it.
I don't like the fact that your AI gets map vision and extra resources at higher difficulty levels it kind of detracts from the already impressive accuracy it has to mimicking real players because of the diversity of strategy it uses.
It would be nicer if at higher levels of difficulty the AI uses either harder to execute strategies and/or executes them better while also increasing the APM it does and also learns finer micro, like constantly kitting zealots with marine/marauder till the zealots are dead then kill stalkers, or if it does efficient drops or poking with vikings etc.
I would love if you could develop and AI that works like that, doesn't cheat at higher levels but just plays better overall, would feel like Insane is GM/Master, Very hard is Master/Diamond, hard is Diamond/Plat.
A laudable goal.
There's a reason AIs don't "play fair," which is that is very hard to make an AI that can out-think a person. This isn't just true for green tea, but for any computer-based AI. In fact, as I'm sure you know, there's an entire field of Computer Science dedicated to trying.
I'm sure everybody would like to have a fair-playing, GM AI waiting to play them at any time. I also find the "ohh i see your observer, guess I'll just park my observer on it forever" types of gameplay frustrating since they don't mimic reality, but the truth is it's just the way things have to be to make the computer a challenge for us, at least for now.
If you're really that bothered, pick up the torch and consider a career in computer AI research.
I just wanted to say thank you to those that work on this. I lab this thing about 20 games a day and usually still lose a few. My Masters friends think its terrible and I should ladder, while my Gold and lower friends thinking I'm like a pro. This keeps me SORT of sharp (at least mechanically) with none of the stress of laddering. TYVM
All your work is appreciated, and GL making this thing even more of a pain in the ass <3
edit: Since its the topic of the moment- "cheating" and unit countering are fine with me, it's the human equivalent of scouting- since the AI really cant think. You can still out smart the hell out of it. I spent a LOT of time making AI's with WC3... which had a great interface for it. Once you've spent hours with them, you realize there are serious limitations to what they can and can't do.
While I do understand that something like, unit preservation and effective harass, is something very hard to code, I didn't think it was that hard to make an AI just incrementally get more APM/micro per difficulty and also broaden the range of strategies available to it.
I'm not asking much, just for the AI to not cheat.
I disagree with the people saying there should be a no-vision AI at hardest difficulties.
Starcraft is a lot about predicting your opponent, getting a solid idea of what they're getting without seeing much, and even just guessing and/or unintentional lucky events.
When practicing and/or facing an AI that is (at least currently) weaker than masters level players, or even diamond, it needs all the luck it can get. Having the AI know what you have and know where you are teach players that they need to be prepared for their opponent (prepared for the worst-case scenarios), and not rely on lucky counter-attacks or lucky harassment.
However, I think that currently the AI doesn't even counter what your getting that well at the moment (not sure haven't tested too much and/or don't remember testing I did do). Things like getting bunkers and marauders vs a roach rush, turrets or cannons or observers for Dark Templar or banshees, thors or phoenix for mutalisk, etc.. In fact, that's something I think really needs to be worked on for Green Tea AI that could make it a whole lot better — following a BO, but before even finishing the BO, modifying it a bit (such as detection), or switching to something completely different (like double stargate instead of 2 gate expand) to counter the opposing player's strategy.
Mineral cheating I don't like though, since it's a completely different problem. That said, I could care less if mineral cheating is in ONE difficulty (preferably with player-adjustable income multipliers) such as insane, which is what Blizzard does.
While in an ideal world it would be nice to have an option to be able to disable CPU vision and have it play well, but the problem is that disabling revealed map for the AI results in it being terrible at countering absolutely anything, and scouting is an extremely difficult thing to program effectively into an AI. Humans know generally where the best expansions are to scan, or where they might be hiding tech, or when to gas steal, or all sorts of other things, but without specific programmed guidance, the AI cannot do that. Overall, it makes an unnecessarily easier AI at the expense of a lot more programmer work, which is not worth it until the AI gets good enough to beat most master players.
And if any of you did not know: To play a non-mineral cheating AI, Hard is the best one — the AI itself doesn't improve past hard. To play a non-map-cheating AI, Easy is the best one.
Until today, I had been playing a GTAI that worked fine on 1.4
The ones uploaded by Humdinger (GTAI 104a Melee, Co-Op, vsAI & Custom) maps were working just fine. In fact, they are FAR more challenging on Medium than the Current ones by IcyTournesol.
The new ones are actually nicer, and I like the Aggressive vs Defensive mode options. Very nice. I had been playing the 104a w/o the New Strength Analysis on (it slowed my computer down WAY too much on like 4 bases each)... but still provides a much better challenge than the current ones recommended.
I don't know why this is... but just wanted to give some feedback while this thread is hot. The only other thing I would say is maybe prioritize the Rocks on maps a little higher, especially on XelNaga and Shattered Temple. The thing seems to trip on the rocks quite a bit. Lastly, maybe assigning Map Specific Zones to the AI to build structures in for the Cannon rush and maybe even like a Proxy Rax. ie. Behind the mineral lines of naturals or in smart zones in the main. I LOVE that the AI does these things, and get a giggle every time.
Is it possible to improve some of the building placements? like for example full walloffs with FFE in PvZ or depot-rax-depot walls in TvZ. of course I preface this with saying I know nothing about coding SC2 custom maps, but i shouldn't be able to run lings past the choke points so easily.
It is possible, but it's difficult and takes a lot of work to do. Several months ago I heard of a guy that managed to code an algorithm to make walls at chokes with just depots or something (and I think it still had some problems), but not mixed buildings like barracks, which is more difficult. I haven't heard from him since, I don't know if he's still working on it, but I don't think so. You should just pretend the computer walls off at x time though, or enable the "kill units when they go up ramp" feature the maps have, which I think is super dumb.
