Forums » General Pantheon Discussion

How open is your mind?

    • 9115 posts
    March 8, 2018 3:59 AM PST

    If you could add a new type of AI to an MMORPG (Pantheon) what would you add and how would it work? #PRF #MMORPG #MMO #communitymatters

    • 23 posts
    March 8, 2018 4:05 AM PST

    "Living-like" AI so that NPCs look more like real characters than just NPCs

    • 1315 posts
    March 8, 2018 4:19 AM PST

    Code Name:  The Bloody Tumbler

    Mechanics: A rogue like mob with positional attacks can have The Bloody Tumbler attached.  A Bloody Tumbler will check and see if it has an unCCed ally.  If it does it then checks to see if any PC has The Bloody Tumbler targeted.  If no PCs are targeting it The Bloody Tumbler will target the target of his unCCed ally and move to the optimum location for its positional attacks vs that PC.  The movement it takes is a combination of run and tumble to get into position and may include a special attack in passing.


    This post was edited by Trasak at March 8, 2018 4:20 AM PST
    • 1 posts
    March 8, 2018 4:25 AM PST

    I would have to go with and AI that is given a reason for being where it is and goals.

    Like this group lives here it is their home they protect it.

    This group is moving through the area they will fight till they start to lose and then take off. 

    This group has a strong leader and will fight hard for that leader. But if that leader is killed...not so much fighting and more running. 

    This group is a hunter gather group and your in their territory. So your game. 

    Some of these groups could be reasoned with rather then just killed outright. 

    Maybe group 1 will give you a quest to grant you safe passage. 

    Group 2 might owe the character/group big if they helped them settle in a new area. 

    Group 3 maybe you become the new leader if you challenge the leader. 

    So dynamic ways to interact with the AI besides I just walk up and smash the first mob. 

    • 1315 posts
    March 8, 2018 4:36 AM PST

    Code Name:  The Mocking Miscreant

    Mechanics: TMM does not have any combat advantage and is usually placed on the least dangerous mob in a group.  Unless stunned, asleep, charmed, frozen or pacified in some other way TMM will put out a veritable stream of mocking commentary on the active combat it is part of.

    Every third or so miss on any of his allies by any PC will get called out with a few variations of “%t couldn’t hit the broad side of a barn, maybe we should wear bullseyes”

    When a player takes a hit that is 10% of their base HP he says, “I bet that hurt %t. You going to run home to Mommy?”

    When a player casts a spell that puts them under 25% mana he says, “This is so tiring, I bet you need a nap %t. Then you won’t look so pathetic.”

    When a player fizzles a spell or has it total resisted, “%T, I see you came ill equipped to a battle of wits.”

     

    There are lots of more word choices and triggers.  Too many triggers and there will be constant word spam but that is a little the point.  Players will actively become annoyed by TMM and will target it earlier than its danger would dictate simply to shut it up.

    • 724 posts
    March 8, 2018 4:57 AM PST

    Trasak said:

    Code Name:  The Mocking Miscreant

    Love this one, Trasak! :)

    I'd suggest: "Catch me if you can"

    Probably applies best to timid animals. An NPC of this disposition will (try to) never let a PC get closer to it than a certain distance, if the PC gets closer the NPC will move away in a random direction, gradually getting faster the closer the PC comes.

    • 43 posts
    March 8, 2018 5:03 AM PST

    Honestly, I'd be happy if you found NPC's everywhere.  As in: Bob the Bob is usually found by the NW corner of the bazzar in Thronefast.  But at different times he wanders off to other places; the temple district, south side of town, walks into the tavern (maybe more frequently than his wife would like!), talks with the stable manager and other things like that.  Only it's not just Bob the Bob that does it.  

    • 1315 posts
    March 8, 2018 5:17 AM PST

    I've posted this idea in two other places but thinking about it this idea is really a complex form of Vendor AI and could be used also as an example of non combat environmental AI.  Devs and other readers who have already read it feel free to skim by as nothing has really been changed.  *edit* The only secondary idea would be that rather than an "all commodities" vendor this AI could be for a select commodity on a single vendor but within a city a vendor for each of the commodities will be present. *edit*

    Code Name:  Commodities Exchange Vendor

    Purpose:  Localized commodity vendor who’s prices reflect actual supply and demand and operate at a net negative economy cash flow.

