Design no. 1 - NanoBlob v. 0.2 (defeat replays)

Design no. 1 - NanoBlob v. 0.2 (defeat replays)

Here is where ideas can be collected for the skirmish AI in development

Moderators: hoijui, Moderators

User avatar
Argh
Posts: 10920
Joined: 21 Feb 2005, 03:38

Design no. 1 - NanoBlob v. 0.2 (defeat replays)

Post by Argh »

Edit: version 0.2is now up. One-tenth the size of the original, and at least 10 times the fun. Check out the new units... the modified units... the madness that is NanoBlob...

And, yes... it works just great with KAIBlob. Just finished a two-fricking-HOUR game of crush against one AI on Ashap Plateau (that map is incredibly hard, if you don't win on an early rush, but I think I learned some things from this go around).

http://www.wolfegames.com/TA_Section/NanoBlobs.sdz

Here's a replay showing one way that you can very easily defeat KAIBlob, for those of you who think this game design is "boring" or "pointless" because you always get pwnd by the Blob ;) KAIBlob uses maximum efficiency... but that doesn't mean it's very smart. Watch and learn...

http://www.wolfegames.com/TA_Section/06 ... 67b3-0.sdf

2 vs. 2... great game with FizzWizz... good fun, lots of silly stuff :-)

http://www.wolfegames.com/TA_Section/06 ... 67b3-1.sdf
Last edited by Argh on 17 Jan 2006, 09:16, edited 5 times in total.
User avatar
krogothe
AI Developer
Posts: 1050
Joined: 14 Nov 2005, 17:07

Post by krogothe »

lol, its quite crazy, and could prove fun (doubt it will have lasting appeal with 6 units however, but im sure youll add more), Personally i quite liked the idea (think i seen something very similar in a WC3 mod/map)

couple of "little" problems however...
in a single minute i managed to make 43 units, being about 80% efficient...
in 2 minutes, in theory you could have 1000+ units, and in a big map that would be easy to do before the enemy gets to you in significant numbers.
By simply making an equal mix of attacking units (by cost) and attacking together with them even a simple AI would absolutely (for the lack of a stronger word...) rape any human in under 5 minutes if its playing at a speed any greater than 0.1...
But i dont wanna be a bastard and put you down, the idea is cool as hell and once the games properly balanced, i might even have a go at making an AI for it (provided it proves to be fun yadda yadda) while i wait for the cheat interface to work!
good to see you making some real work :)


EDIT:
in fact, it was so trivial i decided to do it... it took me 5 mins to code KAI to play it (poorly, but fully). Worked first time and with both sides...
Source has already been replaced back to original KAI so no upgrades will be made
Last edited by krogothe on 14 Jan 2006, 03:02, edited 2 times in total.
User avatar
krogothe
AI Developer
Posts: 1050
Joined: 14 Nov 2005, 17:07

Post by krogothe »

here it is...
not as good as a human can be but 800 units in 5 mins should keep you busy. Shouldnt stall and just expand stupidly...
Should be compatible with the next versions if you only adding more units...
http://www.fileuniverse.com/?p=showitem&ID=2157
(hmm funny... gamemode posted and 1 hour later a pretty much complete AI is released for it. Check the post times! :D)
edit2:
hmm had 2 games and im bored of it already :(
edit3:
im sobering up now...
what the hell lol? theres no strategy, risk taking or anything that you said! Its not been presented to RTS gamers because its not RTS! just build as fast as you can and attack, an APM (actions per minute) contest! I accept you made a challenge for an AI to be made, thats fine, but Im not making AIs for your mods if they come with this kind of false advertising :?
Chocapic
Posts: 556
Joined: 16 Oct 2005, 03:35

Post by Chocapic »

well, im on to what krogothe said, if you want ai's to beat humans let time speak for itself, dont make these kind of mods destroying the interesting gameplay and action and everything of a very nice ->>RTS<<-
User avatar
Argh
Posts: 10920
Joined: 21 Feb 2005, 03:38

Post by Argh »

<chuckles>

You're finding it boring, eh? Out-microing and out-thinking a vast horde isn't working for you?

Thanks for making an AI that we can work against, we'll keep messing with it :)
User avatar
smoth
Posts: 22309
Joined: 13 Jan 2005, 00:46

Post by smoth »

why is this in the ai forum?
Targon
Posts: 99
Joined: 16 Dec 2005, 05:15

Post by Targon »

I thought he was doing this for the people making the "works with everything" no cfg AIs in order to test if the Ai would work with everything without specific instructions.
Chocapic
Posts: 556
Joined: 16 Oct 2005, 03:35

Post by Chocapic »

a thing with these concepts doesnt work ok , with none ai unless its changed, because the mod itself doesnt make much sense..
its just a fps game or rpg..
User avatar
krogothe
AI Developer
Posts: 1050
Joined: 14 Nov 2005, 17:07

Post by krogothe »

I believe it would be pretty easy to change existing AIs to work with it, I only had to change KAIs buildtree and comment out the mex code.
Argh, this game has potential in some ways, but the repetitiveness and lack of strategic elements is letting it down.

By strategic elements i mean things that are worth capturing/controlling/obtaning that give you a tactical advantage against your enemy. Its mexes/geos/high ground/fortified positions/commander in spring. Each and every successful RTS game has strategic elements of some kind, and I believe the more dramatic their effects the better it helps the game. Nanoblobs only has the builders as a strategic element and they are not too well done (eg, having 100 or 500 builders makes only a 30 sec difference, and if you lose them all you will 99.999% of the time lose). The comm having some special abilities would help, as well as some sort of strategic feature on maps. (thats one of the reasons people dont like metal maps except for the occasional bash)

Every action a player takes should produce a meaningful and noticeable consequence to the flow of the game It doesnt need to be immediate. your game needs too many actions and they dont amount to that much change in the overall game. Its repetitive!


