Baldur's Gate III News

[Black Elk]

Congrats!


Yeah that portrait bug was annoying! I had it happen to both Astarion and Shadowheart. The solution I read was to respec class at Withers, then click "add class" on lvl up (as if to multiclass) but then click cancel. This worked for me to restore the portrait which hasn't weirded out since. Kinda bizarre, at first I got excited because I thought maybe it was a portrait mode, but alas, just a bug I guess.

Anyhow, looked like so after we levelled up on the beach...

You can see how it looked for me, as if her portrait was smaller and offset to the right. I wasn't sure if it's because Lae'zel and Shadowheart were both killed on the Nautiloid? In my first game on patch 8 I got humilated by the Matriarch TPK, so I had to restart anyway.

This time I'm trying a Death Cleric, Shadow Sorcerer for double Tolls lol

Then here we are again at the Gith Creche several levels later. I was playing for a while, like 4 hours or so with the whack portraits before I googled it and saw that trick in some random reddit post. Now at least we'll look sharp for the Zaithisk

So far that was the only bug I saw that I hadn't already seen in the earlier patches. Portrait stuff always catches my eye and grinds me to a halt till I can figure out a solution hehe. Initially I thought my install was borked, cause I was getting a bunch of weird lag, but then I realized somehow my laptop had been reset from Discrete Graphics to Silent Mode, probably when I was using it on the battery maybe, but anyway that was a relief. I thought Nvidia was doing me dirty, but it was just the MSI user mode. Took me a while to figure out though, the cutscenes were all choppy till I switched back. Seems to be ok now though. Fingers crossed!

 

[OrlonKronsteen]

Yeah that portrait bug was annoying! I had it happen to both Astarion and Shadowheart. The solution I read was to respec class at Withers, then click "add class" on lvl up (as if to multiclass) but then click cancel. This worked for me to restore the portrait which hasn't weirded out since.

Another trick, apparently, is to make like you’re going to change classes at level up, and then revert back to the original class (although actually switching to multi should work, too, if you are planning to do that). I tested this twice and it worked, though of course twice isn’t statistically significant.

I just don’t want to play that way with a AAA title, though, and I don’t want to run into a game-breaker after sinking 30 hrs into a run. Glad it’s working for you, Black Elk!

In other news, it looks like Patch 8 may have fixed the awful visual bug with sorcerer robes?

 

[mlnevese]

Hi everyone, I have a question about patches. Obviously, if you install a full patch, e.g. Patch 8, you need to start a completely new game for it to take effect. But does anyone know about hotfixes? Can you install a hotfix and continue with a current save?

Every well written game will place save fixes on the patch. I remember a major patch for Pathfinder Wrath of the Righteous where they fixed a bug in an area and they included a fix for saves for those who had already been affected by the bug. The game would fix the info in the save file the next time you loaded it.

Patches are more of a problem for those modding a game.

 

[Antimatter]

Figured I'd share some player statistics about BG3 vs Elden Ring vs Diablo 4 playtime on Steam.

Elden Ring’s player engagement is through the roof: 45% of its Steam players have played for 100+ hours

Elden Ring’s player engagement is through the roof: Over 45% of its Steam players have played for 100+ hours Elden Ring is one of the most successful premium games of all time.

alineaanalytics.com

How does Elden Ring’s playtime distribution compare to other time-sink RPGs?​

While Elden Ring’s playtime distribution is impressive, let’s not look at it in a vacuum.

We’ll be looking at our Steam data for Elden Ring (15.7 million Steam players), Baldur’s Gate 3 (14.6 million), and Diablo IV (1.5 million). As you can see, there are some distinct player engagement patterns.

All three games are engagement powerhouses. The attention economy is oversaturated, but gamers will make time for quality titles. Case in point: The highest distribution category for each game – by a significant margin – is the 100-500 hour group:

• Elden Ring leads overwhelmingly at 42.6% in 100-500h, a testament to its vast open world, build diversity, and replayability via New Game+ modes
• Baldur’s Gate 3’s 32.3% in the 100-500h bracket underlines its replayability, driven by permutations in narrative choices, romances, and endings
• Diablo IV, while lagging slightly (26.7%), still sees a dedicated cohort, likely sustained by seasonal updates

Of the three games, Diablo IV has the highest share of players in the 10-15h, 15-20h and, 20-50h groups. This suggests that many players are seeing the game’s story through, but aren’t necessarily playing the endgame content and grind loop.

That said, Diablo IV leads in the 500h+ tier (2.3%), suggesting a niche of hardcore grinders. Elden Ring and BG3 – despite their depth – cater more to completists than perpetual players.