    Game Mechanics:  Commodities are defined as base raw materials that have not been processed in any way other than the initial harvesting process and are used in the crafting system. This would include iron ore, coal, spider silk, deer hides, clover leaves, cod fish, pork bellies, cats eye agate, and yellow clay.  This would not include odd ball ingredients like unicorn horns, Uncle Bob’s Peg Leg, or a dragons tooth, these items would typically be no drop or individually valuable enough to be on a merchant or sold player to player.

    Each local commodity exchange would have a list of all the commodities traded and a target amount of inventory and a maximum inventory for each commodity.  The base amount offered by the exchange for the commodities is the base indifferent, average charisma vendor sale price.  Each exchange will have its own target inventory.  This target inventory will be based on how often a commodity was purchased in the last week and a function of how far away the commodity is produced (the further away the commodity is produced the higher the target inventory).  The maximum inventory for each commodity will be twice the target inventory.

    Now the economics/math heavy bit.  Every 12 hours the average inventory levels are checked vs the target inventory levels.  The vender purchase price will increase or decrease at a percentage of the base vendor price inversely proportional to the average percentage of the target inventory. Every day that the average inventory remained below 25% of target inventory vendor offered price will go up an additional base cost times the number of consecutive days that the average inventory has remained below 25% of the target inventory.  This will force the vendor price up rapidly until it matches the player market value in that local area. If the current inventory is above the target inventory then the 12 hour value of each sale is decreased by the current percentage the inventory is over the target inventory.

    The commodity sell price will be one of two prices.  If the current inventory is less than the target inventory then the current sale price is 10% more than the price last paid by the commodity market for that commodity.  If the inventory is currently over the target inventory then the sale price is the average of the price the commodity exchange paid for the commodities currently in inventory, alternating removing the highest and lowest value from the history to maintain the average while removing the outliers.

    The commodity market will only do business in full stacks of commodities not in partials.  The target inventory will be 1/14th of the total number of purchases over the previous week with a minimum set by the initial inventory targets for that local exchange.

    Possible incentives: This will open an entire play style based on moving commodities from one local exchange to another local exchange as well as offer a ready market to both buy and sell harvested materials for crafting or harvesting focused players.  The system is guaranteed to reduce the net amount of cash in the game economy though an individual can move money from lazy people to themselves.  The exchanges will also make low level crafting cheaper as it will increase the availability of low level materials but will likely increase the costs of high level materials as they are easier to liquidate.


    This post was edited by Trasak at March 8, 2018 5:31 AM PST
    • 1399 posts
    March 8, 2018 5:38 AM PST

    I thought a thread like this would be good when listening to the last live stream with Kilsin and Zip and Zip mentioning that "programming was way ahead of modeling"... my thought's "we need to give Zip something to do!"

     

    Barricader came to mind.

    Think of a room in a dungeon, or a shack in the woods, with an open door. NPC wandering around out side of it, farming, harvesting, hunting, sitting on the porch in a rocking chair, whatever.

    As the adventurers approach and agro him he heads to the door goes in closes and lock it. he is out of reach.

    after X-amount of time, he cautiously opens the door and peeks out then goes back to whatever he was doing.

    it would take a root or a snare at the least to counter this behavior and he will not fight unless he has too (for example rooted)  

     

     

    Now EXPAND on this (or a whole new behavior of partially blind) with changing up the agro trigger (is there a thread on this yet?) commonly agro is based on distance, with a line of sight modifier (can't see through walls) what if Movement was all this mob could see. You could be standing right next to him and nothing, But if he saw you MOVE he would agro and start heading to his room. This could happen WELL out of spell casting range IF he saw you move, he barricades. If his back was too you, you could move in closer, or you could approach from behind the building.  But as soon as you started to cast Root....

    Think the kids game Red Light Green Light

    • 69 posts
    March 8, 2018 5:55 AM PST
    The Chatterbox

    A mob that talks smack customized to particular players in the fight. It will mention things like your class, gear, race, etc.

    The Hater

    This guy hates a particular player in a group and will maintain aggro on that player no matter what. Maybe have vocal cues like with Chatterbox to indentify a Hater.

    The Coward

    This guy is much more likely to flee and hide, will return only if it looks safe.
    • 319 posts
    March 8, 2018 6:00 AM PST

    The one thing i think is most important to me would be an AI that allows mobs to assess the size of the party atacking it and adjust it's damage and hit points to the crowd. So if you have a solo pc attempting a mob then his strenghth is equal to a solo player. Not equal to the layer attacking him but equal to  a solo player of his level. So if a mobs is attacked by a solo he has a chance to defeat him dependent on the levels of both. If a level 15 pc attacks a red con he will get wiped as he should be. But if a level 15 attackes a white con it would be an even fight as it should be.

