Sensuki

The rationale has already been explained multiple times, just not officially in an update or whatever. Josh Sawyer worked on Neverwinter Nights 2, and he is a D&D player and plays a lot of tabletop and turn-based games that probably have AoOs and he is used to the concept of movement having an opportunity cost (which is almost never good in real-time games except for siege units like Trebuchets in AoE2 and Mortar Teams in Company of Heroes). It also doesn't even have to be the case in turn-based, people are just used to the cargo cult design of movement and non-movement actions sharing the same action points in turn-based games. also I think they were actually thinking about MMO aggro mechanics first, but since 'the backers' said that we didn't want it, we got the D&D style version instead. But AI Targeting and Crowd Control abilities are the way to do it properly in real-time without overriding player issued commands.

I don't have a selective memory I recorded a full Let's Play of Icewind Dale in October. https://www.youtube.com/playlist?list=PLIRfCmyR7ijJU11xQVn2SXIwmeuajTkom So? The Melee Engagement system actually gives you the means to do this without an ounce of effort, that's what it's for. The Melee Engagement system only prevents a very specific instance of kiting - when a melee unit actually reaches a ranged unit. All you have to do is avoid that situation and you can still kite the bejeebus out of anything. Yes, encounter design is a big part of fun combat. I could have done the same thing against any group of melee enemies where I had the chance to metagame the encounter and in PE the designers (unfortunately) have gone out of their way to make sure that the player doesn't get ambushed the moment they enter an area, which often occured in the IE games, so it's very easy to metagame encounters.

On the topic of Melee units not being able to control an area in the Infinity Engine games, I recorded this video the other day to demonstrate how to switch aggro in Baldur's Gate 2, since the AI is very simple. It's silly, but that's how you do it. I will do another one for Icewind Dale: Heart of Winter tomorrow or something. IWD:HoW has much smarter AI targeting. The premise that melee units couldn't be sticky in the Infinity Engine games is not true, it just requires understanding the AI targeting clauses.

or just remove it and spend that time on something more important

Here's my first video with it removed, I'll be making more of these, but if you play the same way as the game promotes currently in Pillars of Eternity, it doesn't play any differently.

Here is one of the videos I recorded on the subject of looking into kiting on Pillars of Eternity, it contains information on how the AI targeting works in Pillars of Eternity and demonstrates the normal way in which kiting occurs in these style of games.

