Larian Studios Forums Baldur's Gate III Baldur's Gate III - Mega-threads and Guides Stealth & Ambush Game Mechanics Megathread

They might look a bit too much but have in mind that:
• that horrific image I did is like an example to explain the functionality, not the look and feel of what could come out. It could be way more easy on the eyes (I like the Mark of the Ninja call too)
• the character circle only applies for characters when the character has armor with disadvantage in stealth checks. If you have many you probably aren't sneaking around in encounters (and shouldn't), and if you have just a few, it would be useful to have a clear indication of where you can go.

On the different noises:
• Using spells breaks the stealth, so you are getting caught from way farther.
• Attacking breaks stealth also, unless missing the attack (I think) which is really weird and should be changed.
• And finally jumping, it would make sense to have it make more noise (so widenning the cuircle let's say) but I think is clearer if you just can't go there.

So, there aren't really many things to do while sneaking, most of them would break stealth, and some I wouldn't make the produce more sound because they are probably just annoyances with out purpose (like making noise from coating a weapon with poison).


The application I have in mind is somewhat of a deterrent, if you have those type of armors just don't try it, and the solution seems simple to understand from the user standpoint, and to make (I think) from the devs standpoint.

Many good ideas here. However, I would have liked to see an even playing field for better gameplay. I'd like a gatekeeper mechanic for all potential Stealth situations where a hidden Stealth (lowest nearby party member) contested by NPC perception (highest nearby NPCs) check is required for any stealthing. Implement realistic modifiers like using light-sources in darkness leading to a penalty. Such a mechanic would make the Pass Without Trace Ranger-spell (+10 party wide Stealth) very viable.

1. Very much for a passive hearing radius triggering hidden Stealth checks. Makes the Silence spell more viable as it allows for creative use in such situations.
2. Starting turn-based phase for the entire party when combat is initiated is a necessity. Perfect surprise turns should be the exception, not the rule.
3. Wider, longer and hidden NPC vision cones. NPCs should be less predictable, glance around more.
4. AI refinements. Like a small party of goblins being more defensive, perhaps retreating to alert their allies. Reacting in a believable manner to player actions. An intelligent creature would for instance be on alert (hidden Perception check?) after first having been tricked by Minor Illusion and definitely react to taking damage by alerting allies.



Don't give Larian an excuse. Exploits could be considered as core mechanic in their games so far lol. Player agency and cheap laughs is a priority - while balancing and immersion/realism seems to be considered a resource intensive chore. Another word for "self-policing" is "self-nerfing", and the problem with that should be obvious: If you happen to like stealth and smart plays you're just sh!t out of luck, cause the system is so dumb you feel cheap for engaging with it. Now you can't play a certain way in order not to trivialize challenge. However, long before most players realize the system is basically a broken exploit, the lack of balance will have heavily incentivized players to engage in repetitive tactics that makes for a more predictable and boring gameplay than would be possible in a properly balanced system. Besides, optimization is a fundamental aspect of RPG gameplay. You try to find character build synergies and tactics that plays to the strengths of party. Self-policing/nerfing is the antithesis to that.

Fortunately Larian stated in the very early stages of EA that they were aware of the pitfall of making some combat tactics too strong: That it would incentivize repetitive and more boring gameplay. Larian said they therefore wanted to make more varied tactics viable. I really hope they remember those words...especially considering it kind of goes against their core game design philosophy.

I for one like the idea ...

But i think it would need same mechanic as sight have ... you know, Light > Partial Shadow > Deep Shadow
So ... from the far one HUGE ring where we would make easy stealth checks ... at mid range one conciderably smaller ring, where we would make conciderably harder stealth checks ... and on personal distance VERY hard stealth checks.

Yes.

If you check the screenshot, there is no way you can stealth around Crusher without stealth check ... and that is the point here, isnt it?

---

I would say if the question is, how often would it make sense in general ...
I mean, there is a waterfall on that screenshot ... have you ever ben this close to waterfall? Do you really think you would hear someone walking?

I’ll just reiterate support for points already made.

- Only rogues get hide as a bonus action at level 2, as per TT rules.
- No hiding in plain sight during combat, replace sight cones with 360 degree sight up to vision-blocking objects, again as per TT rules.
- Sniping automatically reveals location unless you have sniper feat, as per TT rules… seeing pattern here.
- (Overland encounters at range, like Githyanki patrol, heavily favor party with longer range. Give such foes a response to close distance and/or call in dragon strike.)
- Acute hearing, enhanced smell, tremorsense (bulette!), life sense and dragon senses are all-directional, expand sight cones to 360 degrees for such creatures.
- Companions nearby not in combat switch to turn mode and get inserted into initiative order.

Stealth in TT has similar problems as BG3, so I’m understanding towards Larian for having to implement tricky rules.

