Originally Posted by scottymoore2010

i like this idea and i support it all the way. like what someone said befor when multiplayer isnt working instead of just fighting uke motionless, you fight a uke that skills increase every fight. and yes it would teach new players how to run and comprehend new moves. but i think the hard part would be the cutscenes.

Originally Posted by vBulletv

Question to all of you: why the hell do you keep posting on this old ass thread!? It is like year old, which probably means it is NOT going to be put in-game, so stop trying and wasting space.

Originally Posted by Dafe

I stopped dev because Hampa ran out of funds for it or something, but it was pretty much usable. I've always regretted not releasing it though, so I might just open the source up on GitHub or something. It was pretty ambitious. I wanted to make it so people could compose their own bots via a web interface, and I'd written a PHP replay parser and MySQL analytics to select openers.

I don't remember what War was working on, or if he was just helping test or something, but he didn't describe it very well.

It worked via a bunch of modules that worked as a series of sieves, so at the beginning of the turn, the first module could get first crack at setting joints, and whatever it didn't set was available for the rest of the modules to adjust as secondary (tertiary, etc.) priorities. You could change the order of the modules to change the general behavior of the AI, to play more safe, or be more aggressive, etc. Every module took an string input of joint states, and returned the same, so it was like some functional chainable monad crap, before I knew what that stuff was, and before functional programming was cool. It was actually inspired by the Final Fantasy 12 gambit system. Fully composed bots were also subclasses of the modules, so you could use someone else's bot as a module in your bot.

So you might have a tripod module that detects threat of DQ, and sets joints appropriately depending on orientation (detection courtesy of my spherical region lib) to set up a stable tripod, and then other joints to attack, or switch the order to be more aggressive and save from DQ if possible with what's left over.

This is a replay testing the tripod module (Uke's movements are all AI, and I'm just trying to destabilize him):
http://toribash.mydistraction.net/AI/balanceDemo7.rpl

This is a demo of a composed wushu bot, which had a module with a randomizer for hand picked wushu openers, and the tripod module. I don't remember if there was more to it than that.
http://toribash.mydistraction.net/wushuBotDemo.rpl

These are replays testing the attacking modules. Just a right arm aim module, focused on the head (Angery is the account I created to test the AI online):
http://toribash.mydistraction.net/botDecap.rpl

A right leg aim module, and a power enhancing module (e.g. if the aim module says to contract right pecs, the power boost module would also left rotate torso, etc.), targeting the head.
http://toribash.mydistraction.net/botKicks.rpl


Personally, I think GA bots are dead ends, unless you can give it the right parameters to look at (spherical coordinate positioning of points of interest is a good start).

The aim modules I used were fairly simple, conceptually. You define a 3D spherical fragment (like a half sphere) region that's fixed relative to a joint, like a shoulder pad that moves with your shoulder as you move your arm. And if the target appears inside that shoulder pad region, you raise the shoulder. Then you have a corresponding region on the underside of the shoulder that triggers a lower shoulder, and repeat for other joints, until you get fairly fine control like in that decap replay.

I don't remember anything about Lua, so if I get this going again, it's going to take a while. I wrote this crap in 2008.
-----


This is possible if you look at a body part's angle relative to attached body parts. Sorry about necro. I haven't looked at the forums in a long time. I'd thought about making a script that let you hotkey stances, but since you spend so much time floating around in Toribash, it didn't seem worth the time.