You should never be able to foretell who will CERTAINLY win the game until the very end eg, better play should help your winning more and more but not guarantee it until the very end, there should aways be the possibility of an amazing comeback, which can be one of the most exciting things about a game (your game has this one good.)

There are more things about making a better game, but these three are probably the main ones, and even major game developers forget about these points, making sucky games that seemed good on paper. Want a good example? Master Of Orion 3! It is poor on those 3 elements (especially no 2 and 3) and therefore has been voted one of the worst games ever by many people. It looked hell amazing on paper though.


It will be tough making interesting games that differ too much from the TA style since the spring engine seems remarkably inflexible for gamestyles FF, AA, XTA, SWTA are very similar in the gameplay elements, its just they look and feel so different. A game like WC3:FT would be your best bet for making cool and crazy games.

We could have a chat about this if you want, just drop me a PM.
User avatar
Neuralize
Posts: 876
Joined: 17 Aug 2004, 23:15

Post by Neuralize »

AI crashed right as I was about to beat it on Prime. Quitter.

The mod is cool, however, it needs an archer unit, something to counter the Knights at range. :P

You should play a bunch of different varients of this mod. Like with air and sea units :P
User avatar
krogothe
AI Developer
Posts: 1050
Joined: 14 Nov 2005, 17:07

Post by krogothe »

New version is better, it has potential...
heres some suggestions:
-scrap wreckage, it wrecks the game(pun intended) on maps with tight passages.
-Make the comm weapon less friendly-fire prone. ATM it kills his whole base (guided rocket machinegun?) (accurate anni laser?)
User avatar
Min3mat
Posts: 3455
Joined: 17 Nov 2004, 20:19

Post by Min3mat »

freaking awesome :twisted:
me + NoiZe just beat 2 allied KAIBlob
maybe give it a handicap ^-^
User avatar
krogothe
AI Developer
Posts: 1050
Joined: 14 Nov 2005, 17:07

Post by krogothe »

i found the source for it on sourcesafe and made a new ver...
try that before bragging again :D
User avatar
AF
AI Developer
Posts: 20687
Joined: 14 Sep 2004, 11:32

Post by AF »

Has anybody seen if it works with NTAI TestBuild 4?

If not then I'll do build 5
IMSabbel
Posts: 747
Joined: 30 Jul 2005, 13:29

Post by IMSabbel »

where can i find this blob-ai?
IMSabbel
Posts: 747
Joined: 30 Jul 2005, 13:29

Post by IMSabbel »

Wow, this is actually quite addicting and cool...

But i would really suggest to give the wrecks a VERY low health (like 5hp or something), because the 1000s of dead bodies being rendered increase the cpu load a lot (and kills pathfinding)
User avatar
krogothe
AI Developer
Posts: 1050
Joined: 14 Nov 2005, 17:07

Post by krogothe »

Okay ill post new AI ver in a couple mins
keep in mind its *very just slightly* harder than last time.
Blob KAI v0.2
http://www.fileuniverse.com/?p=showitem&ID=2171
mongus
Posts: 1463
Joined: 15 Apr 2005, 18:52

Post by mongus »

Alantai.
Buils 4 works, but it only got 5 const kbots, (arm and core=) and did nothing else. no crash.. and btw.. build 4 is broken eh? last time it just build stuff but never attacked.
User avatar
Argh
Posts: 10920
Joined: 21 Feb 2005, 03:38

Post by Argh »

First off, my apologies to everybody who responded with negative comments on the first version- yes, it was very crude, and not as fun as it could have been. Although we did still find it fun as a human-vs-human experience, we didn't really take into account the sheer chaos that fighting an exponential AI was going to cause :-)

Fun was not really the goal for the first version, however. I was literally walking a friend (and fellow game developer) through the process of making a simple TA mod, and we wanted to build something that current AIs could not address properly. We built the entire game design from idea to finished game in 5 hours... the refinements that went into the second version took me about 5 more hours (a lot of playtesting, tweaking, playtesting again- I am a huge fan of playtesting).

I did the second one to address some of the obvious weaknesses of the game design, tweaking a large number of areas, such as the turn-rates of the various units, build/maintain costs, the way that the weapons worked, etc- a lot of changes, big and small, to smooth things out, repair things that didn't have a purpose, etc.

We were very surprised that AAI didn't work- of all of the current designs, it was the one I expected to "figure things out". But now that we have something that does work, we'll proceed to add new gameplay elements.

Fixing the corpses sticking around will not be a problem... and I agree with Krogothe, that really hampers the AI's ability to give humans a good game.

Version 0.3 will feature at least one more unit. Will that break the current build of KAIblob?

And has anybody played with the Wolf unit yet? I'd like to hear feedback on the self destruct... I don't think that the AI should use that, personally, but I want to know if it's too egregious to even let the humans have.

Anyhow, thanks for the comments, both good and bad. Next time I build one of these, I will probably PM the AI devs and give them a 24-hour sneek-peek and chance to revise their code, so that I can beta-test it better and provide a less crude initial experience for players.
Chocapic
Posts: 556
Joined: 16 Oct 2005, 03:35

Post by Chocapic »

We were very surprised that AAI didn't work- of all of the current designs, it was the one I expected to "figure things out". But now that we have something that does work, we'll proceed to add new gameplay elements.
lol ?
you are changing the all the concept of the game, in where a normal RTS ai builds resources for further buildings structures or units, you changed it all around.
how do you expect the ai's to work ? they cant!
unless we are making RTS/RPG/FPS ai's.
Then yes they would work fine[/quote]
Post Reply

Return to “AI”