Ironic juxtaposition: the 30 year old text based Rouge has monsters that can follow you from room to room; these rooms are analogous to Andors Trail's individual maps.ncer wrote: In addition you would only have monsters move around on maps within quick reach of the player... There's absolutely no need to be thinking about the entire game world to simulate larger monster movement.
You could even place spawning points for this near the exit towards that other map so the monster actually appears to be coming from that map.
On the newer Baldur's Gate all you have to do is exit the map screen to ditch pursuing monsters.
I think freeing monster handling from the map portion of the code, so monsters are their own entity and not simply a map attribute, would open up options for game play.
As it is the game is biased heavily towards "first strike" juggernauts versus balanced offensive/defensive strategies because of a few monster behavior reasons:
- no following or pursuing by monster means that a PC can juke amongst the monsters attacking a each individually with minimal risk
- monsters are frozen during combat
- no combat advantage for monsters ganging up on PC. I proposed a simple adjustment for that over here.
With a behavior attribute table you can do creative things like:
- monsters that follow
- monsters that lurk timidly
- monsters that cluster in herds
- monsters that bumble around randomly (like now)
- monsters that will flee
- monsters that guard other monsters (by being aggro while following a fellow "guarded" monster that flees from the PC)
- monsters that don't attack until in a group (instinct for wolves or imagine being nibbled to death by cats)
Just a simple table with boolean behaviors (awake/asleep, avoid/follow, passive/aggressive, moving/standing still, others?) and maybe geographic settings defining a limited wandering range when not pursuing and a preferred grid location when not interacting.
With a sufficient set of basic behaviors (that need not be boolean) complex behaviors will arise just as with genetic algorithms.