58 réponses à ce sujet

[Drasca]

It doesn't make sense to engage a group of enemies only to have every single one decide to run down the hall and around the corner at someone else who wasn't in line of sight or attacking them in any way.

 

The next time you open a room full of spiders, stop attacking and watch what happens. They'll bee line toward one person until interrupted. Most enemy groups have mixed parties, each preferring different players, or the players aggro before the enemies arrive within attack range, and the enemies change targets after their attack is completed.

 

If you play enough with the same group, you'll notice how stalkers might always go for the same dude, over and over (if not intercepted), or the Demon Commander always preferring Archers first... and most archers choose to shoot him, to confirm the damage aggro level. If they have the discipline not to shoot him though, aggro from DC can be pulled. If not, aggro will not be pulled without taunt (or Archer death).

 

You can also play stupid aggro trick games with the enemy, if you have a player aggro'ed on them and luring/kiting them from a distance (preferably a higher form, like damage or taunt), and it is predictable. You can literally bodyblock an enemy aggroed on someone else, but they won't attack you, but instead get frustrated trying to repath toward their target.. because they're outside the distance where they can make decisions (and change decisions to attack you instead).

[DrKilledbyDeath]

Sorry, not sure I follow. Is what based off empiric or superstition? The agro system on other games?

[Drasca]

Is this based on empiric evidence (like viewing the tables), or superstition (like "more dots")?

 

You can test my claims yourself if you have a few players willing. All it takes is having more than one class, and letting the routine enemies run to smack you (since most of them can't hurt you much if you've got higher armor), then observing who attacks who.

 

You've mixed up terms. Empirical evidence is done through testing. Viewing tables is data-mining.

[DrKilledbyDeath]

The next time you open a room full of spiders, stop attacking and watch what happens. They'll bee line toward one person until interrupted. Most enemy groups have mixed parties, each preferring different players, or the players aggro before the enemies arrive within attack range, and the enemies change targets after their attack is completed.

 

If you play enough with the same group, you'll notice how stalkers might always go for the same dude, over and over (if not intercepted), or the Demon Commander always preferring Archers first... and most archers choose to shoot him, to confirm the damage aggro level. If they have the discipline not to shoot him though, aggro from DC can be pulled. If not, aggro will not be pulled without taunt (or Archer death).

 

You can also play stupid aggro trick games with the enemy, if you have a player aggro'ed on them and luring/kiting them from a distance (preferably a higher form, like damage or taunt), and it is predictable. You can literally bodyblock an enemy aggroed on someone else, but they won't attack you, but instead get frustrated trying to repath toward their target.. because they're outside the distance where they can make decisions (and change decisions to attack you instead).

The problem with this is I usually pug, so playing with all sorts of random classes. And I usually don't play a class more than a few games in a row so I wouldn't notice any of this. It's the MP side of a SP franchise, just go with the status quo on your first try.

[Drasca]

It makes sense for real life players to beeline for high DPS NPC's, so why not have NPC's do the same? I don't think the AI in this game is particularly intelligent, but we have to admit there are unique strategic advantages and unpredictability to some of the quirky aspects of this game's Artificial Retardation (Bioware's innovative variation of Artificial Intelligence).

 

It really isn't intelligent and not meant to be. Each follows a fairly simple script in order to minimize load on the hardware. Good game AI is designed to lose to the player, but in stylish ways. BW's actually done an admirable job in this respect.

 

However, most people don't understand the aggro system, or don't care to, yet make ludicrous claims despite that willful ignorance.

[Drasca]

The problem with this is I usually pug, so playing with all sorts of random classes. And I usually don't play a class more than a few games in a row so I wouldn't notice any of this. It's the MP side of a SP franchise, just go with the status quo on your first try.

 

I'm not sure if the aggro system in SP can be tested in the same way (or be considered valid), but I don't see why it wouldn't work the same way. Just drop the game down to 'easiest' and watch what happens.

 

I'm not surprised people wouldn't notice, for the very reason you mentioned and those I've mentioned. I've noticed though, after being informed of the data-mined threat tables.