In this video I have a party of three Rogues and a Druid. The Rogues are (naturally) equipped with Two Handed weapons. I use the Druid to cast an AoE spell on Medreth’s group and then abuse how the AI Targeting, pathfinding code and Melee Engagement system interacts to score no less than 7 disengagement attacks. The Rogues can only engage one enemy at a time and have normal engagement range. In this video I have (for maximum effect) three Fighters. I use a reach weapon to attack a House Harond Guard so that I do not provoke engagement from him, I then move away and lead him into the Engagement range of two nearby Fighters. When the House Harond Guard is hit with a disengagement attack, engagement ends, and the House Harond Guard is interrupted (which is like a ministun) this gives me the time to move my characters back so I can repeat the process. You will notice none of my characters make a single attempt to target the House Harond Guard. Obsidian has stated that they think the problem with Melee Engagement is visual feedback, well as these videos demonstrate, there are far more serious problems than that. Here are some of the components of either the Melee Engagement system or the game itself that make these abuses possible. Engagement Range – Engagement range is a circle around a character, even the default engagement circle size can be overlapped so that multiple characters can engage a moving enemy at the same time. This is magnified by things that increase the Engagement range, such as Fighter Defender Mode. Multiple Engagements – Enemies can be lured past units that qualify for multiple engagements (Fighters with Defender mode, and units that take the Hold the Line talent) for multiple disengagement attacks. I would also like to point out that there is a bug with Multiple Engagements at the moment where new Engagements are not added once an engaged target has been killed. To qualify for a new engagement you have to move which provokes disengagement attacks ... yeah. Attack Resolution system – I can ensure that I have very high accuracy for my disengagement attacks by maxing Perception, using items that give bonus accuracy and casting accuracy buffs with a short cast time (such as Dire Blessing) on my party before enemies get in range so that I score many critical hits on disengagement attacks Two Handed Weapons – Two Handed Weapons do the highest damage, and since disengagement attacks are free of recovery time, equipping two handed weapons ensures maximum damage on disengagement attacks Save Scumming – I can save a game and attempt this tactic multiple times until I get lots of critical hits on my disengagement attacks so that I kill the most enemies with them. AI Targeting system – There is no loop that constantly re-evaluates targets for enemies. There are a set of initial targeting clauses (which I explain in a video further down in the thread, in the kiting section) and there is target re-acqusition based on a set of conditions. You can make it so that enemies are ‘blind’ to the trap you have laid for them. The way AI targeting interacts with multiple engagements and engagement circles makes it possible to score multiple disengagement attacks on multiple opponents. Pathfinding – The pathfinding in Pillars of Eternity is not very good, the current code does not handle multiple enemies moving around in the same space very well and there is no loop that constantly checks for first shortest path to my knowledge. Unit pathfinding acts kind of the same as the AI, in that there is only correction made when a certain set of conditions are met. The way the pathfinding system interacts with multiple engagements and engagement circles makes it possible to score multiple disengagement attacks on multiple opponents. If you've gotten this far, you’re probably trying to think of solutions to this problem that would easily fix or alleviate the situation. The most obvious one is to limit the amount of disengagement attacks units can qualify for in a period of time. Give them a cooldown or something. You’re probably thinking something like one per engaged enemy every four seconds, or something like that. Yes, that would make it less bad, but it would not fix the abuse. I would still be able to get an initial round of disengagement attacks off on groups of enemies and it would still be the absolute best tactic to use at the start of combat. In the video where I have six Fighters with Defender Mode on, if I also had Hold the Line as a talent, that’s still up to three disengagement attacks per Fighter, and perhaps 3-5 Fighters could engage a single enemy running past the clump in Defender mode. With High Accuracy and Two Handed Weapons, the disengagement limit does absolutely nothing to prevent the carnage. Another obvious one is to improve the AI targeting of enemies (which Obsidian really need to do anyway). However to try and avoid the situations that are happening at present, AI would have to run a loop that constantly checks for the closest enemy. A Disengagement attack cannot be scored against a unit that moves straight up to a target and engages them. This is how some of the disengagement attacks are occuring, because enemies are not targeting the closest player controlled unit, but rather, the unit that opened combat. The pathfinding brushes them across the edge of other unit’s engagement circles and thus, disengagement attacks are scored. However while this would also contribute to less disengagement attacks in the three Barbarian or three Rogue scenario, it would still not prevent many of the situations from happening (such as 6 Fighters) and would do nothing to save the poor House Harond Guard from the disengagement attack kiting from multiple player units. There does need to be a loop that checks for new targets (also something that will be demonstrated in the kiting section), but always targeting the closest enemy isn't exactly smart AI , and this is one of the problems with the Melee Engagement system – it promotes simple AI targeting because of the fact that the “first enemy engaged by” targeting clause that is part of the Melee Engagement system would override most ‘smart’ AI targeting clauses for melee units in the first place and melee enemies always re-evaluating for the closest enemy would open up kiting abuse and also disliked by many of the players. Another conclusion you might have come to is that enemies might need to be able to path around Engagement circles or be able to determine when to cross engagement circles of player units. This raises a bunch of concerns such as how that would interact in different environments such as open spaces, and corridors ? How would it interact with dynamic environments (such as the player moving a unit left and right over and over again) ? How would it interact with engagement circle overlapping ? How would it interact with general pathfinding with multiple units ? It seems like a solution that would need very thorough research that is best undertaken during pre-production and with three to four months left on the project, this is not a solution that is even remotely viable to look into. And the last thing that you possibly might have thought of while watching the videos or reading the above solutions is that what if disengagement attacks were scored only when units left Engagement range, rather than it checking for any movement ? This is actually how the system used to work, back in BB v257 and v278 and it was actually during v278 that I discovered that the Melee Engagement system was abusable. I moved my BB Fighter next to a moving beetle and when he stopped, the beetle clipped the edge of his engagement circle and received a disengagement attack. I believe Obsidian changed how Melee Engagement worked from that system to this system because it was unclear when units would qualify for a disengagement attack and showing the engagement radius around units would add to the already cluttered visuals of the game. Doing it this way might make the system a bit less abusable, but not completely abuse free, and it would require extra UI elements most likely so I don’t think this is a path Obsidian is interested in pursuing. I believe that I have sufficiently proven that on top of the 2D art, programming and VFX/animation time that Obsidian plans to dedicate to making the Engagement UI clearer, it would require some serious brainstorming from design and programming to remove the exploits that I have found with the system. There are no doubt, avenues that Obsidian could take to make the exploiting harder to do or less bad, but I do not think they will be able to completely remove them without changing something about the fundamental design of the system. In my opinion, the easiest and best solution to this problem is simply to remove the Melee Engagement system from the game. I believe that both of the problems that the Melee Engagement system was created to address can be solved independently from the Melee Engagement system with things already present in the game – Melee Stickiness can be achieved through AI Targeting clauses and crowd control abilities, and after doing some research I have found that the specific case that of kiting that the Melee Engagement system prevents is actually not feasible even with Melee Engagement removed. What’s more I think that the Melee Engagement system is detrimental to the gameplay and removes many of the tactical elements from combat after the opening and contributes to making combat less fun than the Infinity Engine games. Some elements of tactical play that the Melee Engagement system heavily penalizes or prevents are re-shuffling formations mid-melee to take advantage of the terrain, tactical retreating and switching aggro. The lack of tactical movement in combat after has begun puts a large emphasis on the strategical aspect - the combat opening, while removing the tactical aspect - reactive play. I have been doing research into the removal of Melee Engagement and it looks like making the combat playable with it removed would only require some changes to things such as AI targeting clauses, various abilities related to engagement, and some animation and programming time to fix bugs related to attack animation interaction moving targets and run animation/combat idle animation blending. It looks like it will require MUCH less effort than trying to fix the system and will also remove clutter from the screen (Engagement arrows, custom selection circles for Engagement). I have not concluded my research yet though, it will still take me a few more days of testing, so expect a thread on that in the near future.

