Larian Banner: Baldur's Gate Patch 9
Previous Thread
Next Thread
Print Thread
#474297 01/10/13 07:32 AM
Joined: Apr 2013
A
arkand Offline OP
apprentice
OP Offline
apprentice
A
Joined: Apr 2013
I am not a professional game designer, thus why I am going to ask this question. Why do games not incorporate a learning machine for combat?

I was playing with the unity engine and built a learning machine to test this out within a small mock up I did and thought it was more fun than current combat systems in many games. Basically, I built a system that allowed the game to analysis pattern within the way I fought and the game changed and adapted to that what I did.

One thing that annoys me anymore and in many games now is that instead of increasing the ai of Mobs the developers increase the mitigation of the mob, the hit multiplier for larger hit and increase overall health of the mob. This is tedious and boring to me. If a mob couldn't kill me before and the increase in damage does not make a big difference the only thing that changes is the duration of the fight not the challenge, much like D3 and there monster powers. If a learning machine was used instead the ai could be tuned at given intervals to increase actual ai and challenge based upon how well the learning machine is built.

Examples, of what I did consisted of;

spell caster uses a flame spell the mob puts up a ward and counters with a lighting spell. If I used a ward myself the mobs could learn to start to use a magic debuff when a ward is used, then cast.

When in melee close range if I kept using a normal forward hack with a sword, the mob learned to parry and riposte. If I was able to bring a shield up to block the attack, the mob starts to learn the pattern and behavior I am using to fight and instead might parry, feign a riposte and instead take a step back and counter with a spell.

The combat become a strategic real time situation, instead of the boring systems many game incorporate of finding a certain build and spamming one constant ability over and over. Instead all skills become more relevant and players cannot use one over powered ability to curb stomp a game.

Joined: Dec 2006
old hand
Offline
old hand
Joined: Dec 2006
The simple answer is that learning systems are hard to test, and plan, and balance. But, that doesn't mean they're a bad idea, far from it. I think that if we'd have sufficient resources, we' probably try out a few things in the learning machines department, but alas, we don't have that luxury for the moment.

Joined: Jan 2009
veteran
Offline
veteran
Joined: Jan 2009
Still though, the reactive AI in Dragon Commander is really good at building counters to your force. Even if it's not a learning AI, it does a good job of faking it.

(Although I wish it would learn to not build a second Battle Forge at the start, because Hunters counter all of those units, making it too easy.)


Moderated by  Larian_QA, Lynn, Macbeth, Raze 

Link Copied to Clipboard
Powered by UBB.threads™ PHP Forum Software 7.7.5