     But if a group of level 15's attack this same mob the mob gets a bonus to attack and defense as well as a bonus to hit points according to the number af players attacking it.

     The second thing I would like to see in AI is if you are a level 20 player you will not be pestered every step by level 5 mobs. If you do attack a less than grey mob you will never get experience but you should not get any coins or drops if  the mob is so low.  If you are looking to get bat wings or bones for spell ingredients than buy them from a lowbe who can surely use the coin.

    • 1315 posts
    March 8, 2018 6:10 AM PST

    Code Name:  The Balcony Bandit

    Mechanics: The Balcony Bandit loves to push players off ledges.  The Balcony Bandit will be equipped with a combat charge ability with a knockback effect.  During a battle it will periodically check to see if any player on its agro table is within 2 knockback distances of a Z drop in a straight line from the Bandit to the player. 

    Every time that the BB checks for targets, say once every 30 seconds, it picks one target that has at least 50% of the highest agro that is within a straight line distance of 2 pushback lengths from a ledge and charges.  The charge does minimal damage but pushes the player backward in a straight line and knocks it to the ground.  The player must then stand back up before continuing combat.  The obvious danger is that if you do not move to more than the knockback distance from an edge the Balcony Bandit can and will knock you off the ledge, possibly to your death.

    After the push back charge the mob returns to its original target.

    Just imagine this on a mean goat half way up a cliff climb.

    • 2756 posts
    March 8, 2018 6:15 AM PST

    If anyone didn't see the recent Voices of Terminus chat including Tod 'Zippyzee' Curtis, a programmer with VR and an AI specialist, I suggest going and watching it.  Some great stuff in there, thanks guys: -

    https://www.youtube.com/watch?v=OAa8STNzQes

    As for the question: I do understand Tod's views on not 'over-engineering' encounters, Well designed encounters shouldn't 'need' to learn how to get 'better'. There's 'need' and there's 'want' though. Just because an encounter works doesn't mean dynamics elements wouldn't make it better. Sometimes the monsters could even change in a way that means the encounter is easier.  Everyone has an 'off day'.

    I also understand the view that having encounters that change too much may frustrate those that like to 'learn' how things, especially like raids, 'work' in order to repeat and beat them, BUT I would still like to see more learning and dynamism in those situations.

    I think 'knowing' how monsters will react and how encounters, especially raids, will happen is one of those 'old skool' features/mechanics that can be left in the past or at least seriously changed.

    It would be *much* better to have encounters of raids play out differently, at least have monsters pick different tactics depending on current group make up and/or historic data on how those encounters have been going.

    Sure, some wouldn't like it. Some would prefer the old way. Some always do. Don't be afraid VR! Innovate! Please!

    Examples: -

    If an orc village gets wiped out every hour for a week, have them call in the wandering orcs and stand closer together for mutual protection.

    If a red dragon regularly comes up against parties that are totally fire resistant, have him hire a mage that can nullify the attackers' resistences (that will need to be killed quickly).

    The AI can be as subtle or as bold as you like, but *some* change will give things a totally different, more dangerous, more exciting feel, even if most times it doesn't kick in, knowing it *could* has an immense effect.

     

    • 1921 posts
    March 8, 2018 6:41 AM PST

    NPC Lieutenants that would use a dimensional-door consumable to summon adds that would raise the encounter difficulty to that set by the PC group leader.

    • 3852 posts
    March 8, 2018 6:44 AM PST

    I can't think of anything enormously creative so I'll just endorse what Disposalist said.

    One of the problems MMOs have is keeping people interested after they hit maximum level and do the dungeons/raids a few times. I could elaborate but I think we all understand this dynamic.

    Creating new raids and dungeons takes time and has obvious downsides - constantly adding even better gear is worse. 

    Having the existing raids/dungeons work differently each time would be a really nice way to keep people interested for a longer time.

    • 626 posts
    March 8, 2018 6:52 AM PST

    Uncontrollable - AI 

    Upon being CC'd or Rooted this Mob would go into a Rage or Trigger the Bersek dispostion. Trying to controll this mob type would result in a fight you won't forget. 

