Spring RTS Engine

el_matarife wrote:These were removed a long time ago, in Absolute Annihilation 2.2, and have never been in BA since BA was forked from AA after this change took place. When you say last year, I think you mean Sep 09, 2006 when AA 2.2 was released.

Not impossible since I cannot really remember. Got to ask though, do you know why they where removed?

Something I see missing in the current BA is the tech progression in all vehicle types. Look at Kbots for example, you have T1, then T2 factory and finally a T3 gantry.
I think it should be the same with ships (T3 shipyard for flagships), hovercraft (T2 hover platform and T2 hovers missing), seaplanes (T2 platform and T2 seaplanes missing) etc
Also, if the long range plasma vehicles are ever introduced Id say they should be in a T3 vehicle factory.

Something I just remembered and might be considered a bug is the size of the missile the atomic bomber fires, its HUGE and it just looks silly...

Damn, Im full of opinions today :)

Regards
/Doxs

Doxs wrote: Not impossible since I cannot really remember. Got to ask though, do you know why they where removed?

It either seemed like a good idea at the time or that was the beginning of the developers complete mental breakdown. I think its a little of column A, a little of column B.

el_matarife wrote:

Gota wrote:IF an AI was developed for a specific mod i don't see why AI wouldn't become much better at playing that mod.
It could have tons of harcoded things and it would work much better i think.

If that was true, why aren't AIs totally dominant in most RTSes instead of reliant on cheats to provide a "challenge" in most cases? (Challenge in quotes because honestly, most RTS are not challenging even with cheats.)

Doxs wrote:I have not read the entire thread so i dont know if someone has reported this yet, but I find a few units and unit types missing in the game.
Specifically the following units:
The old mobile long range plasma cannon tanks built in the T3 factory.
The ships with the same long range plasma cannons.
Cant remember from what mod and Im at work, but it was a videly played mod last year. (AA/BA 3.* or something?)

These were removed a long time ago, in Absolute Annihilation 2.2, and have never been in BA since BA was forked from AA after this change took place. When you say last year, I think you mean Sep 09, 2006 when AA 2.2 was released.

nobody spends much time on an AI??whats the point of there is multiplayer and scripted campaigns...
But as a hobby over many years a good AI could have been developed for BA by just hardcoding many things..

an llt...llt placement can depend on a few things like distance of enemy start location and from your own,distance from near by llts or other defensive structures,distance from mexes,how much of the range of an llt's spot can be covered by the llt due to other structures in its way or terrain.
If an AI can take a map and Analyze its terrain for the mod,like the map small divide,and than Analyze the metal map it will decide that the best place to place llts are at high spots where they cover more ground near the middle where the concentration of metal is highest..and its on the way to the start position of an enemy.

Gota wrote:

el_matarife wrote:

Gota wrote:IF an AI was developed for a specific mod i don't see why AI wouldn't become much better at playing that mod.
It could have tons of harcoded things and it would work much better i think.

If that was true, why aren't AIs totally dominant in most RTSes instead of reliant on cheats to provide a "challenge" in most cases? (Challenge in quotes because honestly, most RTS are not challenging even with cheats.)

Doxs wrote:I have not read the entire thread so i dont know if someone has reported this yet, but I find a few units and unit types missing in the game.
Specifically the following units:
The old mobile long range plasma cannon tanks built in the T3 factory.
The ships with the same long range plasma cannons.
Cant remember from what mod and Im at work, but it was a videly played mod last year. (AA/BA 3.* or something?)

These were removed a long time ago, in Absolute Annihilation 2.2, and have never been in BA since BA was forked from AA after this change took place. When you say last year, I think you mean Sep 09, 2006 when AA 2.2 was released.

nobody spends much time on an AI??whats the point of there is multiplayer and scripted campaigns...
But as a hobby over many years a good AI could have been developed for BA by just hardcoding many things..

an llt...llt placement can depend on a few things like distance of enemy start location and from your own,distance from near by llts or other defensive structures,distance from mexes,how much of the range of an llt's spot can be covered by the llt due to other structures in its way or terrain.
If an AI can take a map and Analyze its terrain for the mod,like the map small divide,and than Analyze the metal map it will decide that the best place to place llts are at high spots where they cover more ground near the middle where the concentration of metal is highest..and its on the way to the start position of an enemy.