Now that season 4 maps are confirmed, can IcyTournesol please put Metalopolis (w/o close spawns) back into the rotation of GTAI maps that are kept regularly updated?
The AIs don't even seem to do what they say they do.
A build labeled tank did like a 7 marine +1 tank attack, and then got marauders, marines, medivacs, and some vikings and hellions(along with a few tanks) — how is that build tank? I would consider a tank build to be marine-tank, which I suppose he did, but it was only 1 tank and like 7 marines before getting other stuff.
There was no reason to be getting vikings or marauders or medivacs either, since I just massed mostly zerglings with some banelings. If it switched to automatic play, shouldn't it be able to deal with just 1 or 2 units better than it did?
On July 15 2011 21:47 ptanhkhoa wrote: _ Also update the new version that change the strategy core. After the first strategy failure, instead switching to random commander, it will switch to general purpose commander that can counter to any strategy. But it need more testing and feedback.
Is that why it did that? Getting a bit of every unit is a terrible idea until/unless it's late game like 20 minutes deep. It's definitely not the way to attempt to counter any strategy. Is that sort of behavior still around? it seems like it.
I also selected a 14 hatch 14 pool commander, but [the 3rd game] it did a 7 pool instead — is that intentional?
Hi, I've been messing around with the 1.17 version on the EU ladder and its much improved!
The two big bugs that stand out are: - The AI never breaks the rocks down for expansions and faster pathing between bases, sometimes it expands next to me instead of taking the safer gold.
- Workers seem to glitch out when an enemy unit comes within a certain proximity and get stuck. E.g. leaving a marine at a watchtower caused 30 drones to glitch while trying to transfer, or putting a unit behind the mineral lines prevents all mining.
Great map where I've practiced on alot! Too bad there are a few bugged maps on the server now, especially the 'b' or 'c' versions (example Xel'naga GTAI 1.14b) where there are no larva's for zerg or just no option to select cpu's buildorder.
I'll try out the new 1.17 version of the map. Still, great job on this map and it really helped alot! I know a Terran player who was bronze (with a little ladder anxiousness), began to practice alot with the GTAI and when he started laddering again he instantly climbed to Platinum. Keep it up!
Tried it for the first time today, very impressed. I tried a new build as Zerg vs Very Hard normal AI toss and just completely owned, I didn't even have to think, scout or react, just macro according to the build. Tried vs medium GTAI, a commander using a ground force, and I actually had to work for it, he attacked way faster than the very hard AI, at a time where my build really didn't have any defences.
To have an AI which can do realistic builds in a decent manner? Now that's nice.
I wanted to give a try to the AI's 4 gate in PvP on Tal'D'arim, but I must say I'm a bit perplexed at what I saw. When I scouted him, I saw he took 2 gas and didn't reserve any chronos for warp. How is that a 4 gate ?
On October 02 2011 07:59 Destructicon wrote: I like the concept and principle of this AI, however I have a problem with it, one small fault that keeps me from getting it.
I don't like the fact that your AI gets map vision and extra resources at higher difficulty levels it kind of detracts from the already impressive accuracy it has to mimicking real players because of the diversity of strategy it uses.
It would be nicer if at higher levels of difficulty the AI uses either harder to execute strategies and/or executes them better while also increasing the APM it does and also learns finer micro, like constantly kitting zealots with marine/marauder till the zealots are dead then kill stalkers, or if it does efficient drops or poking with vikings etc.
I would love if you could develop and AI that works like that, doesn't cheat at higher levels but just plays better overall, would feel like Insane is GM/Master, Very hard is Master/Diamond, hard is Diamond/Plat.
A laudable goal.
There's a reason AIs don't "play fair," which is that is very hard to make an AI that can out-think a person. This isn't just true for green tea, but for any computer-based AI. In fact, as I'm sure you know, there's an entire field of Computer Science dedicated to trying.
I'm sure everybody would like to have a fair-playing, GM AI waiting to play them at any time. I also find the "ohh i see your observer, guess I'll just park my observer on it forever" types of gameplay frustrating since they don't mimic reality, but the truth is it's just the way things have to be to make the computer a challenge for us, at least for now.
If you're really that bothered, pick up the torch and consider a career in computer AI research.
One workaround could be to continually update what the AI does whenever there's a shift in metagame.
i just tried some GT maps and the AI is pretty good but i noticed last game i played into a longer macro game the game was very choppy and during battle unplayable
I've noticed some problems with the AI. I wanted to see how much gasstealing helps against a protoss who wants to go stargate or DT tech, but gasstealing completely breaks the AI and it starts to expand like crazy. I did it against a terran as well and the terran had 3 bases by the 10 minute mark and was building another CC next to his third base.
Man, you have no idea how much I use this. It's important to find a map that was properly created, as there are some people who have published non-working versions of this.
I've amassed around 500 games (far more than I play on ladder) working on my macro, as this AI is able to macro like a beast if you do not abuse it in certain ways. The best advice I can give is just that, to strictly avoid using any abusive technique that messes with the AI.
There was an AI called Fyn AI that used real scouting patterns and no map vision. I'd really like to see a way to use Medium or Hard and *not* give map vision to the AI. It can often times build a perfect counter to units that you are keeping well-hidden.
According to SC2Gears, I've played 680 matches against GTAI and can definitely say it macros as well as any Diamond player, and has incredible micro at times. At other times, the micro can be somewhat stupid but you can usually "play nice" to let it recover from any bad mistakes. It's important if you're playing at a Plat/Diamond+ level to allow the AI some breathing room and challenge you.
As someone who has had a very difficult time talking myself into laddering, I've shot up the ranks using GTAI for 95+% of my practice and would love to know if I can help or offer any feedback.