    Counters? - Memory Wipe would be required to cause this mob to forget being bound or controlled and revert its Rage to normal. 

     

     

    Claustrophobic - AI

    When engaged if Mob has more than 3 players within 10m it will go into a Panic. This would trigger one of two reactions. 1. Being the alarmist reaction. 2. Being the enrage reaction. Fight or Flight - He will run for help, and decide to Fight it out in a heat of rage and fear. 

    Counters? - Don't stand so close. :)

    • 27 posts
    March 8, 2018 7:04 AM PST

    I think it would be fun to have a disposition that is more quick to target healers.  I'd say test it with no more than one per group of mobs at first.

    When you see that mob switch to the healer faster than expected, you can either CC it to get it later, or have all DPS focus it down if you don't have extra CC to spare.

    • 411 posts
    March 8, 2018 7:44 AM PST

    I'm not sure what exactly defines a "type" of AI. If we're looking large scale, then I would like to see an adaptive population system or an economic supply/demand system.

    Adaptive populations: If you have a dungeon (or any population of NPCs) that are being ignored, then they start to venture out (I think Zippy is already doing this part). On the flip side, if a specific camp is being farmed on a regular basis, then additional patrols or spawns are created and sent to that area until it is secured. Eventually (over 6-12 hours) any group staying in a specific area will be overwhelmed and will have to leave the camp or die. If the entire dungeon is jam packed with players, then the number of spawns and patrols will start growing large, which would create a hectic and dangerous environment. Maybe the coin they carry also gets reduced in order to pay for their hired help. Another variant would be that if a specific area is being camped, then instead of increasing spawn rates to take back that area, those spawns are just shifted to other areas of the dungeon.

    If we're talking about dispositions, then I think some of the suggestions here have been good. Also perhaps...
    Strength in numbers: Enemies want to run if they have less than 3 allies in combat with them. A whole population (of gnolls for instance) of this enemy type would be challenging and fun to deal with.

    • 1315 posts
    March 8, 2018 8:08 AM PST

    Reposting an idea from the named mob spawn thread.  This idea kinda ties into both Nightsongs' and Ainadaks idea and is more of a non combat behavior structure than an in combat AI disposition.

    Code Name: NPC life cycles and variable redeployment spawn points.

    I’ve always liked the idea of a combination redeployment points and life cycles over static spawn points and place holders.  Redeployment points are locations in a zone where creatures will spawn and emerge from.  In a standard dungeon there may be as many as 20 redeployment points.  Meta game these redeployment points will become prime camp locations.

    Life cycles are exactly what they sound like.  While spawned a mob will move from one point to another, often taking long rests to do an action while at the point.  The flow is intelligent for that specific mob and possibly influenced by factors caused by players.  Each life cycle will have a couple of options of redeployment points when respawning. 

    Once the mob spawns it walks, unless its life cycle includes running, to the nearest point in its life cycle and does a task which could just be a 90 second wait cycle.  After the cycle is completed it moves onto its second location in its life cycle and does another action, or inaction, cycle and moves on.  Add as many cycles as make sense for that mob.

    In an orc fortress as mentioned earlier I can think of a few standard life cycles and a few dynamic life cycles.  A simple guard life cycle would spawn somewhere in the guard barracks area and stand around getting ready for the day running some complaint scripts, he would then move onto the mess hall and complain about the food.  After that he could walk out to his duty station to relieve the guard on duty.  The guard would wait there on duty until the other guard completed its entire mobile life cycle and came back to replace him.  An added bonus this could be a pair of orc guard’s who’s scripting is all based off each other and are linked spawns that cannot be split by character pulling methods.

    The dynamic version of this life cycle would be that if more than 50% of the guards are dead and there are PCs in the fortress then the Orcs go on alert and move in groups of 4 rather than 2 and skip most of their life cycle to reinforce the walls.

    The Orc King could spawn in the royal bed chambers with his 2 body guards.  He could go from his bedroom to his dining hall.  From the dining hall he could move to his throne room and sit on the thrown yelling at people and issuing orders.  All of the mobs that are scripted to interact with him will have one life cycle if he is spawned and another if he is not.  From the throne room he can move onto his advisors then to the dining room and finally back to his bedroom.  Most of his life cycle would be based on time of day and his respawn would be game days later.

    There are lots of ways you can go with this but making fully fleshed zones with all mobs having life cycles will take a bit more time than just dumping in static spawn locations.  The pie in the sky would be for all named spawns to have some form of active trigger or accelerated respawn mechanic for those truly farming.

