- Companion AI in NWN leaves a lot to be desired, even when the game is modded with Tony K's Henchman Inventory & Battle AI. Is there a chance of employing full party control and marquee selection of party members in the future, à la the Infinity Engine games, or at least the ability to mod it in? This could improve single-player game-play immensely and make Aurora more attractive to your Infinity player-base, too.
Trent Oster - Neverwinter Nights was never intended to be a Baldur’s Gate game. Baldur’s Gate poses the question of “what if you were the sole hero of a D&D adventure and you met interesting companion NPCs along the way”. Neverwinter was much more intended to be a collaborative, multiplayer experience. Neverwinter is “what will you do with your character on a D&D Adventure?” The rest of the party was imagined as players thinking the same thing and running their characters on the same adventure. We added companions into NWN to address playing the game as a single class character and needing access to other skillsets to fully enjoy the game. NWN companions can be improved, but we currently have no plans for a BG-style control scheme for companions.
Neverwinter Nights is not Baldur’s Gate. Baldur’s Gate is a novel about a single character with companions who join her. Neverwinter Nights is a journey for you and your friends. The concept from the very start was, and still is, that you play Neverwinter Nights in multiplayer with your friends. Neverwinter Nights is core multiplayer, bottom to top, you build your content and share it with other players.
The last thing we want to do is to make Neverwinter Nights into Baldur’s Gate.
This is exactly what I meant when I said direct control goes against the spirit of Neverwinter Nights.
As for breaking mods, there's probably a good chance enhancing the AI could break the AI mods, but maybe they can make the old behavior available for the old mods, or it might not be feasible to keep that backward compatibility at all. If the AI got enough of an improvement, it could just make the mods obsolete anyways, just like a lot of the community patch. I think most of us would be happier with better AI built into the game, but I suppose it could break a module that was built with an AI mod as a requirement. I'm not sure if there were a lot of 3rd party modules that required AI mods, though.
So there won't be any kind of BG/ID style mods for NWN EE? No chance for Storm of Zehir type campaign? That's a shame. I guess I'll have to stick to NWN2 or wait for Pathfinder Kingmaker then.
OK if we accept that there won't be direct control of henchmen, that still leaves:
1. At least allowing you to manage level up and inventories (as an option) 2. Allowing options on how the AI works (like you have in NWN2, where you can tweak when/if they shift from ranged to melee weapons, etc.) 3. Generally improved AI responses, like maybe when they get below 50% HP they attempt to heal themselves
At least these would reduce the main frustrations people have with the AI, although it will never be a patch on human control.
It would also be solved if you could get them to add the following script functions similar but more open than the familiar and DM systems. PossessCreature(object oTarget) UnpossessCreature(object oTarget) GetIsPossessedCreature(object oTarget) GetCreaturePossessedBy(object oTarget)
This would open up the possibility of more interesting spells but it would also allow custom content to allow henchman control.
I finished my run through of SoU and HotU (thankfully before the newest patch broke sounds for me ). When I finally got my shadow dancer level, I found hide in plain sight really messed with the henchmen AI. The henchmen would try to stealth in combat or remain stealthed once the fight had started, and just sit there through the whole battle. Maybe I'm just imagining it, but asides from the stealth issue, the AI in HotU did seem a bit better than SoU. I don't think i noticed the bad spell loops Deekin and Xanos were having in SoU.
They've got an item on the Trello board now, for providing customizable AI. It looks like some of the issues we've mentioned are covered there. I just really hope the standing around doing nothing/looping half casting a spell and stopping gets fixed.
I finished my run through of SoU and HotU (thankfully before the newest patch broke sounds for me ). When I finally got my shadow dancer level, I found hide in plain sight really messed with the henchmen AI. The henchmen would try to stealth in combat or remain stealthed once the fight had started, and just sit there through the whole battle. Maybe I'm just imagining it, but asides from the stealth issue, the AI in HotU did seem a bit better than SoU. I don't think i noticed the bad spell loops Deekin and Xanos were having in SoU.
They've got an item on the Trello board now, for providing customizable AI. It looks like some of the issues we've mentioned are covered there. I just really hope the standing around doing nothing/looping half casting a spell and stopping gets fixed.