I've found that there is a threshold where if your macro is good enough, and your army is strong enough, the AI will simply never attack. However, this teaches some valuable lessons in of itself. If my macro slips, I deserve to be pummelled to death by a preponderance of units.
Zergling runby functionality would be excellent (or any runby functionality). Since the AI seems to attack move from the front of your base to the back, there is currently no good way to be punished for having a poor wall. An example is a 6 pool or 12 drone rush, where the computer will attack a half-built barracks to death before attacking your workers, ruining the attack.
Thanks so much, and I really hope to see further revisions of this mighty macro tool. I can't thank you enough for giving me an outlet to improve my skills vastly (I have a career in BW which helps) by giving me limitless ladderless games with a more than willing opponent.
Do not ever gas steal or build structures in the opponents base as it will draw the ire of their workers and put the AI far behind in macro, never to catch up again. This AI is capable of beating Diamond level players if done right, and I can't recommend it enough.
I love the GreenTea AI with Macro or Die. Haven't been able to find it on NA server. Glad it's back. But this might discourage me more from laddering practice lol.
The new ladder maps can be learned safely by noobs, just type 'Entombed GTAI' or 'Plateau GTAI' into the search bar, and you'll find them under Icytournesol as usual.
Yeah when it goes all Flash 1500 apm crazy my PC is like WTF-> 0,1 fps, kinda hard to pull off any storms :/ P.S. Of course the natural solution is to activate the APM limit. For my PC about 600 makes it playable. It does impare it's micro+macro ability during a maxed battle, but it is unrealistic otherwise anyway.. P.S. #2 Nope, even the minimum of 200 apm still seems to overwhelm my system sometimes.
I've been using it a lot using SC2Allin1 and it has been great... however, I changed to Darkgrid hotkeys yesterday, and since SC2Allin1 doesn't work with custom hotkeys (you need to edit gamehotkeys.txt which isn't compatible with the formats used today) it has gone from awesome to useless
I just wish there was a better custom map system so one could find the real deal... on EU, Nex has only published a very small amount of maps, and I don't want to use other publishers since they aren't official so I can't know what version of GTAI is used etc.
PvT against very hard really makes me wonder why ladder terrans do the passive 1rax expo instead of pressuring... Seriously, I have yet to find a way with which I can punish his 3rax pressure and get an expand at a similar timing with him.
Despite the lag I really like it though! Best ai out there.
On January 06 2012 04:41 Thobrik wrote: PvT against very hard really makes me wonder why ladder terrans do the passive 1rax expo instead of pressuring... Seriously, I have yet to find a way with which I can punish his 3rax pressure and get an expand at a similar timing with him.
Despite the lag I really like it though! Best ai out there.
Keep in mind that everything greater than medium has increased mineral income over time.
On January 06 2012 04:41 Thobrik wrote: PvT against very hard really makes me wonder why ladder terrans do the passive 1rax expo instead of pressuring... Seriously, I have yet to find a way with which I can punish his 3rax pressure and get an expand at a similar timing with him.
Despite the lag I really like it though! Best ai out there.
Keep in mind that everything greater than medium has increased mineral income over time.
What do you mean by over time? I checked the replay and it seemed to get 5 per worker...
On January 06 2012 04:41 Thobrik wrote: PvT against very hard really makes me wonder why ladder terrans do the passive 1rax expo instead of pressuring... Seriously, I have yet to find a way with which I can punish his 3rax pressure and get an expand at a similar timing with him.
Despite the lag I really like it though! Best ai out there.
Keep in mind that everything greater than medium has increased mineral income over time.
What do you mean by over time? I checked the replay and it seemed to get 5 per worker...
From the op
4/ Is this AI cheat ? - Depend on AI level: the first 4 level doesn't using resource cheat. + Very Easy : It will stop attacking in 10 minutes. + Easy : No map vision. + Medium, Hard : AI has full map vision. + Very Hard: Only cheat resources in later game. Limit is 1.5 times normal resources. + Insane : 1.5 times resources at the start.
---
On January 07 2012 16:00 Hopelessnoob wrote: how are you guys playing this now? I used to get the one on bnet but it does not work now . sc2allin1 cannot be downloaded
Mine works beautifully. Make sure you don't have it on Very Easy, or its in passive mode for like 10 minutes.
Just a note to the creators, the Aggressive mode On/Off is a great idea, but KNOWING which it's going to do isn't. This should be a random setting when you start.
For example, I know Terran is going to send marines out early if I set it to aggressive. So I make an extra Spine and it just suicides into it. However, if I leave it off and just let it push when the build orders specify, I know I can get away with minimal defense and just leave a Ling out front of it's base so I know when to start making lings.
Making this a random setting, would be equivalent to a player that you don't know is going to 2 rax pressure you or 1 rax expand. You have to scout and be aware of what it's doing, rather than knowing in advance. I would just remove the setting from the interface selector, and have it be randomized. OR, you could have 3 settings, on/off/random... for the aggressive state.
have tried GT AI 1.17 Nex on Hard in B.net and it didnt work... AI just did nothing
build 4 gateways and after that stopped doing anything but producing 400 workers ...
Tried some GT AI 0.86 by someone else and it worked kinda. However i got owned pretty hard midgame and was surprised how this could happen since i had it on "Hard" since OP says no cheating at Hard. But at least this version after 12 minutes it had spend more ressources than me by about 5500 minerals I mean yeha my macro is not that good but lol 5500 minerals less mined after 12 minutes ?