    Trasak


    This post was edited by Trasak at March 8, 2018 8:08 AM PST
    • 557 posts
    March 8, 2018 8:22 AM PST

    Rather than suggest yet another behaviour for Zippy's pluggable AI system, I'd like to look at the larger question Kilsin poses "add a new type of AI to an MMORPG".

    In past games, NPC essentially came in a few flavours:

    In the first group, I would put static NPCs like most merchants who never move unless aggroed and have an extremely limited response vocabulary.  You could add most roamer mobs to this list since they were essentially dumb like the static NPCs, but with pathing.

    In the second group, I would put the mobs from the first group who had an associated quest, so they were slightly more responsive on a limited set of keywords and might have a limited path they walked which included scripted dialogue so they appeared to be doing some chore or task and not merely roaming mindlessly.

    In the third group, I would put the boss and mini-boss mobs who had scripted events associated with combat cycles.  For the most part, these mobs were like the first two groups in that they were static and uninteresting until hailed or aggroed to trigger their scripts.

    In each of the above, referencing "AI" to describe their behaviour is rather a loose, inappropriate or at best generous use of the term.  If we look at Oxford's dictionary, "intelligence" gets defined as the ability to acquire and apply knowledge and skills.   "Artificial Intelligence" describes computer systems able to perform tasks normally requiring human intelligence, such as visual perception, speech recognition and decision-making.   In EQ, the NPCs exhibited none of these properties.  They were no closer to passing a Turing test than an electronic game of tic-tac-toe.   Arguably, we've never had AI in an MMORPG.

    Nor have we had convincing life in most MMORPGs if we use Oxford's definition where life "includes the capacity for growth, reproduction, functional activity and continual change preceding death".   Where life defined on a per mob basis hasn't really existed, Ultima Online attempted to create a sort of ecology.  The players destroyed it almost immediately at launch and continued to do so through several attempts to rebalance the system. 

    Since HAL 9000 is now 17 years overdue, we need to look for alternative ways to make our game worlds feel alive and filled with residents of varying intelligence.

    There are a few basic rules that I believe should be applied to NPC behaviours to make Terminus come to life.

    1.  No NPC should be completely stationary.  A merchant in a stall is still going to move around.  Maybe they are arranging merchandise or going through the motions of selling goods or occasionally walking to the cart behind their stall for more supplies...   They don't stand around like mannequins.

    2.  NPCs should interact with each other.  In town, this might be conversations as they move around and encounter other NPCs.  The range of dialogue could be anything from discussing the weather to giving hints to nearby players that are perhaps only heard if you have the right quest or perception ability.  Maybe one merchant is angry about taxes or a competitor.    Don't just give NPCs a name and a stall.  Give them a backstory, some jobs they do and pathing that relates to their tasks and backstory.   Do the same for mobs in camps.   Maybe one orc is making tea, while another roasts a halfling on a spit and two more fight over last night's loot.   If you want mobs to feel alive, you have to give them tasks and personalities and they can't all be clones (appearance and behaviour) of each other.   As they say, life has a purpose and variety is the spice of life.

    3.  NPCs should respond to natural speech to a greater degree than the extremely limited scripts of the past.   If I mention "weather", they should say something relevant.   This sort of rudimentary system has been around since the early AI research days of projects like ELIZA.   ELIZA is now 50 years old so we're looking for an upgrade.   The primitive ELIZA ran in BASIC and was even ported to the emacs editor, so implementing some sort of natural language processor in an MMO client given today's hardware should not be a huge challenge.   

    4.  NPC speech responses need not be limited to interactions with players.  If Farmer Brown hear's Farm Hand Bob mention "weather", he may respond accordingly, which may, in turn, cause him to say something where he mentions his cow, which makes his wife say something about their goat, which reminds Bob to go mend the fence.   Each NPC has a limited number of tasks and triggers, but to the player/observer, they're not constantly having the same conversations.   Each NPC has a limited number of triggers and certain events require multiple conditions, so we're not constantly hearing seeing the same events in the same order.   To the player, it feels as if they have witnessed a moment in time of life on the Brown farm.   We don't need huge variation in the vocabulary of each NPC, just enough to give them a few different ways to respond to events and a touch of randomness which prevents the same few NPCs from always being in the same place, doing the same things.   If only Bob and Farmer Brown were having the conversation, the fence never gets mended.  It may be a few hours before all three participants are in the farmhouse together and Bob mentions the weather.   Nor do NPCs need to be constantly speaking.  IRL, people who are in proximity to each other continuously often don't speak for long periods.   Just look at the average married couple.  :-)