... I need somebody with a human touch .. hey you, always on the run gotta slow it down baby, gotta have some fun. As you all know, Melee Engagement has been a hot topic on the forums for the last month or so. I have been meaning to do an official topic on it for some time, and I feel I can delay no longer after seeing some shocking news in the recent Kickstarter Update. The PE dev team’s conclusion from their internal playthrough is that the Melee Engagement has problems with visual feedback. While this is a problem, there are far worse problems with the concept, design and implementation of the mechanic than there are with it's visual feedback. Obsidian have stated that they are going to devote precious 2D Art time, UI programming time and possibly animation/VFX time to making engagement clearer. This pledge, combined with the serious problems under the hood makes me really concerned about the amount development time and resources that will be spent on *trying* to fix a mechanic isn't a core mechanic of the game and isn't essential to invoking a modern Infinity Engine experience. So I would like to ask Obsidian to stop right now, consider my words, and the evidence that I am about to present before you spend any more development time on this mechanic, because it has serious problems that you may not be aware of and in my opinion the game will be much better off if it is cut, and those resources spent elsewhere. What is Melee Engagement and how does it work? Melee units in Pillars of Eternity have an invisible circle around them that defines Engagement range. This circle is slightly larger than melee range by default and can be increased by abilities and buffs. Melee units can engage one enemy by default, a limit which can be raised by abilities, talents and buffs (a Fighter with Defender mode on and the Hold the Line talent can engage up to 3 enemies in BB v364) When a unit moves into the Engagement range of an enemy, and that enemy has free engagement slots, that unit is engaged and an AI targeting clause stops their movement, cancels their current action and orders them to attack the first enemy they are engaged by. If a unit attempts to move while engaged, they suffer an automatic, invisible disengagement attack, free of recovery time, from the currently equipped primary weapon of the units they are engaged by. Regardless of the outcome of the attack, engagement ends after the attack has been made If the unit is hit by a disengagement attack they play a long interrupt hit reaction that stops their movement. There is no physical limit to the amount of disengagement attacks that a unit can score in a period of time. Does the Melee Engagement system fail it’s intended purpose? No it doesn't. It makes the first enemy (or enemies) that run by a melee character, stop to attack that melee character. That is the intended behavior and that is what happens. It addresses one specific case of the one specific style of kiting cited in the Kickstarter update – ranged units kiting melee units. When a melee unit engages a ranged unit in combat, the ranged unit cannot retreat without suffering a disengagement attack. This problem is meant to stop the player from kiting the AI melee units with bow wielding characters. Multiple methods were actually implemented to try and prevent this specific scenario – bow damage is pretty low, recovery time was previously paused while moving and now recovery time after making a ranged attack is slowed while moving. Those are it’s two intended goals as listed in the Kickstarter Update. It meets both of them. One other noticeable thing the system achieves is it heavily penalizes and disincentivizes movement in combat after the initial melee has begun. Josh Sawyer is a heavy tabletop player and has a lot of experience with turn-based games, and has often talked about movement coming with an opportunity cost in combat, so one can conclude that this behavior is also intentional. So if the system meets it’s design goals, then what's the problem ? There are many problems with the concept, design and implementation of the Melee Engagement system, but the most glaring issue is that it is completely abusable by the player. Since Melee Engagement range is a circle around a character and is larger than that character, one can theoretically overlap those circles so that when an enemy unit moves into engagement range of one character, it moves into engagement range of other nearby characters. Moving to attack the first enemy it was engaged by provokes a disengagement attack from the rest of the enemies that it is engaged by. Here are some examples of what happens when that occurs: In this video I have a party of six Fighters. These Fighters have Defender mode which allows multiple engagements and increases their engagement range (larger circles). This allows me to overlap the engagement range of all of my characters. Since disengagement attacks are free from recovery time and are a standard attack, I equipped them all with Two Handed Weapons to maximize damage output. In this video I have a party of three Barbarians equipped with Two Handed Weapons. These Barbarians have normal Engagement range and can only engage one enemy, but they have an ability called Carnage, which makes lower damage AoE cleave attacks on nearby enemies. Carnage is also procced on disengagement attacks.