Even for my bad macro this isnt possible without AI cheating.
goonna try that +gtai +17c
edit :
Player GT AI 17c from Steeno (or something) and it worked and AI didnt mineral cheat on Hard At minute 12 it had somehow spend 450 ( not 5500 like that stupid map last time that did massivly cheat) minerals more than me overall (no probe cutting i guess)
However it wasnt really so much better than Blizard AI on very hard. Just a tiny bit better but not much.
On February 04 2012 23:59 DontLoseSightOfIt wrote: How do I get this? I am in SEA, is it available in SEA?
Or is there a website that shows all instructions? GT AI is basically AI that gives more challenge than Bnet AI right?
Everyone on SEA has a free NA account. You should activate it.
Oh yes I forgot that.
So, after that, what to do? Is there a list of instructions? Sorry guys, I just have my internet capped, so I cannot even go to the main website to check.
Would someone be willing to list out the steps here? Would love to play GT AI
EDIT: Ok, so far I have...the SCallinone thing. I have no idea what this is but I did my best. I am up to the part where I have to add maps. I am lost here, I cannot open the normal BNet maps..
The AI is in a class of its own as long as you don’t break it with funny builds. For standard play it really does make you a well rounded player, because the AI can macro like an animal and can micro almost as good.
Though, I too sometimes have anxiety about playing ladders and so all I do is play GTAI. I can macro pretty damn well (70-80 average APM without a ton of spam) and can micro decently. I can beat the AI on hard consistently (as long as I’m not goofing off) and can beat it on very hard occasionally, but yet, I’m a gold level player. Why? Because GT as great as it is…is nothing like a real opponent. Real opponents have wonky builds and wonky play styles that you just don’t expect. There really is no replacement for ladder play, but it at least gives you strong fundamentals to fall back on.
I'm sorry if this has already been asked: Is this AI less smart (or have lesser macro or micro) on Easy than on Very Hard? Or are the only differences between the difficulties the ones listed in the OP (map hacking, additional resources, etc)?
If the Very Hard AI is actually smarter and plays better, is there any way to play against that AI without having it cheat?
On April 14 2012 11:22 Empirimancer wrote: I'm sorry if this has already been asked: Is this AI less smart (or have lesser macro or micro) on Easy than on Very Hard? Or are the only differences between the difficulties the ones listed in the OP (map hacking, additional resources, etc)?
If the Very Hard AI is actually smarter and plays better, is there any way to play against that AI without having it cheat?
I would assume that Very Hard isnt smarter than Hard. You don't want an AI to cheat if there is an other way. I also find it quite strange to practise against the Very Hard AI. Especially psychologically if you sit comfortably on 3 base vs 2, but he produces just as much.
On May 02 2012 08:19 Smackzilla wrote: Is there anything that really describes the commanders? The in-game descriptions are a little terse.
In short, I want a commander that does a 1-1-1 fast siege push to practice my gasless FE in TvT. Is there any AI close to that in green tea?
Practice partners I would do it for you, but finals...
Honestly there are so many variations to the 1-1-1 that even if there was a program that could execute a build against you I'm not sure how helpful it would be.
I love the layout of the map, and would love to just lab some games on it with the AI.
Also, as far as how the AI are performing, the 1.17 is good, but does some weird stuff (never clears rocks, attacks its own units sometimes). I find the most challenging version of this AI to be unique to this singular map...
...the "GTAI 1.04a iCCup Testbug" by Dvalen (on NA) seems to be leagues ahead of all the other AI. I don't know what makes it so, but it surely is. I play with ALL the check boxes off, except for the APM limiter on and turned down almost all the way (since the new APM counter is retarded). I play it on Hard difficulty.
Thanks to all the people who have contributed to this AI
On May 28 2012 16:39 Laserist wrote: I wonder if it is possible to turn that AI into a script and throw it into the ladder. It would be interesting to see how it will ranked in that case.
I would guess gold, maybe silver, on the highest level w/o income cheating at all (hard). It might even score platinum if it were lucky, but the ladder is a lot more difficult that it was a year ago.
This would be given that people thought they were playing a person... so they played and scouted as they would a person without trying to exploit it.
I'm sure it would be apparent after a while that the opponent was throwing away some units strangely, and not harassing at all, and not doing basic micro techniques... but I'm sure everyone has a bit of that anyways. It does have a good variety of builds though, when you play it on random. Biggest weakness is that the Protoss AI doesn't have a FFE build, but I'm sure its early Zealot Immortal push would at least keep it in silver.
On May 28 2012 16:39 Laserist wrote: I wonder if it is possible to turn that AI into a script and throw it into the ladder. It would be interesting to see how it will ranked in that case.
Would that be legal under battle.net terms of use?
*serious question*
I would find it very interesting to see AIs matched up against players however not everyone playing ladder might agree to running into what is essetially a bot(even if it would have to be quite sophisticated in order to be a challenge for most ladder warriors).
On May 28 2012 16:39 Laserist wrote: I wonder if it is possible to turn that AI into a script and throw it into the ladder. It would be interesting to see how it will ranked in that case.
Would that be legal under battle.net terms of use?
*serious question*
I would find it very interesting to see AIs matched up against players however not everyone playing ladder might agree to running into what is essetially a bot(even if it would have to be quite sophisticated in order to be a challenge for most ladder warriors).
This bot sees what you are making and makes the counter. Unit for unit. So, even for that, it would not be legit. A real bot on ladder would have to just play based on scouting and such. Very hard to make.
It is DEF not allowed under the TOS. Anything that automates your gameplay is not allowed. If you have a macro keyboard that presses 4sd for you by pressing 1 button, that is against TOS. Let alone, and entire script for game. They should have some sort of forum for developing AI's though. That would be awesome.
On May 29 2012 03:41 CursOr wrote: This bot sees what you are making and makes the counter. Unit for unit.