Elden Ring was marketed heavily and received incredibly high review scores, but it’s clearly not for everyone. Elden Ring has the highest share of players who played for under an hour. Its notorious difficulty curve likely filtered out casual players early.

These splits reflect each title’s core identity: Elden Ring as a repayable challenge, BG3 as a narrative odyssey, and Diablo IV as a cyclical grind.

 

[WarChiefZeke]

Can't believe I bought this in early access and yet still haven't finished a full playthrough. I have no good reasons for why this is so, I enjoy the game well enough. but when you drop a game this heavy it's hard to pick up from where you left off.

 

[Antimatter]

Baldur's Gate 3 turned two, and Larian shared some of the in-game analytics, very interesting:

Baldur's Gate 3 - Community Update #35 Baldur's Gate 3 Turns Two - Steam News

Happy birthday, gang! Baldur's Gate 3 turns two - hard to believe, isn't it? It feels like only yesterday we were streaming our launch-day Panel From Hell from a cathedral in Ghent and earning a TikTok ban in the process for our druidic love crimes. But time flies. Two years later, Baldur’s...

store.steampowered.com

As for what's next for Larian, here is what Swen said:

 

[Antimatter]

Baldur’s Gate 3 ranks among the best-reviewed games of all time and won Game of the Year from countless media outlets. It has sold more than 20 million copies, Swen Vincke, Larian’s founder, chief executive officer and creative director, said in an interview. That makes it one of the best-selling games ever, which gives Larian the resources to keep expanding but also ramps up the pressure for the company’s next game.

Speaking to me on Dec. 11 in a hotel suite in downtown Los Angeles ahead of the awards ceremony, Vincke said Larian plans to do an early-access release of Divinity, as the company has with previous games, although it’s unlikely to be out in 2026. He wouldn’t offer many specifics about the new game other than to say it will continue to iterate on the studio’s previous work.

“This is going to be us unleashed, I think,” Vincke said. “It’s a turn-based RPG featuring everything you’ve seen from us in the past, but it’s brought to the next level.”

“Baldur’s Gate 3 was a good game and I’m proud of it, but I think this one is going to be way better,” Vincke said, noting that the underlying systems of Dungeons & Dragons were difficult to translate from tabletop to digital gaming. “Here, we’re making a system that’s made for a video game. It’s much easier to understand.”

For this game, the developers are making some big bets. They recently switched to a new engine — in video-game parlance the technology and tools used to make a game — which has led to some growing pains that Vincke says will be worthwhile. The team is hoping to improve their systems for streaming content into the game and doubling down on the cinematic storytelling that worked so well in Baldur’s Gate 3.

Larian is trying to find ways to cut down on development time and aims to finish Divinity in less time than Baldur’s Gate 3, which took six years to make because of its scale and Covid-19 disruptions.

“I think three to four years is much healthier than six years,” Vincke said.

https://www.bloomberg.com/news/newsletters/2025-12-16/-baldur-s-gate-3-maker-promises-divinity-will-be-next-level

 

[WarChiefZeke]

Baldur's Gate 3 turned two, and Larian shared some of the in-game analytics, very interesting:

View attachment 17982
View attachment 17983
View attachment 17984
View attachment 17985
View attachment 17986
View attachment 17987

Baldur's Gate 3 - Community Update #35 Baldur's Gate 3 Turns Two - Steam News

Happy birthday, gang! Baldur's Gate 3 turns two - hard to believe, isn't it? It feels like only yesterday we were streaming our launch-day Panel From Hell from a cathedral in Ghent and earning a TikTok ban in the process for our druidic love crimes. But time flies. Two years later, Baldur’s...

store.steampowered.com

As for what's next for Larian, here is what Swen said:

I find it mildly creepy that games are spying on us to that degree. Leave me alone, man.

 

[Antimatter]

Well, first of all, you're getting notified the developer will use your in-game data when you install the game and accept the EULA.

Then, analytics play a significant role in the purpose of Early Access. There are a few interviews from Larian where they explain how working with in-game data allowed them to identify and figure out potential issues, such as game-breaking bugs, bad combat balance, or even optimization issues (especially bad FPS spikes) etc. I remember they improved Fort Joy in D:OS 2 in terms of even more ways to get out of there, adding player agency, basing their decisions on the in-game analytics.

And of course, eventually this data can be used for marketing/fun/trivia purposes, which happened there.

Support - Larian Studios | What information are you collecting?

When you submit data, the data we gather helps us improve the game. How much gold do you have, when do you level up, where do you die, when do you equip a different weapon. We don't know your name, email address, IP address, or your Steam nam...

larian.com

"When you submit data, the data we gather helps us improve the game. How much gold do you have, when do you level up, where do you die, when do you equip a different weapon.