[capn233]

I'm not surprised people wouldn't notice, for the very reason you mentioned and those I've mentioned. I've noticed though, after being informed of the data-mined threat tables.

 

Could you post a link to information about the tables?

[Drasca]

Could you post a link to information about the tables?

 

Sadly I don't have them. I was informed of their existence, and had to test it out myself the hard way.

 

Edit: I don't have data on all the kits relative threat table, just the most common ones. Generally expect your archer and keeper to be top of the list. There are oddballs among them, like Revenants love Arcane Warriors, and Spiders love necromancers.

 

Getting the combination for everything without data mining is too much work for me alone though, and I am not saavy enough to data mine the threat tables.

[DrKilledbyDeath]

Pretty sure Drasca is one of the only people who actively tests these things, and also posts here about it. I would generally assume he's checked anything out that he comments on.

[LearnedHand]

I just want to confirm the basic agro principles Drasca is laying out. We've all had those games where you feel like you are kicking ass with impunity and some poor sop seems to keep getting hit and dying. Other games it feels like you have a bulls-eye on your back and you are constantly fighting to stay alive. Chances are its not about your skill or lack thereof but about party makeup and pre-set agro distribution. 

[Drasca]

Chances are its not about your skill or lack thereof but about party makeup and pre-set agro distribution. 

 

Just keep in mind it doesn't have to stay that way, either lack of skill, party composition, or aggro distribution. The game isn't random, though humans can be (pugs especially)

[capn233]

Sadly I don't have them. I was informed of their existence, and had to test it out myself the hard way.

 

Edit: I don't have data on all the kits relative threat table, just the most common ones. Generally expect your archer and keeper to be top of the list. There are oddballs among them, like Revenants love Arcane Warriors, and Spiders love necromancers.

 

Getting the combination for everything without data mining is too much work for me alone though, and I am not saavy enough to data mine the threat tables.

 

Yes the jump from observation and data to conclusion is what I am wondering about, this would be settled definitively if someone had a script viewer for this game and could just take a look.  Not meaning to sound argumentative or anything, but jumping over here from ME3 is a bit of an adjustment since there are pages of data and math for most everything, or I can hop on the script viewer and look for myself.

 

I am mainly wondering why they would have set it up to target specific classes, which makes me wonder if it is actually looking at something else, or if it is a phenomenon that results from a bug, and isn't actually intended.  Not that it is the same team, but after looking at ME3 where it seems int and floats have about a coin flip's chance of being used correctly I wonder if similar issues aren't present here.

 

The little experiment on routine does seem interesting, I might try and convince some buddies to fool around with that.

[Drasca]

jumping over here from ME3 is a bit of an adjustment since there are pages of data and math for most everything, or I can hop on the script viewer and look for myself.

 

Yep, insufficient data and math really sucks. I end up having to correct people's bad arguements & conclusions, terrible math analysis and poor data (and gathering techniques) quite a bit.

 

If you find tools to data mine, please do share the tools & results.

[LearnedHand]

Having differential threat levels based on class is intuitive. Some classes have high dps and are squishy, why wouldn't you design AI behavior around that? 

[DrKilledbyDeath]

Because it makes the MP more complicated than it needs to be. No offense to those who are way up on leaderboards, but I don't get how they can play this game for so long. I play 2-4 matches and need to do something else. WoW, Everquest, FF14, DC Universe, those are games I'm willing to learn more about enemy agro, most effective DPS, etc. You can play those types of games for years and not become mind-numbingly bored  Why they think your average gamer would ever want to go that in depth for such a subpar game (in relation to other multiplayer games) is beyond me.

[capn233]

Yep, insufficient data and math really sucks. I end up having to correct people's bad arguements & conclusions, terrible math analysis and poor data (and gathering techniques) quite a bit.

 

If you find tools to data mine, please do share the tools & results.

 

ME3 was on Unreal, so from that standpoint it was a different ballgame.  And to be fair I am not sure if anyone was actually looking at scripts before about a year after release.

 

 

Because it makes the MP more complicated than it needs to be.