If it did this it would be much better than it is.
the AI/AI-developer bothers me because the AI is not designed to be particularly reactionary. It runs builds randomly (many of the builds suck) regardless of the opponent's race, so it will oftentimes just get stomped if the opposing player performs the proper reaction.
A good AI should select a build based off the opponent's race, and even potentially the opponent's build order. From there they should go into countering what the opponent is doing as soon as possible.
It's not the easiest way to make an AI, but it's the only relevant way — because that's how this game is played.
Part of the reason why builds aren't updated is because they take so long to make in the current AI engine. (I've had experience with it. I'm also the person who implemented the walling algorithm for the Terran AI)
The current design of the AI engine is very limiting, which makes the task of scripting build orders much more complicated than needed. In addition to that, the AI lacks planning (admittedly, none of the SC2 AIs currently offered have any sort of planning), and therefore any reaction that the AI has, doesn't fit into the big picture.
But let me assure you that there should be something ready for showcase in a month or two...
Can somebody direct as to how to download GTAI maps after patch 1.5? I type it into the search box and nothing comes up, which is how I used to do it. Thanks!
On August 01 2012 12:50 Grantalf wrote: Can somebody direct as to how to download GTAI maps after patch 1.5? I type it into the search box and nothing comes up, which is how I used to do it. Thanks!
God I hope this can be fixed somehow, while I get a good amount of practice from ladder, this was amazing for practicing against certain builds, as well at times when my ladder anxiety is high.
Green Tea AI for patch 1.5 is not yet ready. We hope to have a workable version ready soon ^.^ (porting approximately 20,000 lines of code with only 2 developers is a big task!)
The production system has completely broken down due to the patch. Please wait patiently while we developers write a new (and vastly superior) one to come in the next version.
On August 02 2012 06:32 [JT]Landau wrote: You can find maps in "Arcade" mode, but I tried Cloud Kingdom v. 1.18, it does't work...
Is there a way to download them instead of just playing them through the multiplayer portion? I hate to clog up my game history with gtai practice games.
On August 02 2012 12:45 LookNaph wrote: Hi. This is Naph.
Green Tea AI for patch 1.5 is not ready. We hope to have a workable version ready soon ^.^ (porting approximately 20,000 lines of code with only 2 developers is a big task!)
The production system has completely broken down due to the patch. Please wait patiently while we developers write a new (and vastly superior) one to come in the next version.
On August 02 2012 12:45 LookNaph wrote: Hi everyone. This is Naph.
Green Tea AI for patch 1.5 is not yet ready. We hope to have a workable version ready soon ^.^ (porting approximately 20,000 lines of code with only 2 developers is a big task!)
The production system has completely broken down due to the patch. Please wait patiently while we developers write a new (and vastly superior) one to come in the next version.
On August 02 2012 12:45 LookNaph wrote: Hi everyone. This is Naph.
Green Tea AI for patch 1.5 is not yet ready. We hope to have a workable version ready soon ^.^ (porting approximately 20,000 lines of code with only 2 developers is a big task!)
The production system has completely broken down due to the patch. Please wait patiently while we developers write a new (and vastly superior) one to come in the next version.
Cheers, -Naph
Good luck on GtAI for patch 1.5!
I just wished blizzard would take your mod and make that the AI used in custom games. I tried playing against the standard AI and it just wasn't the same.
note: there seems to be reports that the AI gets stuck after several minutes into the game. Personally I have not been able to reproduce this issue. Leave a report if any of you do.
It's the same one, minus the micro function. Probably to get some basic functions running smoothly with the new patch first, do some testings, then maybe the more advanced stuffs will be re-integrated later on.
So, can I just go to arcade, search for "GTAI" and play? Because that is waht I used to do before 1.5.0. I went to custom, search GTAI then download and thats it. I could play GTAI in single player after I have downloaded the map.
On Americas server, I uploaded 'GT AI Light Antiga Shipyard' to the multiplayer custom games (not arcade). However, as noted above, the search function is pretty buggy and most of the times the map doesn't show up in the search result. Alternatively, you can go to 'create game' --> 'popular' and keep going down until you find the map, lol. I've played a few games on it, so hopefully it isn't too far down. Once you find the map, you can bookmark it.
I'll test the map a few more times to make sure there are no complications, then I'll upload more ladder maps.
On August 09 2012 06:44 DigitalisDestructi wrote: On Americas server, I uploaded 'GT AI Light Antiga Shipyard' to the multiplayer custom games (not arcade). However, as noted above, the search function is pretty buggy and most of the times the map doesn't show up in the search result. Alternatively, you can go to 'create game' --> 'popular' and keep going down until you find the map, lol. I've played a few games on it, so hopefully it isn't too far down. Once you find the map, you can bookmark it.
I'll test the map a few more times to make sure there are no complications, then I'll upload more ladder maps.
Hi, I haven't been able to find any maps with tags of "green tea," "greentea," "gt," or "gt ai" at all on NA. Are they functioning and published still?
Edit: nevermind.. Search function is indeed terrible.
Have you tried the 'create game' --> 'popular' --> 'all' selection? I think you have to click on 'show more' like five times before you come across 'GT AI Light Antiga Shipyard' by omgJCaBomb. Remember it's under custom games in multiplayer, not arcade.
Thanks DD, it's working great. Any tips for getting this in a map? Can't seem to find the 2 XML files that came with previous versions of GTAI so I'm stuck putting it into maps.
1. So, where is this available? In starcraft -> custom or arcade? 2. What do I type into the search bar, GTAI? 3. What Server is this on? I think its only NA and EU at the moment?
I've been testing out the map for a few more times now with no problems so far, so I'll start uploading the entirety of the ladder map pool.