Honestly, what is the problem to allow US to chose which way we want to play? I mean we had this option in NWN2 and people were happy. But now, when zealots hear about some changes in the AI, UI, additional gameplay features etc, they claim they know better what others want. This is after all "Enhanced Edition", not just Higher Rez mod.
As an owner of every enhanced edition on multiple platforms, I'd be really disappointed if they didn't improve the AI. There has to be something there to motivate me to buy this rather than keeping the old edition -- graphic filters simply won't cut it. Knowing that the AI has been improved would definitely provide serious motivation.
Not splitting the community is incredibly important for this game and adding improved AI as a listed feature will go a long way to bringing people over to the enhanced edition. Compatibility should be 100% providing you you don't change the original scripts just add new scripts, and then update the campaigns with these new scripts.
Updating the OCs henchmen to uniform patch level is something that I rather expect in an enhanced edition. Giving full control might be out of the question, but all of the OCs and premium modules should, at least, be up to date with henchmen inventory control, buffing options, and some of the other very basic updates that Bioware never had the sense to include. I'm really hoping that Beamdog doesn't drop the ball here, too.
I bought NWN:EE only for 2 reasons. To see full party control and improved AI or atleast in same level like Tony K. AI. For example full party control could be as an option turn on/off so both side could be happy. And at all it seems like devs just do not agree with everything what community asking for. So for what we are paying a money then...
Thats a little bit unfair towards the devs-
I actually agree with what you want (or if not full party control than at least way better AI and control over the AI) but I also dont think the devs should always do everything the community wants even if I am one of the people who wants it- I'm actually against voting on trello (even though I am voting because thats the system they have) for reasons I wont even get into- But the community isnt always right and democratized development doesnt always lead to the best choices being made.
As long as the devs are listening to us, reading our discussion and interacting- Thats good enough IMHO (and obviously progressing with the development, lol) but party control (or AI control and improvement), mod organization and graphics updates are my main 'wants right now' and sadly its not looking too bright for any of those ideas- But I feel more than pleased with my purchase even if none of those things come to pass for whatever reason.
So I'm a little worried that the Trello card makes no mention of cleaning up the stupid. After playing more, I've realized stealth and invisibility seriously messes up the henchmen AI. They just sit in stealth not helping at all, even when enemies have seen them and are beating them down. There's the spell loops, particularly visible in SoU and HotU (although I see it less in other modules it seems like). There's also the fact, they don't need to be in striking distance to attack with ranged weapons or cast spells, they frequently run right up into combat with a ranged weapon, then proceed to cast spells provoking attacks of opportunity, when those spells would be perfectly castable well away from the melee. The swordflight mod really makes the limitations of henchmen come into focus, with it's high difficulty.
Trent was also mentioning the clunkiness and slow reactivity of the AI in the last livestream, so I hope that gets fixed. There's no reason for a henchmen to stand around for an entire round, and not close with the next enemy until the following round.
The teleporting next to the main character never seems to work in a useful fashion, either. They never appear next to me when they've gotten stuck on some doorway or other object. They only seem to teleport when it would be most annoying, like right onto the trap my character is disabling. I find I have to issue stand your ground while disabling traps to keep them from stepping on the trap I'm disarming.
I think if the game releases with the AI doing things like this, it's going to be a huge negative on the reviews of all newcomers to the game. Those of us with nostalgia for the game are willing to overlook more, and take it in the context of when it was made, but newcomers likely won't, and we'll see a ton of negative steam reviews.
I'm definitely 100% by the stuff that's on the existing Trello card, but think the stupid needs to be added to it, as well. Fixing path finding, adding self preservation, scriptable AI, and not nuking me all sound good. I just don't think it covers all the bases that need to be covered.
Since this post sounds super negative, I'll leave it on the upbeat and say that I really appreciate beamdog working on this game and fixing up stuff.
You only gain direct control over your animal companions when a DM homebrew rules it and disregards the rules. Normally you use handle animal checks with several bonuses to order them to perform actions and tricks you have trained them before as opposed to telepathy that arcane spell casters get with familiars. Their intelligence doesn't change and if it does with a spell like awaken then it's not longer eligible to be an animal companion. Few exceptions exist such as the Exalted Companion feat which allows magical beasts to be companions, depending on communication options you can give complex tasks. There are different levels of communication between summons and all sorts of companions, such as emphatic, telepathic, simple commands, shared language, through simple gestures.
