I mainly released this version to give ppl an ai to play with for the new spring version. there might be some balance issues since it's a very quick release this time
Changelog:
AAI v0.55:
- AAI is now compatible with the modified ai interface of spring 0.70
- Extended map learning files/every mod now creates own map learing files
- AAI tries not to send builders in setors which turned out to be too dangerous
- Fixed a rounding related crashbug
Edit:
AAI 0.52 (compatible with 0.67b3):
http://www.fileuniverse.com/?p=showitem&ID=2445
AAI 0.55 (compatible with 0.70bx):
http://misc.twam.info/alex/AAI_0_55.zip
AAI 0.55 - compatible with new spring version
Moderators: hoijui, Moderators
AAI 0.55 - compatible with new spring version
Last edited by submarine on 25 Feb 2006, 23:29, edited 3 times in total.
Can't talk... too busy playing...
Seriously, AAI just keeps getting better. The new mod support is great! I've been playing mostly Final Frontier, with a little Gundam and Xect vs. Mynn thrown in. The only crash I got with this latest release (0.52) was on a buggy map (Core Prime Industrial Area). At this point, AAI is by far the most stable AI.
Good work, Submarine!
Although when posting your updates in the future, it might be better to rename an existing forum thread rather than creating so many new ones... just a thought :)
Seriously, AAI just keeps getting better. The new mod support is great! I've been playing mostly Final Frontier, with a little Gundam and Xect vs. Mynn thrown in. The only crash I got with this latest release (0.52) was on a buggy map (Core Prime Industrial Area). At this point, AAI is by far the most stable AI.
Good work, Submarine!
Although when posting your updates in the future, it might be better to rename an existing forum thread rather than creating so many new ones... just a thought :)
sorry guys, i'll promise to keep things better organized next-time. there have been so many other things the last few days (end of my semester, exams, cleaning up my flat, transfering to my girlfriend) so i got a little bit confused.XigXag wrote: Although when posting your updates in the future, it might be better to rename an existing forum thread rather than creating so many new ones... just a thought :)
thx for your positiv feedbacks :)
I have to concur with IMsabbel... awesome piece of work on FF,- once I changed the max_metal limit to 1000000 from 6000 (why is it necessary anyway?) it started sending piledrivers against my rather cramped base on comet catcher (I hadn't expanded being under constant assault from the three trained up ais on zero penalty)(they're a bit harder once they figure out which units are good for attacking with! 
)
Anyway she was a bit white knuckled for a while and I had my first invasion force slaughtered by core battleships that one of the ais was saving up for a rush. hehe
good work man, keep it up!
)Anyway she was a bit white knuckled for a while and I had my first invasion force slaughtered by core battleships that one of the ais was saving up for a rush. hehe
good work man, keep it up!
oh shit, i forgot that ff has higher metal costs and income than most other mods. basically i introduced this to give players the opportunity to prevent aai from building too expensive units an thus lacking ressources to build other assault units.mecha wrote:I have to concur with IMsabbel... awesome piece of work on FF,- once I changed the max_metal limit to 1000000 from 6000 (why is it necessary anyway?)
but you can easily disable that feature by setting higher values...
btw: since i dont play most of the mods aai supports the mod-cfg files are most likely not quite optimal. if anyone created better ones (like yuritch for sw:ta) feel free to send them to me so i can include them in further releases
-
GrOuNd_ZeRo
- Posts: 1370
- Joined: 30 Apr 2005, 01:10
excellent job there, WD is actually working, w00t!
It did crash on Small Devide but not before letting me have a long and challenging game, I was winning anyway.
The AI refused to build higher end stuff frequently, it mainly swarmed me with infantry and co-in planes.
I haven't played OTA/XTA/etc yet, I might try TAFF :)
It did crash on Small Devide but not before letting me have a long and challenging game, I was winning anyway.
The AI refused to build higher end stuff frequently, it mainly swarmed me with infantry and co-in planes.
I haven't played OTA/XTA/etc yet, I might try TAFF :)
-
Optimus Prime
- Posts: 755
- Joined: 03 Oct 2005, 14:31
ok i tested the ai and it works fine :)
Well done!
If it is allowed to make some suggestions:
1. Let the ais try to first build the best mine they can (lvl3 before lvl 2 before lvl 1). So they would only build lvl 1 as long, as they have only lvl 1 constructors. Later in the game they would build lvl 2 and then lvl 3 like humans do.
2. Let them build more economic buildings. The eco is very weak in comparison to a human. More metal makers and energy facilities.