Yeah it's mip map loss. Not sure what causes the door's jagged edges though.

Since when am I promoting this? Look up my post history, you'll see me frequently requesting and advocating for better AI targeting and robust target re-acquisition. Melee Engagement removes the need for a lot of it, which is one of the (many) reasons why I want it removed.

The player generally has a party of 3-6 characters and the situation is A LOT more complicated than you think. You can't use isolated examples like that. Here are some of the questions I would be asking in that situation How many units does the player have How many units does the player have that are currently engaged is there pathing space available to get to the Wizard what afflictions is that unit affected by what is the current health of that unit that would attempt to target a low health character how many enemies is the player fighting how many are still alive will they all attempt to beeline to the back and attack the wizard With the engagement system it makes NO SENSE for enemies to change targets like that, because they could be killing themselves from disengagement attacks. The player can abuse the system. However with no engagement, it would make a bit more sense that an enemy a certain distance away from a low health character would re-acquire them as a target if there was available pathing space.

Enemies rushing for the backline at the start of combat is not smart, nor challenging AI though. Any enemy that tries to run through Engagement will suffer damage from it and dealing damage to enemies is exactly what you want to do. That damage is FREE, it doesn't cost the player a single thing. If enemies rush through Engagement suffer a disengagement attack and I make my melee enemy hit them at the same time, that's one extra hit that I wouldn't have got if they had just attacked my Fighter instead - which is really stupid and makes it easier for the player to win.