So D&D actually has a lot of granularity and differences in the level of control you have over allies but none of that matters if you can directly control everything. In that case you essentially assume they have constant telepathy and magical domination over all their allies which cannot be dispelled.
No, the player simply role-plays more than one character. If the multiple character's are not able to communicate directly that is role-played, just as it is if characters controlled by different humans who are not able to directly communicate with each other is role-played.
If you have ever met a RL dog, you would know that they don't need super-intelligence to co-ordinate effectively with humans.
You only gain direct control over your animal companions when a DM homebrew rules it and disregards the rules. Normally you use handle animal checks with several bonuses to order them to perform actions and tricks you have trained them before as opposed to telepathy that arcane spell casters get with familiars. Their intelligence doesn't change and if it does with a spell like awaken then it's not longer eligible to be an animal companion. Few exceptions exist such as the Exalted Companion feat which allows magical beasts to be companions, depending on communication options you can give complex tasks. There are different levels of communication between summons and all sorts of companions, such as emphatic, telepathic, simple commands, shared language, through simple gestures.
So D&D actually has a lot of granularity and differences in the level of control you have over allies but none of that matters if you can directly control everything. In that case you essentially assume they have constant telepathy and magical domination over all their allies which cannot be dispelled.
Yes but again- This isnt tabletop.
The DM is playing the henchmen effectively- The Ai does not... The DM is allowing you to discuss strategy and then implement that strategy.
If we were sitting at a tabletop a million things would be different- this is a computer game with limits and unable to creatively adapt to situations. I'm not playing at a virtual tabletop where were all being controlled by humans- generally I am the only human in an AI controlled world with massive limitations .
Full party control would go a long way in not only making the game seem proper in terms of tactics (My wizard should not be the one charging in the room first or trying to hold a doorwall but if I play a wizard in the computer game, that is how it plays out while the fighter is following behind) but also in terms of fun.
Those that dont like it would either not use or or disable it in mods they create.
This doesn't need to be an argument over FPC vs improving the AI, they're not mutually exclusive. Even if, and that's a big IF, FPC is implemented, I know I'd still want the AI improved. The less switching back and forth between characters I have to do, the better. I still don't think FPC is going to happen, but hey maybe they'll do it.
The fact it isn't table top means that it can more effectively track and implement those nuanced rules without putting the need for book keeping and rules application on players in a live game. The computer simply handles all the necessary things automatically.
The fact that table top games are made for human interaction is one of the drivers for why old editions received criticism and why 5e is rules light with many simplification and more emphasis on theater of the mind. So it's incorrect to assume that a computer game is less equipped to handle complex simulations whereas a table of friends don't want to spend 4 hours figuring out the grappling rules rather than playing.
An animal companion in this case is still an animal with it's own will and capabilities as well as the inability to communicate normally and less capacity to understand complex tasks. Hence you have the handle animal checks to learn and use certain tricks, such as guard this spot, sic on an enemy, flank the target. It's not always going to go perfectly to plan even if you are an epic level dog whisperer.
If you objective is that you are controlling a small task force or an army them full control might be desirable. On the other hand if your objective is to have a game where you are a character with different kinds of associates that may not all be telepathically linked then full control is counter productive. I'm not judging ppl for wanting to make or preferring either type of game, but there are obviously reasons the rules of D&D don't use full control as the standard.
If your design goal is to recreate that experience and heavily simulate interactions then simplifications like full control of associates are obviously counter productive to your stated design goal.
So I'm a little worried that the Trello card makes no mention of cleaning up the stupid. After playing more, I've realized stealth and invisibility seriously messes up the henchmen AI. They just sit in stealth not helping at all, even when enemies have seen them and are beating them down. There's the spell loops, particularly visible in SoU and HotU (although I see it less in other modules it seems like). There's also the fact, they don't need to be in striking distance to attack with ranged weapons or cast spells, they frequently run right up into combat with a ranged weapon, then proceed to cast spells provoking attacks of opportunity, when those spells would be perfectly castable well away from the melee. The swordflight mod really makes the limitations of henchmen come into focus, with it's high difficulty.