We don't know your name, email address, IP address, or your Steam name, and we don't know your location. We are only interested in what goes on in the game."

I'd say that overall, eventually, games with in-game analytics, have better quality than those without it.

 

[Black Elk]

Larian is getting dragged pretty hard right now across the board, or at least from what I can gather. Earlier this week it was because of comments about trying to use genAI in their development process. This morning it's about their hiring for the writers room. I think one could probably add to the pile there. Maybe something about an excruciatingly long Early Access period as a way to offload Quality Assurance work onto their EA playerbase? and the post launch tinkering on that front still taking fucking forever, even then lol. Or I don't know, maybe some sour grapes remain from their bouncing on the whole thing at the end, instead of delivering the D&D Expansion/Sequel everyone expected as the follow up to being awarded GOTY (although we did get a dozen extra sub classes no one really expected, which was pretty cool.) I think it's telling though that Generative AI was the tricksy genie in the bottle on this one. I think people are particularly salty there, because it's killing off the arts as a viable profession, and doing so much faster than most people would have anticipated. Especially right now, during the holidays, when we're all being subjected to a torrent of trash AI in commercial advertisements. Quality control is falling off a cliff, and I would prefer to support developers who are doing something more to resist that, or fight back against that trend, rather than just acquiescing to it as the new shitty reality for everything. I'd rather wait 6 years and pay more for a game that didn't use machine learning to speed things up, than a game that comes out in 3-4 years but which has the scarlet diphthong of AI hanging over it like a dark cloud. Really dampened my enthusiasm for their next release, like a kneecapper before the race has even begun. Like honestly who wants to feedback and QA for robots? It just rubs me the wrong way

 

[Antimatter]

Larian is getting dragged pretty hard right now across the board, or at least from what I can gather.

Swen just responded to this, one last time in 2025. I think their response is pretty good:



It’s been a week since we announced Divinity, our next RPG, and a lot has become lost in translation.

Larian’s DNA is agency. Everything we work towards is to the benefit of our teams, games, and players. A better work day, and a better game. Our successes come from empowering people to work in their own way and bring the best out of their skill & craft, so that we can make the best RPGs we can possibly make.

In that context, it would be irresponsible for us not to evaluate new technologies. However, our processes are always evolving, and where they are not efficient or fail to align with who we are, we will make changes.

To give you more insight, we’ll do an AMA featuring our different departments after the holiday break, in which you’ll get the opportunity to ask us any questions you have about Divinity and our dev process directly.

We’ll announce the date in the new year. In the meantime, I wish you all happy holidays!

IMHO, doing an AMA is the best way to address any concerns, questions, and whatnot, considering actual real developers, not execs, will reply there.

I'm with Dan Vavra on this personally. He was the only exec who stood up for Swen and Larian in this case.

Spoiler

This AI hysteria is the same as when people were smashing steam engines in the 19th century. Swen said they were doing something that absolutely everyone else is doing and got an insanely crazy shitstorm. I've even seen someone accuse us of using AI in KCD2. I don't know anything about it, except that I used Topaz Labs to upscale some of the AI elements from KCD1 and some of the old low res textures. I'm no fan of AI generated art, but anyway, it's time to face reality. AI is here to stay with us. As frightening as it may be, that's the way it is. Personally, it scares me the most in the music because you can't even recognise AI there anymore. On the other hand, you know what I hate most about making games? The fact that it takes 7 years and 300 people and tens of millions of dollars to make. And the fact that Tom had to spend 500 hours in the studio recording completely generic heckling and generic bars. If AI can help me make an epic game in a year with a smaller team like in the old days, I'm all for it. That game will still have an art director, writers, programmers, graphic designers, but they won't have to do the tiresome and boring tasks, they'll have to focus on the essentials. I have ideas for lots of games, but I'm fifty years old and so far it's taken me seven years on average to make one game. If AI helps me realize those ideas faster, I'm all for it. As for the RPG and the actors. Wouldn't it be nice if you could ask ANYONE ANYTHING in an RPG? Like, maybe someone for the road directions? Or what they think of their neighbors? We already have tools (11labs can do it) where you script a NPC, their character, their knowledge and opinions and then they talk to you about anything. For non-story stuff, this is an absolutely revolutionary development from a player's point of view. And you can't record it with an actor because it has INFINITE variations. But what you can record are cutscenes and story dialogue. You dont need an Oscar level performance when ordering a sausage in a pub or when asking how to get to the castle. Programmers have a problem. The work of most of them will probably not be needed very soon. We will have software architects, and AI will do the programming. Very soon. Resisting this is probably as meaningful as resisting the use of sewing machines in the textile industry. Or going back to riding on horses while we could use planes and cars. How many horse breeders lost their jobs thanks to Henry Ford!? Who and how will recognize if the game was programmed by a human or AI? And who would want to spend months of work and hundreds of thousands of dollars when the same thing could be achieved in minutes for free? The whole AI revolution may mean the demise of humanity, nobody knows now, but it may also mean that ANYONE, at a fraction of the current cost, will be able to implement virtually any grand idea. Making a game will be as easy as writing a book. There will be more games. We'll see the return of niche genres. Some games will be better. Maybe just a few. There will be also lot of trash. But who cares? There is lots of average trash in books or music for years It will also be an end of most big publishers. And Hollywood as we know it. Nobody will need their money and resources anymore. What we saw with boom of indie scene thanks to Steam, we will see with AAA games thanks to AI. Or Skynet will destroy us before all that. Either way, there's no stopping it. I used AI to translate this text from my native language Let the shitstorm begin!

