No Man’s Sky is both a cautionary tale and a redemption tale all in one, and Hello Games is not giving up on its title any time soon!
I would love for some more variety on the one planet. I feel like once you’ve landed and explored a 100m radius you’ve basically seen everything the planet has to offer.
How cool would it be to land in a desert, then takeoff and fly across jungles, heading to the poles for snow etc.
They said, the updates they’ve made have been amazing. You can land on water and fish from your spaceship now!
some of my best planets got screwed in an update and I stopped seeing any point exploring.
of course fishing is the only thing they ever needed to add for me to go full Andy, I’ll have to re-install now.
So this happened to me as well, all the planets I had named reverted back to their original name but were still founded by me…so I thought why bother? and then it just lost its appeal.
I never cared but that would really suck if you did.
I was talking about the biome changes. Lot’s of nice planets turned to trash. My first paradise earthlike with zero weather became some pink, yellow, and peach hellscape with constant extreme storms.
The extreme storms are just time wasters…not immersive at all when you sit in a river for 5 min, or just go afk, make a sandwich & come back for 2 minutes before another storm hits.
Or how about really radically different planets? Right now they are pretty much all the same. Just a vast natural landscape of $BIOME with the occasional point-of-interest dotted about. Where are the city-planets? Factory planets? Planets strip-mined to the core? The giant space wharfs? How can there be trillions of aliens but there are never more than 30-ish in the same place?
I wish they’d make proc-genned interiors for certain planetary buildings, or at least create slightly different architectures. Every abandoned research station, every small outpost, every trade outpost, everything everywhere in the galaxy is exactly the same. Even the ruins you can excavate once in a while are always the same sandrock, tilted walkways with some chests
I think they’re overcompensating.
all this effort should be put into a sequel using generative AI. because this is one of the few perfect use cases for it.
What advantage would AI generation have over procedural generation?
it doesn’t take long to see everything procedural has to offer but you could feed a biology textbook to an AI model and spit out unique creatures all day
Maybe, but would that improve the gameplay experience? In my opinion, if you can generate a million unique creatures or a billion unique creatures doesn’t matter all too much, because I can’t imagine that the average player will ever see more than maybe a thousand. Even as it already exists, it’s pretty much impossible for a player to experience the full breadth of No Man’s Sky’s potential. It would definitely be an impressive technological accomplishment, but not one that I think offers anything substantial to a player.
Personally, I think AI’s use in games is better-suited toward NPC dialog. There’ve been some tech demos showing this, and it still needs a lot of work, but shows a ton of promise. I think that given the right parameters and limitations, there’s more potential for gaming on the LLM-side of AI than anything else. I’ve not played NMS since launch, but I don’t think the current version has much in the way of character development, but that could be something that a sequel could make use of AI for in a sequel, I think.
Not seeing the full extent is not the issue, the issue is not seeing basic variety. They need to let their algorithms run more wild with what they generate, imo. If every animal looks roughly the same in a game about exploring a vast universe, then the algorithm has failed and the game suffers massively for it. I haven’t played enough NMS to say if that’s an issue with the flora and fauna, but it is an issue with the planets.
Setting aside the merits of an AI-generated world, that shit is computationally expensive and already causing environmental issues.
They’re working on their next game already, Light No Fire. No genai, as far as I know, which is a win. The algorithmic approach they are taking is already good enough, it just needs to be used to generate biomes (which they are doing in the next game).
it’s been a fun journey so far. cool