You seriously underestimate the difficulty of writing an AI. You can't just hardcode everything. Every time you place a building you're making a decision based on hundreds of variables. What's the map's metal distribution like? What is the terrain like? What are the other players doing? Which side are you? And so on.

Don't even try to tell me otherwise, I've taken classes on AI and written them myself. Writing good AIs is very hard even for games with very limited rule sets (like chess or othello). Besides that, a good AI takes a lot of processing power. I suppose theoretically you could write a minimax AI for a game like BA, but there are so many possible moves that no computer in existence today (or even ten or twenty years from now) would have even the slightest chance of being able to run it.

The only way a hardcoded AI could work is if it was designed for a specific map against a specific player with a specific playstyle that's exactly the same every single time.

BeefofAges wrote:The only way a hardcoded AI could work is if it was designed for a specific map against a specific player with a specific playstyle that's exactly the same every single time.

Sounds like BA to me!!

BeefofAges wrote: Don't even try to tell me otherwise, I've taken classes on AI and written them myself. Writing good AIs is very hard even for games with very limited rule sets (like chess or othello). Besides that, a good AI takes a lot of processing power. I suppose theoretically you could write a minimax AI for a game like BA, but there are so many possible moves that no computer in existence today (or even ten or twenty years from now) would have even the slightest chance of being able to run it.

He's 100% right, look at AIs for Go, even the best ones running on supercomputers can't even play an average human opponent. Balanced Annihilation is probably orders of magnitude more complex than Go.

Gota wrote: nobody spends much time on an AI??whats the point of there is multiplayer and scripted campaigns...

You'll notice in most games with scripted campaigns the AI will usually start with a fully built base laid out by a human. Why? The AI sucks at designing well defended large bases, and the AI sucks at expanding properly. You give them a fully built base so they don't have to expand and they don't have to design a defensible base, they just have to build units and throw them at you.

A BA AI with more hard coded elements would be much, much better than the adaptive, play-any-mod AI's we currently use. It wouldnt be dominating real players, but it would play several orders of magnitude better than what we have.

A comet flashspam bot, using a few simple rules, such as a simple 3-mex starting build order with 2 llts that cover as much of the base as possible from opposite sides of it.

From there it just needs to expand mexes making 1 llts per mex (because thats simple), keep its energy and constructors in pace with its economy, and spam flash in large roving groups that patrol its territory and harass the enemies outlying mexes (Rather than just pouring directly at his base). Since Day/Andy play every map as if its Comet anyway, it should probably be able to play a lot of other maps if it can do that.

that is what NTAI is about. try it.

I made an NTAI profile for XTA than played and ecoed better than most non-elite players, it only failed at micro

ntai is basicly a build your own ai toolkit, it fucking owns

Could you share that profile, please ?

sure thing;
http://spring.jobjol.nl/show_file.php?id=1710

be warned that making a good ntai profile takes a large investment of time to make, this one took about 16 hours

If AF ever releases an up to date version of NTAI and toolkit for easy download without the description "I couldnt be bothered to test this so it probably dosnt work", I'll make up a ba profile, because I imagine ntai could play ba a great deal better than kai/rai

next person to post has herpes

The important question is "if i invest x hours more will it be y times better?or it cant be better than z no matter how much time i invest.