 

[Chronicler]

Yeah, that sounds like the accusations are pretty much accurate.

Doubling down that your detractors are Luddites who would've opposed the steam engine just kind of reinforces the point.

 

[Chronicler]

If they're not using AI, it'd be much simpler to just say that instead of posting some small manifesto about your corporate values or whatever.

Anything beyond "No, you misunderstood, we're not using AI" is basically just debating whether it's correct or good to use AI, which is a debate happening all over the place, but doesn't really change anything about whether the claims of their AI use are accurate or not.

And given that this debate is already happening all over the place, with the talking points having pretty much been repeated ad nauseam at this point, and with people broadly having come to their own conclusions now, I'm skeptical that the AMA is going to convince a whole lot of people to change their stance.

 

[WarChiefZeke]

Well, first of all, you're getting notified the developer will use your in-game data when you install the game and accept the EULA.

That doesn't make it better, really. I have to accept the EULA after I already paid 60$ for the game. I shouldn't be forced into giving up my privacy rights because I made the mistake of paying money for a video game. Corporations of all stripes have gotten way too comfortable with utterly invasive and dehumanizing practices, but that is an entirely irrelevant discussion.

 

[Chronicler]

Back in the day, protecting yourself from "Spyware" was basic internet safety.

These days, "Spyware" isn't even really a word people use anymore, because pretty much all the major software is spyware. We'd sooner need a word to clarify when the software specifically isn't spying on us.

 

[Black Elk]

The contrarian in me wants to push back against this idea that they're only trying to eliminate the laborious tedium, and only so they can expedite the release schedule.

If they say the aim is to cut development time in half (from 6 years down to 3 years) then that must mean the final game is going to cost 50% less right? Right?

Lol

I mean we know how its going to go down. They'll shitcan like half the staff, expect whoever remains to finish on the same timeline. The savings from payroll will go directly to the shareholders or the chief executives or whoever. Would a potential BG4 by some mystery studio only cost us 35 dollars instead of 70 in this new future where nobody is doing the literal grunt work anymore?

I prefer my grunts and claps and chunk out thud sounds to be fully like acted out by professional professionals. Human beings, with some gusto behind it. I want my CA and background environments and character designs doodles to be doodled and noodled by actual people. It makes a difference. The coffee table I just bought at walmart lasts what like 3-6 years tops? When we might still have some handcrafted thing built by shakers that gets passed down through the generations cause it's quality is just all boss like that. They just convinced me that games could be artworks, like by artisans, but now its headlong on to the mechanical LOOM?

I think it would be nice to at least just hear that they're as pissed about these changes as everyone else. You know like a good response might have been 'well we know it sucks, but we're being forced into this stuff by the billionaires, so taste it! I guess.' Like least that would seem more honest lol. Pretending its for the rank and file staff, so they can like take Xmas eve off and get a turkey for tiny tim or whatever, I mean I just don't believe it. What if Machine Learning learned nothing from Marley's Chain party control and so we're stuck with that forever? We're gonna get goosed. They'll probably just scrooge it if we don't grumble every single time it comes up

 

[Antimatter]

That doesn't make it better, really. I have to accept the EULA after I already paid 60$ for the game. I shouldn't be forced into giving up my privacy rights because I made the mistake of paying money for a video game. Corporations of all stripes have gotten way too comfortable with utterly invasive and dehumanizing practices, but that is an entirely irrelevant discussion.

Actually, you can check out and read every game's EULA BEFORE making a purchase. It's linked right there on the store page.