    Up to this point, we're talking more about making the world appear to be alive and not really addressing AI.  If enough different predetermined/scripted systems are in play, events appear to take on a non-repeating life of their own.  This adds immeasurably to immersion for the players.  You feel like you are part of an ongoing story, not merely inserted in a low budget movie set.  Arguably, this is more important than AI.   If mobs interact with each other, we can also introduce a simple ecology where wolves breed based on proximity to each other and to a food source such as rabbits, sheep or halfling children.  The world is a lot more interesting when you can watch the wolves at feeding time.  Even Minecraft had a very rudimentary ecology system where the wolves ate sheep and animals reproduced.  

    6.  Conway's Game Of Life - If animal by animal species interaction is seen as overkill, another approach would be to use a cell-by-cell simulation where the zone could be broken into hexagons and survival/population stats for specific species could be calculated based on adjacent cells.  This could be applied to the wolves, rabbits and bird populations, but it has other, perhaps more interesting uses.  Assuming we have a city which is seeded with a sizeable human population, we have a group of cells which are well established and difficult to change.   Outside the city, several cells away, we have a small orc encampment.  The orcs move with the supply of deer, their primary food source when there are no halfling children to eat.  If the players kill all the wolves in the fields outside the human city, the deer population will increase.  The deer proximity triggers an expansion of the orcs to the field near the city.  The orcs are now closer to town and the town and additional orc cells start to interact.  If the orcs continue to expand into the fields (through cellular automation), they become a larger threat to the human city.   At this point, we have primitive AI driving world events.  The players unknowingly upset the world ecosystem, which prompted an orc invasion.  The players can beat back the orcs and in time, the wolf population automagically starts to regenerate, restoring order to the good residents of humantown.

    7. For us to start to talk about AI and boss mobs, they at least need to be responsive to stimuli.   If the players do X, then the NPC does Y.  If the players change strategy, then the NPC responds in kind.  The in-game INT of the NPC should determine the time it takes to notice the player strategy change and respond.   Defeating a dumb mob might be a case of having to constantly change your attack strategy.  For other mobs, it may be that you need to use one strategy for half the fight, then change so the mob won't use a particular ability.  Maybe in defensive mode, the mob uses early AOE and in an offensive mode he uses late silence abilities on your casters.  By pushing him into offensive for the first half, then defensive for the second half, you avoid both his AOE and his silence.   If all of the melees are behind the NPC to avoid his frontal AOE attacks, maybe he takes a few steps back to thwart their efforts.  You may need to root him and in his offensive mode, he's not rootable.   Of course, the NPCs would need to emote something that indicates they have change mode/disposition.   The critical component here is that the fight is not simply scripted where disposition X is applied at health level Y.  The NPC must respond to player events.

    8. For us to truly consider it AI, the mobs must not only respond to stimuli, they need to learn from previous encounters.  "The last few times the players backed me into a corner.  This time I'm going to try to avoid corners."  "If I stand still, they surround me.  If I constantly back up, their melee does less damage."   Part of this learning could be that over time the mob develops a particular instinct toward specific classes in the attacking party.  "Last time I took out their wizards first.  That worked well.  I'll do that again.  I'm building up a real hate for wizards."   For this to work, the NPC boss should be running his DPS parser, looking at buffs, heals, unattractive armour combinations or whatever parameters make boss mobs more likely to want to smush a particular class, race or player.   "That guy has killed me 9 times.  He's first to go tonight."

    So that's my thoughts on AI.  Is my mind "open"?

    • 945 posts
    March 8, 2018 8:25 AM PST

    -Allow some warrior NPCs the ability to "taunt" anyone that has taunted them within X seconds but is no longer targeting them specifically.   Could call it the Ego AI
    -Alternately, the same type of NPCs could taunt a DPS if they haven't been taunted themselves.  Come-at-me-bruh! AI
    (Allowing players to be affected by taunt mechanics could also be very effective in arena or pvp combat).

    -If a player has killed a named or powerful NPC of the same species (like a named Orc) in an area, Orcs in that area have a chance of getting an offensive buff when they encounter that player while in that area for X duration.  Vengeance AI

    -Could have intelligent humanoid (playable race) use signal arrows/colored smoke grenades (like flares) as part of the arlarmist AI.  The animation for readying a flaming arrow would have to be obvious and take some time so it could be countered the same as countering a spell.

