tl;dr inZOI is an empty prototype that focuses way more on AI technology than anything fun or meaningful, don't buy it right now. maybe when the game fully releases it'll be better but the direction seems kind of flawed from the start.
this game is extremely early access.
think of sims, except instead of a team of people very deliberately putting lots of stuff together for tons of different contexts, it's all run by a shitty AI.
the foundation is fine, and i do like some of the ideas the game is going for, but it really seems like they focused way too much on random AI features instead of making an actual game.
the graphics are also pretty on the surface, but completely fall apart in motion. no matter what settings you're playing on it's going to look bad. i really, really hate modern rendering techniques. sims 2 looks better than this shit because it's actually a stable image and it's pretty clean too.
definitely DO NOT BUY until they really get more stuff here. there's barely anything to do and there are only some extremely specific things you can interact with. going to school is a rabbithole, which is horrible when you consider that the in-game time even on the fastest setting is ridiculously slow. i've got it set to 48 minute days, the lowest one, whereas sims 4's default is 24 minutes. if it could be sped up more that would be fine, but it still goes by really slowly especially when there's like nothing to do. NPCs seemingly never talk to your character, you have to initiate every interaction and they'll often just walk away after you say one thing unless you queue things up.
it does kind of feel like this game is leaning way more into AI features than being a life sim, like why in the absolute fuck did they spend so much time developing AI that can track movements onto a character? there's no point whatsoever.
played for not even 5 hours and yeah, it just made me want to play sims instead. there's really a whole lot of nothing here and that's a pretty good way to describe it. there's a lot of shit, it's just that none of it is actually meaningful or valuable at all.
paralives can't come soon enough.
@beardalaxy I think they really wanted to focus AI because otherwise custom content would require users to have some basic scripting knowledge, which is extremely uncommon for the target demographic of this type of game. It's a brilliant idea in theory, but in practice it results in a ton of bugs and AI hallucination and as @SuperDicq said, if effectively renders the game SaaS, otherwise it would require a REALLY beefy PC to run.
Though AI facial animations are just some useless gimmick for creating social media videos.
@xianc78@gameliberty.club @beardalaxy@gameliberty.club Also I'm not sure about this game because I obviously haven't played it but from a game designer point of view my mind really finds it baffling that they seem to use machine learning mostly for animations and asset creation?
Personally if I would've developed a Sims-like game with machine learning models I would instead use stuff like LLMs for actual gameplay like how the Sims interact with each other, what they are thinking and creating emergent gameplay.
@xianc78@gameliberty.club @beardalaxy@gameliberty.club
otherwise it would require a REALLY beefy PC to run.
You know these kinds of things aren't a "one or the other" type of thing, right?@SuperDicq @xianc78 the game does run its AI locally AFAIK but I'm not sure. You need a 12gb card to use the smartzoi AI option but my card is only 11gb so I don't have access to that and I guess that makes all the zois retarded because they don't have much real programming.
@icedquinn yeah the whole game is "isn't our AI tech cool?" and then you go "yeah it's kind of cool" and then that's it.
the graphics could be nice but they have a lot of problems, like a lot of other modern games do. it is really distracting when things are leaving afterimage trails and flickering.
@icedquinn from what i understand, the post processing and lighting don't update at a high enough rate so when things move quickly (the camera, objects, or lights changing), they leave trails. in order to make them update faster, though, it takes significantly more horsepower. that's why all this stuff has always been done in pre-renders instead of realtime. it looks great in slow scenes and still shots, but completely falls apart in fast shots. it's why i can't use raytracing in most games, the lighting is great but don't move your camera too fast i guess.
this happens with any temporal effect like TAA too, because the applied effect doesn't apply fast enough to the updated image.
so naturally the idea would just be to turn off AA but then it turns into a pixelated mess because the level of detail is so high that you NEED anti-aliasing to smooth it out. you can use FXAA but it's not that great at doing what it needs to and you still have things like the light flickering because that's all based on weird ephemeral temporal shit too.
modern rendering engines like this just suck ass. they are aiming for something so far in the future that is currently unattainable. these games might look passable in 20 years but for now, there is just way too much distracting shit going on in the image for me to say it looks good. games from 2013 look 100% better.
@icedquinn that was the reason behind using just a blur (fxaa) or temporal effects (taa/tsr). turning off anti-aliasing is all well and good but unless you're running at 4K (and sometimes even then) you're going to run into issues with things being way too noisy.
i would rather take a performance hit for MSAA than use any of these new ones. DLAA isn't as bad but it's also a pretty decent performance hit.
@beardalaxy@gameliberty.club Other than it being proprietary slop. It also depends on online services to function so this game will definitely stop working a few years from now when it becomes no longer profitible to keep hosting the expensive machine learning models for this. So don't waste your money.