3. Instead of guarding the factories with all constructrs, let the ai build more factories and or nanotowers.
4. the ai should expand more. I tried Asteroid battle and the ais dont use the asteroids near to them. They only build on the platforms and perhaps one or 2 mines on an asteroid.
That was it for the first. I have to say that i dont have a plan how the ais work, so perhaps its not possible what i think, but that were the first impressions after 3 ai games.
Hold on the good work - and thanks for supporting FF
Well done!
If it is allowed to make some suggestions:
1. Let the ais try to first build the best mine they can (lvl3 before lvl 2 before lvl 1). So they would only build lvl 1 as long, as they have only lvl 1 constructors. Later in the game they would build lvl 2 and then lvl 3 like humans do.
2. Let them build more economic buildings. The eco is very weak in comparison to a human. More metal makers and energy facilities.
3. Instead of guarding the factories with all constructrs, let the ai build more factories and or nanotowers.
4. the ai should expand more. I tried Asteroid battle and the ais dont use the asteroids near to them. They only build on the platforms and perhaps one or 2 mines on an asteroid.
That was it for the first. I have to say that i dont have a plan how the ais work, so perhaps its not possible what i think, but that were the first impressions after 3 ai games.
Hold on the good work - and thanks for supporting FF
no guarantee that it will work but try this:
- increase
MAX_MEX_DISTANCE 4
MAX_METAL_MAKERS 20
in the ff cfg-file. (default values can be found in the example.cfg )
this should make aai build mexes more distant to its base and increase the max. number of metal makers
right now i'm trying to find a proper way to replace cheap mexes with better ones as the game goes on. this will hopefully be implemented in the next release...
"3. Instead of guarding the factories with all constructrs, let the ai build more factories and or nanotowers. "
-> good idea, i'll add that after being done with some othr stuff
- increase
MAX_MEX_DISTANCE 4
MAX_METAL_MAKERS 20
in the ff cfg-file. (default values can be found in the example.cfg )
this should make aai build mexes more distant to its base and increase the max. number of metal makers
right now i'm trying to find a proper way to replace cheap mexes with better ones as the game goes on. this will hopefully be implemented in the next release...
"3. Instead of guarding the factories with all constructrs, let the ai build more factories and or nanotowers. "
-> good idea, i'll add that after being done with some othr stuff
-
Optimus Prime
- Posts: 755
- Joined: 03 Oct 2005, 14:31
ok after some more testgames i came to the following results:
increasing MAX_MEX_DISTANCE - has no effect
increasing MAX_BASE_SIZE - has no effect
the ais economy has to be improved if you want to create an ai, which is better as most noobs. For that i have some suggestions:
- let the ai build better mexes (I know, you know that, but i repeat it, cause of a complete list)
- if the ai gets more than 1k energy plus, it should build metalmakers (better mohomakers) as long, as it gets 1k energy plus or more
- also the ai should build metalmakers as long as it has an metalleak
- if the ai has an energyleak, it should spam energy *immediately*
- the ai should first build the advanced shipyard if it gets 80 metal or more (right now the ai builds every shipyard very fast and supports them with their construction units, but that doesnt help, cause the ai dont get enough metal..)
- tha ai should build the titan yard if it gets 250 metal or more (i saw one ai which tried that with 50 metal)
- the ais construction units should only support production yards, if it has enough metal and energy, else it should support the constructors, which build the eco
- let the ai expand with time. Not only mexes but defenses too. In OTA the ai was only limited by the 250 unit limit, but would have build over the whole map if the limit would be higher (i think this is not as important as the other suggestions, but for a later version it could be usefull if you want to make a real strong ai).
These are my suggestions. I dont know if you can script that, but i m sure if most of these points are implemented, the ai will be very good in eco and as good, as most real players. The eco is the mainpart of a win so you should first improve that before improving their combat abilities. But i think the combat isnt that bad in the actual version.
I hope i could help you.
increasing MAX_MEX_DISTANCE - has no effect
increasing MAX_BASE_SIZE - has no effect
the ais economy has to be improved if you want to create an ai, which is better as most noobs. For that i have some suggestions:
- let the ai build better mexes (I know, you know that, but i repeat it, cause of a complete list)
- if the ai gets more than 1k energy plus, it should build metalmakers (better mohomakers) as long, as it gets 1k energy plus or more
- also the ai should build metalmakers as long as it has an metalleak
- if the ai has an energyleak, it should spam energy *immediately*
- the ai should first build the advanced shipyard if it gets 80 metal or more (right now the ai builds every shipyard very fast and supports them with their construction units, but that doesnt help, cause the ai dont get enough metal..)