     

     

     

    • 213 posts
    March 8, 2018 8:39 AM PST

      

    I'd give them moods and personality.

    By this I mean the Ai would have a list of things it likes and doesn't like at a particular time in game.  It could be random or on a time schedule depending on how the player base reacts.  

     

    Instead of stagnant NPC's who never move I'd make a list of character npcs carrying out daily tasks in each part of the city, near the water, in the pub, everywhere.  Some of them will be singing songs in the pub while waving a horn of ale around, some will be swimming in the water, some will be pounding out some Iron at the blacksmith, npc kids will be playing in the city... There would be an npc who appears to deliver mail... Either way they would all be DOING SOMETHING.  =)   

    Npc's would react to emotes in game..  Maybe a smile would get a smile in return... etc.  When a player makes 50 or completes a dungeon maybe the towns people react when said player enters the city with cheers.

    Npc's would react accordingly to weather in game.. so if it rains you will see them with umbrellas or they will move to a covered area for awhile.  If it snows you will see some shoveling it away or making snowmen.  

     

    Npc scheduled events in game ie.. Fashon shows, fishing tournaments, seasonal faires, holidays, secret events, XP events, racing, lottery...etc.

     

     

    Just some things I'd love to see.

     

    • 89 posts
    March 8, 2018 8:50 AM PST

    Adaptive and realistic camping group AI: Several enemy encampments are established forward of a base fortress

    Not only do the encampments establish guard cycles and patrols, but they have to maintain a supply line to the fortress as well as send out foraging parties

    Patrol groups would stick to their routes but would be a little more alert for cries for help

    Foraging groups would roam about looking for edibles and tracking game, but trackers would also notice when adventurers had recently passed through as well and follow

    Guards for the supply line would be heavily armored and escorting a wagon at intervals throughout the week or as needed

    If players kill the foraging parties or disrupt the supply line, the bad guys in the camp become weaker over time, but the fortress can also send supplies under heavy guard along with reinforcements

    Additionally, if a player party is killed by an enemy group, the surviving enemies receive a slight upgrade to their gear (along with reinforcements with standard gear to replace their fallen) and the leader of the group gets a buff ability that bolsters his veteran party in the next battle

    As long as one of that group of veteran bad guys continues to survive, any party he is with grows progressively stronger with every victory and his buff/debuff abilities increase to make his party ever more formidable

    This kind of rational and realistic group AI could do a lot more fun and complicated stuff, and it wouldn't be terribly difficult to keep track of, but it would diversify the challenge texture of an area set up to allow for it while providing more opportunities for player strategy and environment awareness

    • 2752 posts
    March 8, 2018 9:53 AM PST

    I'd like to see mobs get smarter in time and work toward countering players as best they can. If a group is in a dungeon camp then I'd like to see adaptive spawning where the mobs begin to respawn in the area with abilities that best counter (within their class/mob type repertoire) the abilities last used to kill them. 

    • 83 posts
    March 8, 2018 10:02 AM PST

    Maybe a 'Spiteful' behavior? Crowd-control effects give increased hate and the first damage and first crowd control on it since the last combat reset give a big, flat spike to hate?

    Maybe a 'Frenzy' behavior that reacts every time the top of it's aggro table changes, giving it a short-duration buff to crowd-control resistance, movement speed, attack speed, and/or attack damage. Better not take this one off the tank!

    A 'Territorial' behavior that has it refuse to leave a pre-set area, and if anything engages it at range from outside it's territory it will retreat or seek cover. Might be a good idea to have a band of 'soft territory' that it will enter if opponents retreat into it from the hard territory, but does not aggro people if they enter soft territory from outside the territory. A useful behavior for monsters that depend on a specific environment and flee if forced outside it (be it from being pulled out, knocked out, or fleeing from a Druid who's bringing their own environment with them.)

    An 'Ambush' behavior might include two different modes. An aggressive mode if it proximity aggros onto a player, and a defensive mode where it flees from a fight if it was aggroed from a distance by a player rather than proximity aggroed. Useful on a mob that attacks from stealth or cover that isn't confident about the fight it will take if it doesn't get the drop on it's target. Might include a desperation 'turn and fight' if it's movement speed is impaired while it's fleeing.