On August 10 2012 03:13 Koronin wrote: Thanks DD, it's working great. Any tips for getting this in a map? Can't seem to find the 2 XML files that came with previous versions of GTAI so I'm stuck putting it into maps.
The unitdata.xml tells the AI where to look for each unit's advanced micro function and it's not included in the current version, hence the 'light' version. I think the author(s) forgot to include the banklist.xml which allows the AI to look for existing saved settings from the GT AI options, so I included the file in the latest map update.
On August 11 2012 01:12 dynwar7 wrote: 1. So, where is this available? In starcraft -> custom or arcade? 2. What do I type into the search bar, GTAI? 3. What Server is this on? I think its only NA and EU at the moment?
1. Custom games under multiplayer mode. 2. The easiest way is to just type the first word of the map's name -- 'antiga,' 'cloud,' 'daybreak,' etc. Unfortunately, as noted in some posts above, the search function is highly unreliable and the map doesn't show up in the search results most of the time. 3. Yeah.
The only reliable way to find this map is to go through 'create game' --> 'popular' --> 'all' and keep scrolling down until you find the map. Since the map is getting some exposure now, I think you only have to click on 'show more' for four times instead of five, heh. Remember that you can always bookmark the map.
Will I be able to run the replay through sc2gears? I noticed that there are a lot of custom maps that won't allow sc2gears to read the data on the replay.
I have a new question. Why is is that when I have found GTAI, I downloaded it, then...it wont show up no my "versus AI"? In other words I have to play online...where I have lag...before 1.5.0 I was able to download GTAI, then play offline where there would be no lag.
Bio upgrades from the tech lab all cost 50/50 more than they should. Thor cannons are researched by default. Zerg seems to think drone rushing is a good idea! Something crazy is happening with creep tumors, dozens being laid on the exact same spot.
On August 14 2012 02:04 Willzzz wrote: Got some serious problems with these new maps.
Bio upgrades from the tech lab all cost 50/50 more than they should. Thor cannons are researched by default. Zerg seems to think drone rushing is a good idea! Something crazy is happening with creep tumors, dozens being laid on the exact same spot.
Green Tea does nothing to change the stats/tech of units. You should inquire the person who published the map to find out how this occurred. However, your concerns about drone rushing and creep tumors are correct; we hope to find an easy fix soon.
Thanks for the update on this! Sure helps when you are trying out off races. Just one question, are there any plans to take out the gold bases on the antiga map?
On August 11 2012 08:34 HornyHydra wrote: Will I be able to run the replay through sc2gears? I noticed that there are a lot of custom maps that won't allow sc2gears to read the data on the replay.
Unfortunately, I don't think sc2gears can read the replays.
On August 13 2012 11:58 dynwar7 wrote: I have a new question. Why is is that when I have found GTAI, I downloaded it, then...it wont show up no my "versus AI"? In other words I have to play online...where I have lag...before 1.5.0 I was able to download GTAI, then play offline where there would be no lag.
I have no idea. You did download the maps from the custom games section, right? That should be enough to store the maps' cache on your computer. Maybe you can try deleting the cache folder under the hidden ProgramData to clean things up. Note that it would remove all maps saved on your computer.
On August 14 2012 02:04 Willzzz wrote: Got some serious problems with these new maps.
Bio upgrades from the tech lab all cost 50/50 more than they should. Thor cannons are researched by default. Zerg seems to think drone rushing is a good idea! Something crazy is happening with creep tumors, dozens being laid on the exact same spot.
http://us.battle.net/sc2/en/forum/topic/6232743757 "Some games appearing in the “Join Game” list of the “Custom Games” menu on the StarCraft II side of the client will cause that game to be played using an older balance. As a temporary work around please start lobbies from the “Create Game” page."
EDIT: GT AI 1.21 is out. I've updated all the old GT AI maps rather than the GT AI Light ones since the update has re-integrated the zerg micro. As usual, they're in the custom games section, and remember to use 'create game' to avoid wacky balance bugs. The title format is 'GT AI Antiga Shipyard,' etc.
On August 23 2012 22:13 dynwar7 wrote: The maker of GTAi, did you release it on SEA server? Perhaps that has something to do with why I cant downlaod the map and play single player.
On August 25 2012 21:16 dynwar7 wrote: What about the other creators of GTAI?
Hope someone will upload to SEA... T_T dont know what I would do without GTAI...llo
Hi. Is there a reason why you won't use the SC2Allin1 tool to launch GTAI? It is actually the preferred way to play against the AI; much faster than have someone upload maps online.
On August 25 2012 21:16 dynwar7 wrote: What about the other creators of GTAI?
Hope someone will upload to SEA... T_T dont know what I would do without GTAI...llo
Hi. Is there a reason why you won't use the SC2Allin1 tool to launch GTAI? It is actually the preferred way to play against the AI; much faster than have someone upload maps online.
Still play these maps all day. Hey... just wanna comment here that even though my Terrans are researching Seige, and they have it, and it would certainly be useful in a lot of situations- they are never seiging them. I now these things aren't always simple to fix, and can be overly complex for such a seemingly simple action. But, as of now, the tanks are just rolling around with the rest of their army, unsieged.
The siegetank problem due to the new patch 1.5.2 , with change some parameter. But I manage to fix these problem already, it will be in siege mod when defend and attack as well. Please wait for the new version of GT AI that will fix this problem .
P/S: It's done, check out the new version here : GT AI 1.22
Just extract the AI to desktop and drag and drop it into the ai folder thats what i always do. On the SCALLIN1 web site there is a guide to installing for dummies(that's what its called) any problems will be listed there if not leave a post about your prob and vernam will answer you.
It took me about a minute after i installed the HOTS beta to play with SCALLIN1 its really simple.