Trent was also mentioning the clunkiness and slow reactivity of the AI in the last livestream, so I hope that gets fixed. There's no reason for a henchmen to stand around for an entire round, and not close with the next enemy until the following round.
The teleporting next to the main character never seems to work in a useful fashion, either. They never appear next to me when they've gotten stuck on some doorway or other object. They only seem to teleport when it would be most annoying, like right onto the trap my character is disabling. I find I have to issue stand your ground while disabling traps to keep them from stepping on the trap I'm disarming.
I think if the game releases with the AI doing things like this, it's going to be a huge negative on the reviews of all newcomers to the game. Those of us with nostalgia for the game are willing to overlook more, and take it in the context of when it was made, but newcomers likely won't, and we'll see a ton of negative steam reviews.
I'm definitely 100% by the stuff that's on the existing Trello card, but think the stupid needs to be added to it, as well. Fixing path finding, adding self preservation, scriptable AI, and not nuking me all sound good. I just don't think it covers all the bases that need to be covered.
Since this post sounds super negative, I'll leave it on the upbeat and say that I really appreciate beamdog working on this game and fixing up stuff.
AI improvements are mentioned in the second card, but that card is also for the full party control, while the first one covers "functionality" for henchmen. To which card should we add the part about fixing pathfinding and other suggestions?
AI improvements are mentioned in the second card, but that card is also for the full party control, while the first one covers "functionality" for henchmen. To which card should we add the part about fixing pathfinding and other suggestions?
That's the one I was referencing earlier when I mentioned not seeing cleaning up the buggy behavior. Maybe just append clean up buggy behavior to the name, since it already covers some stuff like the path finding and OnBlocked? To paraphrase it to fit on a card easier, maybe something like this?
Fix spell casting loops Don't teleport henchmen in front of PC (especially while disabling a trap) Fix not engaging while stealthed/invisible Keep out of melee when not necessary for ranged attacks/spells (might be covered under self preservation/flee) Don't stand just outside a room while a battle is going on inside
"AI improvements are mentioned in the second card, but that card is also for the full party control, while the first one covers "functionality" for henchmen. To which card should we add the part about fixing pathfinding and other suggestions?"
I would personally try to add things like pathfinding to both- Fact is, alot of us will probably vote for both because one or the other has to happen- However, I would think making competent and situationally aware AI is going to be well beyond the scope of what can feasibly be accomplished while FPC would be a whole lot easier and imho better anyhow. It really comes down to the developers here.
I dont think its negative either- Just honest. We obviously love this game or we wouldn't be here but part of 'love' is 'honesty' and being able to see where it has faults and issues and AI is terrible.
I too, am replaying the OC and two things stand out for me- The first is that the OC isnt nearly as terrible as I remember it being and TBH isnt really bad at all just a bit generic but its also enjoyable. The second is that the AI is worse than I remember it being.
So far I'm loving the EE overall and seeing the potential for really great things here and a massive success for reasons i'll not even bother getting int- But the AI needs fixed badly and i'm not certain competent AI can ever account for the varied situations (If it can that would be awesome) so I'm for full party control- but one or the other needs to happen and things like pathfinding would need to be done with both imho.
AI improvements would not be appropriate in the full party control card in any case. This is because full party control refers to the player having the decision making control of the associates rather than relying on AI for decision making. It essentially circumvents it rather than improves it. I'd suggest putting it in the first card or the one DrHappyAngry mentions.
AI improvements would not be appropriate in the full party control card in any case. This is because full party control refers to the player having the decision making control of the associates rather than relying on AI for decision making. It essentially circumvents it rather than improves it. I'd suggest putting it in the first card or the one DrHappyAngry mentions.
I respectfully disagree- unless you want this to be a turn based game - As it stands full party control will 'over ride' the AI rather than circumvent.
in many situations we will still be reliant on AI to move around themselves and do basic things and some improvement is still necessary (or at least would be nice to have)
Its not 'either or' imho but more an amalgamation of both in the sense that pathfinding and other things still need improvement regardless of which way things go.