See, this is for Steam, for example:

That "Baldur's Gate 3 EULA" is clickable and links to the document. https://store.steampowered.com//eula/1086940_eula_0

So if you worry about such stuff, you can always do due diligence before making a purchase.

 

[O_Bruce]

I'm with Dan Vavra on this personally. He was the only exec who stood up for Swen and Larian in this case.

I am not with Dan Vavra on this. The technology itself is unethical and inherently so at that, since it is trained on works of hundreds of thousands, artists, be it visual artists, musicians, voice actors etc. without consent, compensation, acknowledement etc. It doesn't matter how you use it, it doesn't change how this technology came to be.

It also is not my problem that nowadays videogames' budgets are overblown. It is not my problem that Dan Vavra doesn't give a single crap about authenticity. It is not my problem that Dan Vavra doesn't understand that autistic number of voiceovers and insignificant details isn't necesssary for the videogame to be good - people don't play videogames to count pores on NPCs skin or experience hundreds of dialogue options, with each single one of them being pointless. It is not my problem that Dan Vavra doesn't understand that having an idea is nothing unique. He also doesn't understand that an initial idea is usually a bad one, and thus creative ideas require development. Shitting out every idea you have via generative AI will just result in slop. For creative people, this is common knowledge. For Dan Vavra, not so much.

"Making a videogame will be as easy as writing a book". Sure, maybe. But writing a good book is a lot harder than someone out of touch might think.

Yes, there is a lot of bad art, music and so on. What Dan Vavra doesn't understand is that those "bad" works are necessary in order to eventually being able to make "good" work. Bad art is a stepping stone towards good art. Even great artists produced tons of bad art at some point. With art, the bad we have is there for a reason, at least part of it. With generative slop, it is only bad. It won't make people better at it. If anything, it will only make it harder to find authentic human creations. Hell, it already does that.

We won't even start talking about how outsourcing your cognition to AI does to your cognitive funcitons. We won't talk about AI psyhosis. About envoirmental factors. About it effects on marketplaces. AI might be the future, but it will be a worthless future. Not to mention, if this technology was so amazing, there wouldn't be need to showe it into everything regardless of customer feedback.

So, fuck you Dan Vavra, with everything I have, including a cacti.

Now, with that is over, there is a lot to say about videogames, their budgets, crunches and so on. Customers are not responsible for balooning budgets, budgets mismanagaments, demands from marketing department or anything shareholders say. If anything, businesses should remember and go back to the basic: you are serving customers first and foremost. That's it.

 

[Antimatter]

Well, I don't want to argue much. You have good arguments, but I think I can see where Dan is coming from on the issue. I'll say this for the record:

- KCD and KCD2 were created without AI
- They are absolutely amazing games with fresh things
- A lot of people worked thousands of hours to create them
- He is very proud of authenticity; these games wouldn't even exist if he weren't
- He does understand how VO works; these games have an insane number of VO lines; actually, the game has a Guinness record for the script length
- VO work in KCD2 is insane
- He does understand a lot about creative ideas; he's a game director on these games, and they have a ton of creative solutions
- He is also a writer who has written a lot in his life
- No, he doesn't deserve at all to be sent to hell. His works deserve recognition, respect, and, actually, more players playing them
- He didn't suggest shitting every idea via genAI
- He didn't suggest outsourcing cognition to AI

But all of this is off-topic. Swen also didn't suggest any of that.

Swen, Kojima, Vavra, are all people in their 50s. They don't have a lot of time left to create games. I can understand how they worry they wouldn't be able to create more than 2 games in what is left for them. Every game company is also a business. If other competitors start using something and achieve something, you need to compete.

 

[O_Bruce]

Before anything, I can totally understand that when there is a lot of good faith build, you can forgive some things or have much more positive interpretation of Dan's words. I get it. This is just not the case for me, since there was no good will build there beforehand.

With some things, like two of last points (about ideas and cognition outsourcing), I specificially said "We won't even start talking about how (...)", exactly because Dan Verva haven't mentioned those factors - either out of ignorance, or for the sake of his argument. That doesn't change the fact that those factors are always present with gen AI. If generating concept art, scripts, ideas, voicovers etc. with gen AI isn't outsourcing cognition to severe degree, then I don't know what is. And because it is so fact, convenient and easy, there will be a lot of generated things by default.

About ages of people like him and Kojima, ok. They're getting old and want still more to be done. Too bad it is while not giving a damn about consequences, in the long term especially.

And about competing in a marked, there are alternative strategies, on is very underutylized and I call it "Steam strategy". Let your competition make terrible decisions and suffer consequences. You, on the other hand, just offer your customers good service. Is it really worth chasing artifically sustained technological bubble that will burst sooner or later?