I cant seem to get on the web site at the moment but im pretty sure you need NET framwork to play. Can you not download that from somewere?Here it is >>http://www.microsoft.com/en-us/download/details.aspx?id=17718 but check the guide for dummies just incase im wrong GL http://sc2.nibbits.com/media/tools/files/20110417225357/SC2Allin1_Guide_for_Dummies.pdf
> but i need advice how to fix it, or may > be , switch off the menu (ai vs > ai games seems to be normally)
I fixed it. Try it out: - press F3 (Custom Games) - Choose a map - Create with Mod - Search for "GTAI" - You will find "Green Tea AI (GTAI)"
I have to learn a lot about Galaxy-Script so I just did a few things to get it back working again. Bugs with new units will remain. E.g. SwarmHosts won't burrow sometimes, even if attacked.
> ai vs ai games seems to be normally
It's supposed to do so. Take a look:
if (checkai) { DialogSetVisible(GOpenMenu, PlayerGroupActive(), false); DialogSetVisible(GOptionBoard, PlayerGroupActive(), false); CameraLockInput(1, false); AITimePause(false); TimerPause(GTRoundTime, false); UnitPauseAll(false); TriggerDebugOutput(1, StringToText("No human player, AI automatically start"), true); }
Since "bool checkai = true;" is true (standard value) until there is a player involved, this is normal behavior.
If you want to use GTAI in your maps ... Quick and dirty: - Find "if (g_start > 0)" in "GTZBoard.galaxy" - Put "g_start = 1;" above it => Starts Random General - Find "if (checkai)" in "GTZBoard.galaxy" - Put "checkai = true;" above it => Auto-Start
Is there any way to have the very hard ai not have 400 extra minerals when its goes 4 gate pressure its comes at 5:30 and they have 2 bases, the third comes up by 6:15. its overbearingly strong. Also, Oracles like don't shot.. ever.
On October 29 2014 21:40 NeoBlade wrote: I fixed a few things, but GTAI cheats with minerals since 2011. It's supposed to and the majority of the players want it that way.
If AI playes a 4Gate you have to scout it and build proper defense. If "Hard" is too hard for you, start with "Very Easy" or "Easy".
Note: Elita-AI builds a Pylon about 0:24 which is impossible for a human player. But it always have been this way. I won't change it.
===
I released a new version. All races are attacking more aggressive now.
Protoss Update: If AI builds an Oracle, it will try to kill workers.
Zerg Update: SwarmHosts Micro is awful. Removed the unit from "CounterUnits()".
Just to clarify, I didn't have a problem with the ai getting more resources. I had the problem with the fact i kill everything. Then magically 2 immortals 4 zealots and 2 stalkers appeared out of no where.
Hello, I have been using GTAI ever since I played it, I'm glad you recently updated it . How do I use GTAI the latest version with my maps that are offline? Do I still have to import it like before? I tried that but it is still not working.
I heard the standard way to do it is like so:
- press F3 (Custom Games) - Choose a map - Create with Mod - Search for "GTAI" - You will find "Green Tea AI (GTAI)"
But I want to play one of my offline maps with custom triggers. How can I use GTAI completely offline?
How do I use GTAI the latest version with my maps that are offline? Do I still have to import it like before?
Yes, you have to import it like you did before.
But before you do, you have to make some changes.
It is important to find this section in "GTZBoard.galaxy"
if (checkai) { DialogSetVisible(GOpenMenu, PlayerGroupActive(), false); DialogSetVisible(GOptionBoard, PlayerGroupActive(), false); CameraLockInput(1, false); AITimePause(false); TimerPause(GTRoundTime, false); UnitPauseAll(false); TriggerDebugOutput(1, StringToText("No human player, AI automatically start"), true); }
To always start GTAI, you have change 2 things: "g_start" and "checkai"
In more detail: If you want to use GTAI in your maps ... - Find "if (g_start > 0)" in "GTZBoard.galaxy" - Put "g_start = 1;" above it => Starts Random General - Find "if (checkai)" in "GTZBoard.galaxy" - Put "checkai = true;" above it => Auto-Start
(Quick and dirty)
Example:
// Neo 2014-10-02: "g_start" is set to 1 to start "Random General" g_start = 1;
if (g_start > 0) // Startup: GT Start Menu
and do the same with "checkai" like mentioned above.
I opened up GTZBoard.galaxy in notepad++ and did what you said, but I don't think GTAI is working, I think it's normal AI. GTAI is usually waaay better so I don't think it's working. I copied my GT2AI 0.20 folder, edited GTZBoard.galaxy and imported into a map that already has this version of GTAI on it (I assume it will just overwrite the file?). Any idea on what's going on?
- Bugfix: 7 Ebays with 1Rax-FE - Increased probability: CannonRush = 12,5% - Decreased probability for Oracle-Attack - GTAI now always reveals its name - Terran attacks with Banshees (if available) - Help AI with Oracle-Production - included "GTAI 0.26" - Refactoring
I may have posted in this thread more than anyone except OP. I had given up on this thing since the HOTS release!
IDK how long it has been up and running but I am so happy!
I've been playing VS hard as Zerg... Terran is super easy. Protoss is super easy. Zerg seems very difficult. Is this just because my macro is bad ... is there any chance that the Zerg AI has some sort of income hack that the other races don't in Hard Mode?
Other than that... my 1v1 is going to to go hell because I love playing the GTAI so much ... thank you VERY much.
> Terran is super easy. Protoss is super easy. Zerg seems very difficult.
:D ... I have to laugh because a student of mine (I am a SC2-Coach), said almost the same thing: "Terran is super easy. Zerg is super easy. Protoss seems very difficult."
Please try again 2 times against all 3 races (6 Replays) and post a Replay here. Thanks!