As the card for providing customizable AI already has covered pathfinding improvements, and people already have voted for it via votes for that card, I added new suggestions there and also changed its title.
It doesn't have to be either AI or FPC but they are entirely separate issues. Note that AI is not dependent on full party control nor is full party control dependent on improved AI. Anyone is free to respectfully disagree but this is the way it is.
As the card for providing customizable AI already has covered pathfinding improvements, and people already have voted for it via votes for that card, I added new suggestions there and also changed its title.
Comments
As for breaking mods, there's probably a good chance enhancing the AI could break the AI mods, but maybe they can make the old behavior available for the old mods, or it might not be feasible to keep that backward compatibility at all. If the AI got enough of an improvement, it could just make the mods obsolete anyways, just like a lot of the community patch. I think most of us would be happier with better AI built into the game, but I suppose it could break a module that was built with an AI mod as a requirement. I'm not sure if there were a lot of 3rd party modules that required AI mods, though.
1. At least allowing you to manage level up and inventories (as an option)
2. Allowing options on how the AI works (like you have in NWN2, where you can tweak when/if they shift from ranged to melee weapons, etc.)
3. Generally improved AI responses, like maybe when they get below 50% HP they attempt to heal themselves
At least these would reduce the main frustrations people have with the AI, although it will never be a patch on human control.
PossessCreature(object oTarget)
UnpossessCreature(object oTarget)
GetIsPossessedCreature(object oTarget)
GetCreaturePossessedBy(object oTarget)
This would open up the possibility of more interesting spells but it would also allow custom content to allow henchman control.
They've got an item on the Trello board now, for providing customizable AI. It looks like some of the issues we've mentioned are covered there. I just really hope the standing around doing nothing/looping half casting a spell and stopping gets fixed.
https://trello.com/c/M27Vvlnw/20-provide-customizable-ai
Thread about it here: https://forums.beamdog.com/discussion/67534/bug-henchman-death-in-shadows-of-undrentide
I actually agree with what you want (or if not full party control than at least way better AI and control over the AI) but I also dont think the devs should always do everything the community wants even if I am one of the people who wants it- I'm actually against voting on trello (even though I am voting because thats the system they have) for reasons I wont even get into- But the community isnt always right and democratized development doesnt always lead to the best choices being made.
As long as the devs are listening to us, reading our discussion and interacting- Thats good enough IMHO (and obviously progressing with the development, lol) but party control (or AI control and improvement), mod organization and graphics updates are my main 'wants right now' and sadly its not looking too bright for any of those ideas- But I feel more than pleased with my purchase even if none of those things come to pass for whatever reason.
Trent was also mentioning the clunkiness and slow reactivity of the AI in the last livestream, so I hope that gets fixed. There's no reason for a henchmen to stand around for an entire round, and not close with the next enemy until the following round.
The teleporting next to the main character never seems to work in a useful fashion, either. They never appear next to me when they've gotten stuck on some doorway or other object. They only seem to teleport when it would be most annoying, like right onto the trap my character is disabling. I find I have to issue stand your ground while disabling traps to keep them from stepping on the trap I'm disarming.
I think if the game releases with the AI doing things like this, it's going to be a huge negative on the reviews of all newcomers to the game. Those of us with nostalgia for the game are willing to overlook more, and take it in the context of when it was made, but newcomers likely won't, and we'll see a ton of negative steam reviews.
I'm definitely 100% by the stuff that's on the existing Trello card, but think the stupid needs to be added to it, as well. Fixing path finding, adding self preservation, scriptable AI, and not nuking me all sound good. I just don't think it covers all the bases that need to be covered.
Since this post sounds super negative, I'll leave it on the upbeat and say that I really appreciate beamdog working on this game and fixing up stuff.
(In my game players frequently directly control NPC allies as well).
So D&D actually has a lot of granularity and differences in the level of control you have over allies but none of that matters if you can directly control everything. In that case you essentially assume they have constant telepathy and magical domination over all their allies which cannot be dispelled.
If you have ever met a RL dog, you would know that they don't need super-intelligence to co-ordinate effectively with humans.