May I ask you to read my full “edit”? I was hypothetically asking potential impact of removing vision cones and using smaller detection rage (smaller than current vision cones, not smaller than hearing circles on the screen shot) - aka. PoE1&2, Solasta, Wasteland3.


I am not clear what is the point you are trying to make. The game doesn’t have a system that would account for ambient noise. As such what is in the environment doesn’t matter, and stealth will work the same way in rowdy goblin camp, as it will in a quiet tomb.

Depending how “hearing” checks would be calculated there could be ambient noise mechanic, similar to how light works for vision cones. We are talking about adding properties to objects in the environment, and honestly that’s something that could only be added on top of basic functioning stealth system. I think the only thing we can afford to dream of right now, is ways to make current stealth system better, rather than adding stuff on top of the system that doesn’t work to begin with.

Why use smaller detection range if you remove cones? They should see & hear out to the same distance.

The game already checks for vision collision, i.e. whether something blocks line-of-sight. Just do this in 360 degrees.

Regarding heavier armour and ambient noise, it's assumed that its bulk makes it both harder to hide and harder to sneak quietly.

I find the idea compelling but one thing that I can't wrap my head around is how the rolls are triggered. Let's say I want to shank someone, so I approach him with stealth. As I close down I will be triggering stealth checks making 3 in total.
How are this checks made? Am I rolling or using a passive stealth check?
Let's say this are the difficulties:
• Far range: Enemy passive perception -5
• Mid Range: Normal passive perception
• Kiss on the neck™ range: Passive perception +5

- If my stealth rolls are separate it's a huge disadvantage because the probability of getting a bad roll increases.
- If you make a unique roll and use it for all the different ranges two possibilities appear:
1) You get the number on your log, you con exploit this system by examining your enemy and checking if your roll works for all the ranges, if not, you go back and try again until you get a desired roll (or use resources to have a better stealth check), as the first ring is the easiest one, it's simple to accomplish so it gives you the chance try and see your rolls before hand.
2) If the roll is hidden is a bit more fun, if the result is bad you can always reload and try again. But I would say that it works properly as a deterrent though and has a gamble component like, how confident are you on that hidden roll?
- If we use passive stealth, it's similar to things said before, you can examine the enemy and make assumptions on that. This number can only be changed with resources, so "Blessing of the Trickster" as it gives advantage would be +5, Pass without Trace it's a +10, so you use them up and it's just maths there, no randomness.

Now, on those basis I think the most compelling one is the "Hidden roll one" because it's the one with the least exploits, but I feel that is a bit convoluted and it's pretty much the same as having the one circle.

I think in that regard we enter in a bracket of Realism vs Gameplay
Usually when playing a stealth based game you start with a simple mechanic and the game adds complexity in steps like, for example, "Styx" game:
-You first learn about the difference in sound between run and sneaky walk.
-Then you learn about hiding under a table, in wardrobe, behind a wall, etc.
-Then how to approach and remove light sources.
-Then kills.
-Then muffled kills.
-Distrations thowing breakable objects.
-Difference in noise produced between walking in soft surfaces, hard ones, and others with debris and stuff.

And so on, and so on.
They build on top of the system and instruct the user as he plays. But that's their gimmick, their whole thing. In BG3 stealth is more like a side option to approach combat, so adding too much complexity probably lacks purpose.


I'm more in line with making decisions based on gameplay purposes than realism.

So, why would I make an armored character have trouble shanking an enemy in melee range?

In DnD melee characters are the ones with the most options to enhance their attacks, example:
-A lvl 8 character, 3 lvls in rogue Assassin + 5 in Paladin could (and I quote a reddit post):
"Any hit on a surprised creature is a crit (because Assassin subclass). So using second level divine smite, on a single round (with advantage) against a surprised enemy, I could be doing 2d8+mods (from the crit on the rapier attack), plus 6d8 (second level divine smite on the crit) plus 4d6 (from critting on sneak attack) and then another 2d8+mods from the second rapier attack crit (and maybe even another 6d8 if I expend another second level divine smite for my second attack in the round)."
On top of that in BG3 you can coat your weapon with poison so the one shot potential is crazy.

Now, it's a Rogue too, and I think he should be able to sneak up on people, but let's make him spent some resources:
-Paladin's heavy amor? Gone. He should continue the fight with less AC at least (though with high dex he could just be using light armor).
-Getting to a main baddie in this approach should be almost impossible unless you have a crazy amount of resources spent on it (Blessing of the Trickster, Pass without Trace, Invisibility), on top of using the smites and poisons.
It should be hard to get him because there should be guards making sure no-one gets inside, standing guard, patrolling; so multiple rolls required.
-He gets separated from the rest of the party. You've might have killed the BBEG guy but now you are far from your party surrounded by enemies and you might not have the AC nor hp to survive.

This is all based on that you gank the most important NPC, if you just do it on a lowly guard, combat triggers and that's it, you don't earn a lot of advantage.