> Do the commands still work? Like -qr for quick restart? > Have they been changed? I tried typing -help in game > and it's not working. Was this removed?
I don't know. I know I did not change anything in this matter. If they do not work, you have to figure out why. They have a low prio to me. "-qr" will probably be fixed in the next few month.
It is obvious that this is still being worked on. I appreciate it.
It depends on the race you play
ZvZ is hard because this thing is constantly throwing units at you ... so a round of drones at the wrong time is your death. I feel like its really making me better at injects and everything while fighting is going on. I am usually a mid diamond player, but this season am high platinum because its fucking rough in there lol.
This replay has the Zerg "Cardis" on Shattered Temple just suddenly spawn in a bunch of units at 12:55...
No eggs, i thought okay maybe they were burrowed and the animation bugged, but upon replay inspection yeah it just conjured up the units. I call shenanigans!
The old AI's like Incubus and Araq were harder, some of these new ones are hilariously bad (besides the spawn hack!)
Thanks for all your work, once again... GTAI #1 mod NA
Thank you very much! I was very happy, when I read your comment.
> but upon replay inspection yeah it just conjured up the units.
I know, but since the GTAI always cheated I do not see a problem. I will take a look at your replay, though.
Spawning units only is a problem, if the GTAI opens with Triple-Hatch-Before-Pool and sends 7 Roaches at 4:46 at your base (7RR). But at 12:55 you should have Watchtowers and see the attack coming.
Spawning units should not happen if you use "Very easy" and "Easy". IIRC cheating starts with "Hard" or "Harder".
If you want to fix it yourself: Well that's very easy :D. You just need to debug about 9.900 lines of code.
Why?
GTAI uses a "Commander-System", which is a good thing, do not get me wrong. Every BO is put into a single file. 21 files per race.
Good luck! Neo
PS: In other words ... I do not have the time to search for it.
Thanks for all your comments and help! Enjoy your new years eve, folks!
New version 12.1
- Bugfix: Terran-AI "Cade" did not work => CRITICAL BUG - Bugfix: Terran-AI "Cade" tries to swap buildings without AddOns - Bugfix: Banshee-Harass for Protoss? AI_RACE == Protoss - Bugfix: Terran creates Oracles - Help AI with Cloak (Terran) - GTAI: Viper-Micro is awful, tried to remove unit - Bugfix: GTAI-Terran always plays Banshees - Bugfix: Contrllr of plyr (Triggering Player) == AI => user - Changed "GTSupply" for Terran - Tested changes to "GTName" - Bugfix: Commander "Kate" SwapBuildings - Included "GTAI 0.29" - Rizzo-Mech-BO => Terran2.galaxy - Preparde GTAI for "Elite" and "Insane" - Added Hellbats and Hellions in AttackWaves - Created first BO "Rizzo-Mech" - Improvements to mech BO - Bugfix: Terran just plays one BO, removed test trigger
I used to use this with SC2AllinOne to play custom maps offline but it looks like these new versions are only on Battle.net. So how would I get the mod only for use in offline?
I did create a map with the mod on Battle.net so I assume I have the mod downloaded on my computer somewhere. Where do I find it? And how do I import the mod for my custom map with the editor?
EPT![[image loading]](http://s328.photobucket.com/albums/l344/ptanhkhoa/?action=view¤t=Screenshot2011-08-1317_50_42.jpg)
![[image loading]](http://i328.photobucket.com/albums/l344/ptanhkhoa/star2/Screenshot2011-08-0921_25_45.jpg)
![[image loading]](http://s328.photobucket.com/albums/l344/ptanhkhoa/?action=view¤t=Screenshot2011-08-0611_12_54.jpg)
. Still need more testing and feedback.
![[image loading]](http://s328.photobucket.com/albums/l344/ptanhkhoa/?action=view¤t=Screenshot2011-06-2511_26_09.jpg)
![[image loading]](http://s328.photobucket.com/albums/l344/ptanhkhoa/?action=view¤t=Screenshot2011-06-2511_27_00.jpg)
. I strongly recommend not attacking the ai for the first 10 minutes or so of the game so it can build up its army and economy. It makes for a much better challenge.![[image loading]](http://www.gamereplays.org/community/uploads/repimgs/repimg-33-217645.jpg)
![[image loading]](http://s328.photobucket.com/albums/l344/ptanhkhoa/?action=view¤t=GTencoder.jpg)
![[image loading]](http://i328.photobucket.com/albums/l344/ptanhkhoa/mappacks.jpg)
![[image loading]](http://s328.photobucket.com/albums/l344/ptanhkhoa/?action=view¤t=Screenshot2011-07-2523_27_02.jpg)
![[image loading]](http://s328.photobucket.com/albums/l344/ptanhkhoa/?action=view¤t=Screenshot2011-07-2523_27_12.jpg)
![[image loading]](http://s328.photobucket.com/albums/l344/ptanhkhoa/?action=view¤t=Screenshot2011-07-2523_27_33.jpg)
![[image loading]](http://s328.photobucket.com/albums/l344/ptanhkhoa/?action=view¤t=Screenshot2011-07-2523_28_05.jpg)
![[image loading]](http://s328.photobucket.com/albums/l344/ptanhkhoa/?action=view¤t=Screenshot2011-07-2817_26_43.jpg)
![[image loading]](http://s328.photobucket.com/albums/l344/ptanhkhoa/?action=view¤t=Screenshot2011-07-2913_15_26.jpg)
![[image loading]](http://s328.photobucket.com/albums/l344/ptanhkhoa/?action=view¤t=Screenshot2011-08-0622_54_48.jpg)
![[image loading]](http://i.imgur.com/SeewINZ.png)