 

Yes, it makes the logic more convoluted.  Need more data on this, it would be one thing if it was more simple like a ranged / melee dichotomy for some things (like Revi pull), but critters preferring Necros is a bizarre design choice if it is indeed coded like that.

[DrKilledbyDeath]

Treasure room guardians serve as a great example. Go into a room alone, pop the chest, horror or revenant shows up looks at you then turns around and heads out the door across the map to fight your teammates. Line of sight should trump any other factor.

[Drasca]

Treasure room guardians serve as a great example. Go into a room alone, pop the chest, horror or revenant shows up looks at you then turns around and heads out the door across the map to fight your teammates. Line of sight should trump any other factor.

 

It is LoS though, but you're close enough that you're considered a group. It is group based LoS. He sees you as a group and targets after that. On the rare occasion that the players are very far from each other, and a new player pops in from lobby that out of line of sight, then they don't have aggro. That is very rare though.

[Drasca]

 Why they think your average gamer would ever want to go that in depth for such a subpar game (in relation to other multiplayer games) is beyond me.

 

They don'jt, because average gamers never do. Average gamers are dumb.

[LearnedHand]

Mastery of any combat game involves learning AI behavior or human behavior. The whole demons souls/darks souls games are all about this. If you didn't have more complicated AI behavior, the limited nature of this game would be even more one dimensional. You wouldn't need taunt abilities if it were simply unchanging LoS agro.

[Jebbadiah Jenkins]

If you really want to get at look at how bad archer aggro is, then just play 4 matches as an archer with the same group, then 4 matches with a hunter with the same group.

 

It's astounding how much difference there is.

 

It's almost like I'm invisible on my hunter. I can literally take a target within an inch of his life before he turns around to look at me. The same target on my archer would beeline for me from halfway across the map, with just a simple auto attack.

 

The passive I wish had a bigger impact on the game is Biggest Threat. This passive, just doesn't do a whole lot, with the way aggro is a bit wonky in this game. I can be on my Lego, and run into a pack of enemies, and they just scurry right past me sometimes, like I don't even exist. 

[capn233]

 It's almost like I'm invisible on my hunter. I can literally take a target within an inch of his life before he turns around to look at me. The same target on my archer would beeline for me from halfway across the map, with just a simple auto attack.

 

The Hunter is invisible though.

 

But seriously, if you were to spec out low level Hunter and Archer both beelining to Easy To Miss, does one actually draw more aggro than the other?  Of course the trees are arranged a bit differently and I don't know what the difference in auto attack damage is going to be.

 

Long Shot does seem to ****** everybody off, but that is because it hits a bunch of units for high damage.

[Gya]

I love it when there's a lolshot spamming archer in my katari games, keeps the heat off me. Kill all the things -> rez archer -> repeat.

[DrKilledbyDeath]

I love it when there's a lolshot spamming archer in my katari games, keeps the heat off me. Kill all the things -> rez archer -> repeat.

Those archers need to keep their distance. I try to stay at the longest possible range where I can still hit them. My vibrating controller tells me longshot can carry halfway across the map so may as well exploit the range!

[akots1]

ME3 was on Unreal, so from that standpoint it was a different ballgame.  And to be fair I am not sure if anyone was actually looking at scripts before about a year after release..

That was because game was supported and there was no way a DLC bypass loader would work properly to allow testing of changing variables as these were enforced to you if you want to play multiplayer together with balance changes. At least, it could have contributed. Coalesced edits were straightforward though and quite simple so anyone could do that. Also, Mgamerz is still digging through and figured out how to mod a number of things only recently. Apparently, DAI is more complex. Some guy mentioned that to change a color of spirit blade, he had to edit six files including some textures.

 

There are a few configuration files in DAI. I don't think there is a single coalesced.

 

The only place that has some info is nexus and the only thing people seriously bothered to mess up there is cheat engine for single player. The Mod Manager has very limited functionality also and that hampers the progress. Also, it seems that very few people actually care. The game is not as popular as ME3 MP, obviously.

 

It seems that DragonAgeInquisition.par file can be opened, edited, and even saved with ME3 par editor. Not sure how this is relevant to current subject of this thread though.