1v0ry_k1ng wrote:sure thing;
http://spring.jobjol.nl/show_file.php?id=1710

be warned that making a good ntai profile takes a large investment of time to make, this one took about 16 hours

If AF ever releases an up to date version of NTAI and toolkit for easy download without the description "I couldnt be bothered to test this so it probably dosnt work", I'll make up a ba profile, because I imagine ntai could play ba a great deal better than kai/rai

next person to post has herpes

Much apreciated ! Thank you ^^

Can someone take a look at improving the Arm rocket spiders? They seem to really suck right now. They're kinda slow, short ranged, don't do much damage, and don't have much HP. At the very least, they should be pretty quick. Arm is all about quick, weak units, right?

Let's compare the Recluse to the Termite:

Arm Recluse:
Health: 1050
Metal/Energy cost: 375/4200
Build time: 8775
Weapon estimated DPS 95 (This is probably wrong since burst weapons don't generate DPS properly on Modinfo. It looks like a burst of 3x 95 damage shots every three seconds.)
weapon range = 550
sight distance = 338
Weapon Velocity = 450
Start Velocity = 100
Weapon Acceleration = 150
Area Of Effect = 72

Core Termite:
Health: 2800
Metal/Energy cost: 805/12605
Build time: 18834
Weapon estimated DPS: 227.27 (One 250 damage shot every 1.1 seconds)
weapon range = 340
sight distance = 364

One immediately obvious issue is that while the Recluse has a decently long range weapon, it doesn't have any line of sight to go with it. It outranges the Termite on paper, but not in a lot of practical situations since the Termite has longer LOS. Next issue seems to be that the Termite has way more DPS since a 250 damage shot every 1.1 seconds is going to be a lot better than 3x 95 damage shots every three seconds. The AOE may help a bit, but rocket units are tough to use in swarms against swarms because of friendly fire. The AOE almost certainly doesn't help in the Recluse's primary role of base busting. The health on the Recluse is more like that of a T1 or artillery unit, not what's ostensibly a T2 combat unit.

My suggestions: Give the recluse 100 to 150 more LOS, boost the health to 1250 or maybe a bit more, maybe look into boosting the damage per shot.

Termite uses a laser weapon.
A laser's damage goes down with range so he does close to the dmg he should be doing only when the enemy is adjacent to him.
however this is the disadvantage of laser weapons but the advantage is the fact they hit a target immediately and are extremely accurate
while rockets or plasma weapons aren't so maybe it can be ignored...

Gota wrote:Termite uses a laser weapon.
A laser's damage goes down with range so he does close to the dmg he should be doing only when the enemy is adjacent to him.
however this is the disadvantage of laser weapons but the advantage is the fact they hit a target immediately and are extremely accurate
while rockets or plasma weapons aren't so maybe it can be ignored...

Right, I didn't mention the damage falloff on lasers. Frankly, the Recluse's rockets are slow enough they miss a lot too, so it isn't a big factor. I also didn't mention the 35 energy per shot charge on the Termite since it doesn't seem too bad compared to the 75 on the Can or Sumo for instance.

Comparing useless units to useless units is silly imo.

I have tried hard to get any meaningful use out of termite/recluse/spider on DSD and Altored (both maps with cliff-shaped holes in the lolololBAteemgaem porc) as well as things like putting them on the hills on Foothills etc, and they always fail horribly as soon as any equivalent cost of non-fail (read t1 veh) mobiles get within range.

Imo compare them stat for stat to flash, then prescribe X value to the largely useless) all terrain ability, then "balance" them off the back of that.

Then realise it will never get changed anyway.

Okay, I can see them failing on DSD, since DSD has enough avenues to get around, and because the cliffs are a good place to put turrets so they're usually guarded... but Altored Divide? There aren't usually a lot of turrets on top of the cliffs of Altored, so you can often use a squad of all-terrains to get over them and plunge deep into the enemy base and eat their econ.

One of the things that makes all terrain units more viable as a surprise attack is just how expensive they are, if they were cheap everyone would just automatically porc their cliffs but as it is people rarely bother so if you can scout out a good undefended attack route all terrain can be useful - not every game but on occasion.

pintle wrote:Then realise it will never get changed anyway.

Maybe both units have already been changed :p