A problem is that this doesn't affect range character:
For them the solution is as real life, have walls protecting important people. I feel like in chapter 1 we had a lot of advantages because many enemies can be neutral or friendly towards us, like Nere or the Goblins at some points. So we got to walk wherever we wanted. This might change in the following chapters. Also, we've had a lot of open spaces to exploit.


On top off all that, I think a good design logic is repetition, the player learns mechanics by applying them over and over again, if stealth is an alternative approach it shouldn't have too much variation, nor complexity as the game progresses. It might sound boring but BG3 have many other things that consume our attention like character/party builds, story paths, RP choices, inventory management, etc.

Personally, I don't enjoy the length of vision cones. Too often character whose vision I am trying to avoid isn't even on screen, which I find unenjoyable. Yes, giving characters that wide of a detection range as you suggest, would probably be "realistic" but I don't think it would make for a good gameplay. At this moment Larian has two extremes. A very long and difficult to avoid detection in vision cones and complete lack of it when flanking the character. All I am suggesting is to even it out - give them "medium" detection range surrounding them.

As I see it, what is vision cones function in Baldur's Gate3 as a game?

1) blocking paths/access to items/chests
2) making ambushes trickier?
3) making pickpocketing trickier.

A medium sized detection circle (when I say smaller, I think smaller than vision cone, but larger than what was in the mockup screenshot above - perhaps 1,5-2x larger - something to be of decent size, but still leave enough free space on screen to manouver around). I think that would fulfill the game's need

OBJECTION!
As I continue thinking of it vision cone's range comes useful to counteract BG3 broken in-combat stealth. With smaller detection range only it would only make it even easier to disappear and sneak attack in combat.

Ok, let's keep vision cone as unwieldy as it is, and just add smaller hearing range radious. And make hide full action.

Some excellent ideas in this thread.

FYI, 5e actually has official rules for audible detection distances. The caveat is that they're only present on the Dungeon Master Screen.
These values could be used to inform detection circles' distances.

Typical stealth activities like hiding, sneaking, pickpocketing = you have to make a stealth roll if you come within 35 ft. Farther than that, you're fine.

Spellcasting (with verbal components) rules as intended seems to disallow whispering ("particular combination of sounds, with specific pitch and resonance, sets the threads of magic in motion"; the existence of Subtle Spell Metamagic), so you have to make a stealth check if you're between 35-70 feet, and you're automatically noticed if within 35 feet. I wouldn't recommend having two visible circles below each NPC representing 35 and 70 ft; the 70-ft circle could be invisible and a nice surprise if players attempt to cast a (non-offensive & invisible) spell while hidden!


More generally, I'm a fan of a combined sight cone plus a (smaller) audible circle.

I see some good idea's and all but what I really want to see changed is the way the NPC react to being attacked by someone that shoots them and then hides again. I don't see a problem with shooting and hiding, its the reactions by the targets that lack any thought what so ever.

Right now they all stand their like nothing happend even if one of them dies. Thats garbage and needs to be fixed. At least combat should be started and they should actively move around in an attempt to find out what is killing them and attack if possible.

Need dumb AI/npcs upgraded to be able to deal with stealthers and I belive this needs to be added to mega threads because its that broken and needs to be looked at and not just scrolled past. I have noticed this since the start of the game but had hopes it would be managed but now fear its going to be this way in final release.

I agree but I feel that "make AI" respond is a much harder thing to pull of that modify ruleset to reign things in to begin with.

How would AI respond? Would they figure out where the shot came from and go investigate? Or should they follow procedurally generated search paths? Getting spotted in stealth is disasterous at the moment, and puts player on great disadvantage, especially if only one of their companions gets spotted. Ineffective AI search would be still useless, and and effective AI search might make attacks from stealth absurdly backfire.

Real time is very quickpaced and clunky, so if there was life reaction I think we should switch to turn based on attack, so we can reposition and than let AI do it's thing. We would be in combat (ideally whole party), so getting spotted would make us a target, but combat would already be in the process so there wouldn't be the issue of one character at a time being dragged into combat que and loosing their actions. But in essence now I am talking about turn-basef pre-combat stealth phase, and I feel it would be a big new gameplay feature, rather than a fix to what we already have.

Yes
/thread

If I'm understanding OP correctly, they're talking about the scenario of shooting from stealth and then hiding at the end of your turn, so that the AI doesn't know where you are and just stands there. Not real-time stealthing before a battle, in order to try to get Surprise.

In 5e, when you attack, you reveal your position. Even if you hide again, the enemies were still alerted to your previous position. Now, if you attack -> hide -> move, you might be safe. But at the very least the enemies should go to your last known location to search you. Alternatively/in addition, the enemies could take the Hide/Dodge actions (if ever implemented) in response to damage.