- tha ai should build the titan yard if it gets 250 metal or more (i saw one ai which tried that with 50 metal)
- the ais construction units should only support production yards, if it has enough metal and energy, else it should support the constructors, which build the eco
- let the ai expand with time. Not only mexes but defenses too. In OTA the ai was only limited by the 250 unit limit, but would have build over the whole map if the limit would be higher (i think this is not as important as the other suggestions, but for a later version it could be usefull if you want to make a real strong ai).
These are my suggestions. I dont know if you can script that, but i m sure if most of these points are implemented, the ai will be very good in eco and as good, as most real players. The eco is the mainpart of a win so you should first improve that before improving their combat abilities. But i think the combat isnt that bad in the actual version.
I hope i could help you.
A couple things I think would improve the AI dramatically:
1a. If the AI is out of metal, do a reclaim order on all metal debris within the base border, or within the base border + 1 sector. This would give a much needed boost to metal, as well as clearing up the piles of bodies that seem to accumulate in a number of maps. I remember someone said you can do a reclaim on area, just set the are to be the sector size from the middle of the base.
1b. The other idea is to have a builder set to patrol to the center of each base sector, and it will pick up bodies as needed, as long as it doesn't see a new construction project while patrolling.
2. I like the multiple factory idea. Not only provides faster building, but also redundancy in case one is blown up, as well as freeing builder units to do more productive tasks.
3. My view on metal makers is that they should be built only after all metal spots within the base confines (or base confines + 1 sector) have been built on. Then, build/activate metal makers for a surplus of 100 power, or moho makers for 1000 surplus.
4. For upgrading extractors, do a reclaim on the metal extractor, then build the moho extractor on the same spot. This gives a nice ROI (return on investment) plus avoids damaging nearby buildings with an autodestruct command.
5. Bit longer term, but I think the learning coefficients need a slightly different focus. I'd suggest ground_light, ground_heavy, air, base_defense, general_building, and sea. There is just too much variation in ground units to have a single efficiency coefficient for all ground units. I'd recommend splitting the light/heavy with speed or hp.
-QMan
1a. If the AI is out of metal, do a reclaim order on all metal debris within the base border, or within the base border + 1 sector. This would give a much needed boost to metal, as well as clearing up the piles of bodies that seem to accumulate in a number of maps. I remember someone said you can do a reclaim on area, just set the are to be the sector size from the middle of the base.
1b. The other idea is to have a builder set to patrol to the center of each base sector, and it will pick up bodies as needed, as long as it doesn't see a new construction project while patrolling.
2. I like the multiple factory idea. Not only provides faster building, but also redundancy in case one is blown up, as well as freeing builder units to do more productive tasks.
3. My view on metal makers is that they should be built only after all metal spots within the base confines (or base confines + 1 sector) have been built on. Then, build/activate metal makers for a surplus of 100 power, or moho makers for 1000 surplus.
4. For upgrading extractors, do a reclaim on the metal extractor, then build the moho extractor on the same spot. This gives a nice ROI (return on investment) plus avoids damaging nearby buildings with an autodestruct command.
5. Bit longer term, but I think the learning coefficients need a slightly different focus. I'd suggest ground_light, ground_heavy, air, base_defense, general_building, and sea. There is just too much variation in ground units to have a single efficiency coefficient for all ground units. I'd recommend splitting the light/heavy with speed or hp.
-QMan
-
Optimus Prime
- Posts: 755
- Joined: 03 Oct 2005, 14:31
1b. The other idea is to have a builder set to patrol to the center of each base sector, and it will pick up bodies as needed, as long as it doesn't see a new construction project while patrolling.
idea is good, but i think it wont work with an ai. Think of a damaged fighter which flies near to the patroling constructor and which is on the way to the enemy. The constructor would follow this fighter and die or at least would spent a lot of time flying after the fighter. A human player would stop the fighter or the constructor, but i dont know if thats possible for an ai.
The rest of the ideas sounds good
idea is good, but i think it wont work with an ai. Think of a damaged fighter which flies near to the patroling constructor and which is on the way to the enemy. The constructor would follow this fighter and die or at least would spent a lot of time flying after the fighter. A human player would stop the fighter or the constructor, but i dont know if thats possible for an ai.
The rest of the ideas sounds good