Ranger and Companion Health need seperate health pools, but if one is dropped to 0 Stamina, the other should take modest damage from it or something. At the moment it's basically this Most Animal Companions do terrible damage in combat There's no point attacking enemy Ranger animal companion at all, as they have higher DT and Defenses than the Ranger does Therefore, just kill the Ranger and they both die That's not good design. Theoretically it sounded like a cool idea, but in practice it is woeful.

Medreth has Sitlettos, that remove 5 DT when determining damage. Will test it myself another day or something.

Why does that matter if you can just kill the Ranger with a few hits and both him and the companion die ?

I don't think it would be like that Cubiq, but enemies all going for the Wizard isn't smart. This game has narrow corridors as well as open spaces, all you have to do in corridors is physically block the pathing space with your frontline so that nothing can get through anyway. Funny thing is that in v278, enemies did go for your backline if you opened combat with your backline. I have a Sensuki vs the Beatles series where I played around the fact that beetles would only target my Wizard. I put him in the center of my other characters and they could never reach him due to lack of pathing space.

There is recorded VO in the game files and there are incantations. They just aren't hooked up to the beta yet. I am not going to post any (more) because they asked us not to.

I am not promoting dumb AI. The Engagement system is, and the people that enjoy it. Refer to previous post in thread If enemies were programmed to make decisions about whether to run through Engagement zones or not, all I have to do is understand what the clauses for it are, and then it is ripe for the same abuse that are currently present in the game (if not, even worse). If I know that enemies are going to try and run through Engagement zones, I can also play around that and cast Fleet Feet on my Wizard and kite them around while my other party members turn them into diced meat. Enemies going for the backline is, most of the time, not a smart decision and not representative of smart AI at all. What's more, there are only three months left on this project and in my email interview with Josh Sawyer from June, he said that since they have to build the AI from the ground up, we probably won't be seeing anything too complex. I am asking for something practical, within the scope of the project. Others, are not.

Related to this http://forums.obsidian.net/topic/69565-364-bugs-with-slicken-prone-disengagement-etc-all-in-15-seconds-span/?hl=slicken

I made a thread a while back saying it would be cool if the Ranger could switch between passive modals that only affected him and the companion (ie damage, DT, Deflection, Action Speed, Accuracy), and also had a few active buffs that only affected him and the companion. Would also be cool if all companions had an ability or two (an active and a passive). The only thing that was taken on board from that entire thread was talents that boosted the companion and Favored Enemy style talents (which were other people's suggestions).

Yeah when an update goes out I always get an email from Kickstarter and a Backer Website email. I am also subscribed to the Announcements thread so any time someone makes a thread in there, I get an email about that too, and I also read various forums that repost this information. The devs could be a lot better with patch notes, a bit better with forum back and forth but KS/email notifications is pretty straight forward.

It was mandatory up until v301, which was an easy patch. In v333 you could do multiple Chanters very well. Now that Paladins have been fixed they are viable too. Fighters are the best at it, but you can run a Chanter and a Paladin pretty well. They can take some hits, INT (+Deflection and AoE) is perfect for them and their auras are very very strong, and provide a lot of offensive capability. Monks can also do it, but I feel that since the advancement change, they don't have enough abilities to spend wounds and thus aren't as good at tanking, but rather good at single-target DPS. I use my abilities as soon as I have enough Wounds, but there's always more Wounds coming in than what I'm putting out, even in armor. Barbarians though, they are NOT good at tanking.

I think it's more than that. Outside of the Shared Health mechanic, the class just feels uninspired for me. There's a bit of 4E Ranger in there, but there's nothing that makes it fun, or that makes it a good class. They don't do very good damage, don't have any fun abilities, and the interaction between the ranger and the bear doesn't really go beyond targeting the same person. They could be much, much more interesting (and useful) than that. Probably the only class I would never pick at the moment, and it will be sad if they reach release in a poor state and one of the companions is gimped because of it.

Excluding Rangers. They are horrible. There are a few that need a bit of help as well (Barbarians, Monks and Wizards spring to mind).

Yeah this is because they added more colors in the new version. They will probably need to shrink it / rework the inventory UI. Shrinking 2D stuff in NGUI always seems to look terrible though ;_;