One thing with having monsters search for a hidden sniper is that you'll more easily be able to pull mobs to each other, like Githyanki patrol towards the Flaming Fist at Waukeen's Rest.

(I've heard of solo players doing just this, but as it is currently, it's tedious.)

I'm not sure if this would feel cheesy or smart tactical play? I suppose it'd be like barrels, up to us players to choose to use or not.

I see, so there is no "stealth roll" to remain hidden after attack from stealth in original rules? If so,than yes, I think it would be fair to give enemies "ping" to investigate even if character remained in stealth. Of course, if character hides when combat is in process 1) it should take full turn, so one cannot attack and hide on the same turn, 2) enemies should remember last seen location and investigate.

Nope, there isn't! This is said directly in the "Unseen Attackers and Targets" rules (PHB p 194-195), and also by implication in the Skulker feat: "When you are hidden from a creature and miss it with a ranged weapon attack, making the attack doesn't reveal your position" --> you ARE revealed if you don't have the Skulker feat, even if you miss.

As for hiding during your same turn, I disagree. Per 5e RAW, only rogues can Hide as a Bonus Action. For everybody else, it costs an action. If e.g., a rogue attacks from stealth -> breaks line of sight -> hides again, they should be able to. This is intended, and is ~required for rogues to remain competitive dpr-wise. If a Fighter wants to Action Surge to attack -> break line of sight -> hide again, they should be able to.

Of course, in BG3, everyone can hide as a bonus action, which is too much imo. Under this implementation, I wouldn't care if Hide was restricted like that, but now we're fixing homebrew with more homebrew.

Totally!

I think that if solutions from this thread are to be used, changing the basic Hide to be used as an action is given.

So, building on top of that with the example given previously:

Rogue attacks -> moves away from enemies cone of vision (and maybe hearing circle) to become unseen and unheard -> uses Cunning action: Hide

After this it might be a mediocre solution but I'll say:
if an enemy wanted to hit the Rogue (based on the criteria that the game currently posses to choose target) then it makes a Wisdom (Perception) check contested by a Dexterity (stealth) check by the player.
• If it suceeds it knows the player's position and can act accordingly.
• If it fails and has other enemies to attend, he possibly attacks them or whatever.
• If he hasn't other enemies to attack he could take a defensive position, where, It could take a Dodge Action (so yeah, I'm asking Dodge Actions to be added too =P ).
The dodge action is actually quite good against Rogues specially because it gives Disadvantage, what could cancel sneak attack, BUT it has a problem in this case, because the attacker should be seen for the dodge to be performed. Maybe if the rogue has to enter in the cone of vision that will allow the dodge and basically put in check the hiding tactic for cheesing.


One problem would be that, let's say you hide and then you do a movement that the enemy shouldn't be able to predict, like, a Rogue + Sorcerer multiclass hides, then use subtle spell to cast misty step to the other side behind a wall, an the enemy just wins the contested check and goes exactly to your position. That would be weird but I think it's pretty fair.
(also subtle spell doesn't work like that in BG3 so kind of a bad example by my part).

Every one of these bullet points above is ABSOLUTELY needed in the game. There are reasons why there are certain feats in the game to allow for various special actions. Certain spells/abilities can just add a background stat/buff that turns the current cone vision into 360 vision (as explained, many, many different effects can cause this).

And the easiest way to solve the crazy interaction of stealth in this game is if you are anywhere near combat, no matter your condition/position/status/ect, you enter turn combat.

This issue is just as important as the previous issues of using food as potions, short rest vs long rest, Advantage/Disadvantage from Height, and Reactions. Treat this issue as such, Larian.

I'll just pop in to add in my general agreement for the points made here.

Yeah, there is zero question that Stealth MUST be changed so that if initiative gets rolled, it gets rolled for all characters nearby - whether they've been detected or not. If there's a character hiding when a combat starts, great, they can act on their turn. But as soon as initiative gets rolled, they're in turn-based mode even if they remain undetected.

If a character elsewhere approaches, then once they get close enough (no closer than 60 feet from a combatant) then they automatically roll initiative and need to wait on their turn to move, even if they approach while sneaking.

Unless your character has the Skulker feat, attacking from stealth should break stealth, period. You cannot attempt to hide again until no enemies have line of sight onto you.

Running away from a combat and stealthing is a problem. Wounded characters could snap-regen to full if the combat stops due to enemies losing track of the player, but that won't help if an enemy gets killed.

I agree that only Rogues of level 2+ should have bonus action Hide. That's basically their thing.

I also agree that in combat, full 360 degree sight, which is blocked by vision-blocking objects makes sense. If a fight has broken out, characters are obviously aware and on the lookout for danger.

People asking for complete gameplay overhauls when release is August? I mean, I get wanting changes, but let's be realistic here, almost everything we've seen gameplay wise is set in stone, it's not changing.