The DM is playing the henchmen effectively- The Ai does not... The DM is allowing you to discuss strategy and then implement that strategy.
If we were sitting at a tabletop a million things would be different- this is a computer game with limits and unable to creatively adapt to situations. I'm not playing at a virtual tabletop where were all being controlled by humans- generally I am the only human in an AI controlled world with massive limitations .
Full party control would go a long way in not only making the game seem proper in terms of tactics (My wizard should not be the one charging in the room first or trying to hold a doorwall but if I play a wizard in the computer game, that is how it plays out while the fighter is following behind) but also in terms of fun.
Those that dont like it would either not use or or disable it in mods they create.
The fact that table top games are made for human interaction is one of the drivers for why old editions received criticism and why 5e is rules light with many simplification and more emphasis on theater of the mind. So it's incorrect to assume that a computer game is less equipped to handle complex simulations whereas a table of friends don't want to spend 4 hours figuring out the grappling rules rather than playing.
An animal companion in this case is still an animal with it's own will and capabilities as well as the inability to communicate normally and less capacity to understand complex tasks. Hence you have the handle animal checks to learn and use certain tricks, such as guard this spot, sic on an enemy, flank the target. It's not always going to go perfectly to plan even if you are an epic level dog whisperer.
If you objective is that you are controlling a small task force or an army them full control might be desirable. On the other hand if your objective is to have a game where you are a character with different kinds of associates that may not all be telepathically linked then full control is counter productive. I'm not judging ppl for wanting to make or preferring either type of game, but there are obviously reasons the rules of D&D don't use full control as the standard.
If your design goal is to recreate that experience and heavily simulate interactions then simplifications like full control of associates are obviously counter productive to your stated design goal.
https://trello.com/c/SzTaJkBM/5-add-new-companions-henchmen-and-add-functionality
https://trello.com/c/lmXsvl13/112-allow-full-control-of-ai-party-summons-henchman-companions-and-familiars
AI improvements are mentioned in the second card, but that card is also for the full party control, while the first one covers "functionality" for henchmen. To which card should we add the part about fixing pathfinding and other suggestions?
https://trello.com/c/M27Vvlnw/20-provide-customizable-ai
That's the one I was referencing earlier when I mentioned not seeing cleaning up the buggy behavior. Maybe just append clean up buggy behavior to the name, since it already covers some stuff like the path finding and OnBlocked? To paraphrase it to fit on a card easier, maybe something like this?
Fix spell casting loops
Don't teleport henchmen in front of PC (especially while disabling a trap)
Fix not engaging while stealthed/invisible
Keep out of melee when not necessary for ranged attacks/spells (might be covered under self preservation/flee)
Don't stand just outside a room while a battle is going on inside
I would personally try to add things like pathfinding to both- Fact is, alot of us will probably vote for both because one or the other has to happen- However, I would think making competent and situationally aware AI is going to be well beyond the scope of what can feasibly be accomplished while FPC would be a whole lot easier and imho better anyhow. It really comes down to the developers here.
I dont think its negative either- Just honest. We obviously love this game or we wouldn't be here but part of 'love' is 'honesty' and being able to see where it has faults and issues and AI is terrible.
I too, am replaying the OC and two things stand out for me- The first is that the OC isnt nearly as terrible as I remember it being and TBH isnt really bad at all just a bit generic but its also enjoyable. The second is that the AI is worse than I remember it being.
So far I'm loving the EE overall and seeing the potential for really great things here and a massive success for reasons i'll not even bother getting int- But the AI needs fixed badly and i'm not certain competent AI can ever account for the varied situations (If it can that would be awesome) so I'm for full party control- but one or the other needs to happen and things like pathfinding would need to be done with both imho.
in many situations we will still be reliant on AI to move around themselves and do basic things and some improvement is still necessary (or at least would be nice to have)
Its not 'either or' imho but more an amalgamation of both in the sense that pathfinding and other things still need improvement regardless of which way things go.
As the card for providing customizable AI already has covered pathfinding improvements, and people already have voted for it via votes for that card, I added new suggestions there and also changed its title.
https://trello.com/c/M27Vvlnw/20-provide-customizable-ai-and-clean-up-buggy-behavior