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

I would like to bring brilliant Invisible Inc. as a counter example.

If they want to stick with their cone vision concept, they should probably at least adjust it to mimic hearing and periferal vision as well, as 1varangian justly mentioned. I see it as a 3m red circle in addition to the cone, centred around the character. Entering it would force a stealth check with high difficulty or a low difficulty and a disadvantage to remain hidden while staying that close to an enemy.

I have no objection to the rules you put forth, or 5e raw rules. Still, I think any stealth ruleset would require a non trivial amount of design work.

Consider the fact that Bg3 doesn’t do group stealth checks. If the party tries to sneak around some guards, because sneaking outside cone of vision isn’t an auto success, every one rolls for stealth individually. That includes the plate-wearing, disadvantage-rolling paladin. The party’s stealth effectively becomes the stealth of the least dextrous, which doesn’t feel great.

So then maybe Larian adds group stealth checks? But how does the game know what a group is? Do we have to fiddle with the portraits? In any case, these questions and more would need answering.

If a single roll fails, then one could just keep their heavily armored 8-dex paladin ~30 feet behind, outside of the 360 degree enemy perception range. This is how it's often done in tabletop; the stealthy characters stealth a bit ahead (30-60 ft are the "standard" distances imo, but it'd be reasonable to shorten that to 20-30 ft for BG3 maps). This would work with enemies having either a single perception ring, or a smaller perception ring and a farther "sight" cone. Essentially you'd risk being detected with your tank a turn away for the chance of getting a surprise round.

IIRC they haven't changed the stealth system at all during EA. They reworked light a bit but that's it.
Personnaly not playing much with stealth in games but the system as is is very uninterresting gameplay-wise and only serves cheesing. I really think they will improve it.

A better vision cone, a hearing radius (which is basicaly a second "vision cone") and ennemies a bit more active to search the players would probably solve most issues.

I hear you. On the other hand, the party's quiet tiptoeing behind a guard's back coming to a sudden halt as the clanking of the clunky paladin's armour alerts everyone within a mile is very much a DnD moment. Not being able to sneak your entire party anywhere you want is the price you pay for having a heavily armoured front line, and having no sturdy front line is the price you pay for the ability to consistently open fights with a surprise round. I believe that's fair.


A perception ring to represent hearing/smell combined with a longer sight cone is a great suggestion. There's no need to know who belongs to what group; anyone who enters a perception ring rolls stealth. If they move into the vision cone or the cone moves over them, they roll again. Light level and obscurity are already implemented, they just need to be meaningfully used.

Ideally there would be some increments based on proximity as well, e.g. a creature gains a bonus to Perception DC against creatures sneaking very close to them and a penalty against distant ones. Vision cones could then be made much larger, as they should be. This would put a stop to the githyanki patrol being completely oblivious to you climbing down a ladder right behind their backs and other immersion breaking absurdities that now occur at every step of the game if you use stealth at all. Sadly that would be far less trivial to implement and properly test, especially this late in development.

Ok if you’re using stealth to trigger a surprise round. I was thinking more of using stealth to sneak past an encounter alltogether. In that case, there’s no getting around the fact the paladin can’t stealth well.

Part of the reason stealth is hard to get right is that it has to work in and out of combat. For example, it would seem like making Hide a full action wouldn’t matter oustide of combat. However, it matters in turn-based mode. If the party goes turn-based then group hides, no-one save for the rogue has an action left that turn. You’re stuck with passing the turn and hoping you don’t get found. Wouldn’t it be awkward if Turn Based Mode then Group Hide were significantly different from Group Hide then Turn Based Mode?

This has reminded me that I've been intending to create a megathread for stealth, as I while I know that it's been discussed lots of times before I always struggle to find and refer people back to previous chat when it's raised again. Which was part of the problem with creating the megathread too, as lots of discussion about stealth that I can find is dotted into threads on wider topics about dodgy mechanics and so difficult to separate out.

However, I thought it better to at least make a start so have moved another fairly recent thread on this topic to the megathreads section and am just about to merge this one with it. If you're aware of any other posts and threads that it would be useful to add to it then please PM me.

I'm not necessarily looking for every mention of stealth to be all in one place (that would get unreadably repetitive) but any good discussions of suggestions about it that might be useful for posterity.

On that, we agree. I’ve spent the last few posts pointing out potential issues with stealth rules, but my point isn’t that these rules are bad (they’re not!). My point is Larian don’t have the best track record of fixing their own design.

To me, the best case scenario for stealth is analoguous to the reaction system: the “fixed” version we got in Patch 9 is quite incomplete. So if Larian unveil a new stealth system for launch, I expect it to be promising and flawed. I wouldn’t hold my breath for statisfying stealth until BG3’s enhanced edition.

In that case the Paladin can take off or swap their armor to one doesn't provide Disadvantage on stealth checks. Risk of getting caught with worse armor for the benefit of skipping the encounter. Idk, I think it's perfectly fine that noisy heavily-armored characters are bad at stealth and can get the whole party noticed. This is a also case where you can make use of consumables/party abilities---Invisibility, Guidance, Bardic Inspiration, etc---to buff the Paladin's stealth and improving your chances; something that should be encouraged in a party-based game.

Why not hide first and then enter turn based mode? If you're already in stealth, then you should still have actions available to you, no? Or does BG3 still consider you to have used your action if you enter TB mode within 6 seconds of clicking "hide"?

I agree that ideally stealth should work for both real-time and TB modes. Though, if I had to choose, I'd prefer it works for TB mode and the game defaults to forcing you into TB mode whenever you're noticed from stealth. I also agree that I don't trust Larian to fix the mechanics without introducing other significant issues..

I admit my reason is subjective: I don't enjoy exploring while stealthing. There are a few spots in EA where Tav has an opportunity to spot an ambush just before it happens; if the party is moving forward as I realize there are enemies about, my reflex is to hit spacebar in order to stop moving. (It's the only way I know of to get the party to stand completely still under BG3's movement scheme...) Once in turn based mode, I start thinking about how to ambush the ambushers, which is when I hit group hide.

That's the rationale I had in mind, though in that case ending the turn poses no threat because the ambushers aren't looking for you, they're just waiting.

I don't know, but am curious to find out! Can anyone test this?

Haven't tested it but hiding does use a resource which reloads each turn. Once you enter turn based mode your resource should be returned to the character at the start of your turn. Purely theoretical..

Interesting. Did seom testing. Hide then immediately go into TBM the character still has a BA.
If i used firebolt and entered TMB before the bolt hit i entered TMB without my action.

Thanks for testing! The results are indeed interesting. It feels like an extension of the system whereby characters entering combat after it starts don't get to act until the following round.

Larian have put some thought, then, in the way the action economy interacts with the real-time/turn based barrier. I guess all we can do at this point is wait and see how it all shakes out in v1.0.

I think vision cones should not be visible? You should need to look at the facing of a character and decide if you'll risk moving to a particular spot. Pixel perfect vision cones are really metagamey. That and sound/hearing checks (out of enemy turn order - so player can react and trigger combat). I recall sneaking my entire armour clad party past goblins because I could see their vision cones and they are frozen in time during my turn with no means to react. Or is this no longer possible? I haven't played in 6 months or so.

There have been no improvements to stealth in the course of EA. All our hopes need to be pinned on the full release!

Not that I personally mind visible vision cones, given the difficulty of assessing where a smallish sprite is looking. I just think they should have a wider angle and be supplemented by 360 perception/hearing circles closer to the character, and a bunch of other stuff related to stealth and ambush should be tweaked.

I agree about the suggested tweaks - and hope for the best. With regards to the cones, you can zoom in quite close to see which direction the character is looking in?
I have also encountered cases when I can't actually see where an enemy is looking (if they are partially hidden behind some obstacle/window) - so it seems bit unfair that I can still deduce what they are seeing (exactly) because I can toggle the cones on? I remember making heavy use of this in D:OS to sneak around, but thee it also struck me a little dubious/cheesy

I think disabling the vision cones doesn't change too much. You will then simply wait more often until the opponent looks in the other direction and activate the TBM. At this point I hope that this stupid bug will no longer occur, which keeps you endlessly trapped in TBM mode. That's my bigger concern...

Bugs will be fixed, I'm more worried about design choices, which I think by this point are not going to change.

I'd agree they won't change now, but the last patch was in December so there's been 6 months, quite apart from any changes that were in progress before patch 9 but not ready to release. Given how much rework some of the (in my view necessary) stealth and ambush changes would take, I wouldn't be surprised if they weren't ready to show in December. Of course it's possible there won't have been changes, but I don't think we can conclude that just from the fact they weren't there in the final EA patch.