Can procedural generation ever produce meaningful content?

Can procedural generation ever produce meaningful content?

Other urls found in this thread:

sciencealert.com/a-novel-written-by-ai-passes-the-first-round-in-a-japanese-literary-competition)
twitter.com/SFWRedditImages

>meant to make everything different
>just makes everything exactly the same

You tell me.

Most Roguelikes are pretty fun so yes.

No

Daggerfall
>small as fuck dungeon with only two rooms
And then there's
>huge as fuck dungeon jesus christ i've been in here for hours I should've used my teleport anchor
Then there's everything inbetween

Yes

Dorf Fort is the best game ever, so yes.

/thread

Not now, in 5-7 years, after Google will really dig in machine learning, we'll see something amazing. Or terrifying.

Can Fear ever be a good player again?

yea, but they're going about it wrong.
they need to make it so it's dynamically seeding based on an origin point and every single neighboring region effects the other region in significant ways when it's being generated.
like, it just needs to do a hyper accelerated full on simulation of the zone to generate the region. using stuff like water erosion, human travel paths down roads etc.

Not until he finds a competent team

What happened to EG, they used to be so good?

Shifts in the meta have fucked over literally every TI champion in Dota history.

it's the nature of dota 2. patches are always screwing with the meta and it hits some teams harder than others and then those teams prosper then another patch and those teams are now shit and... i don't know why people still care about esports anyway

What's so hard about adapting to a new environment? Are all TI champions recent players who have started playing and won the tourney during the same patch?

They changed their roster after winning the biggest prize pool in video game history

Why? I have no idea

But its universally agreed that Aui is a shit player.

The misconception is making procedurally generated graphics, not gameplay.

Sumail sucks dick now. PPD needs to kick him out.

he wasn't bad at pos4 back at ti5. he got kicked because he annoyed ppd too much in the finals.

The universe was created with a few laws, and look on Sup Forums, so, no, something procedural generated can't produce something meaningful

It's VERY hard to adapt with the way Valve handles the pre-TI buildup. Bugfixes and minor balancing tweaks aside, they stop patching the game like one or two months before TI so that pro players can get comfortable with it and be ready for the tourney. Because of that, they develop a lot of habits and strats which thrive specifically in that meta and have trouble shaking them TI is over and Valve drastically changes the game with a new patch.

...

And then EG collapsed and hasn't recovered b/c ppd is such a beta he can't handle bants

Sure, just depends on how well you make your generation, spelunky does it really well.

no

Handcrafted levels > generated levels

>he doesn't remember radiant ai and quests
the tech just doesn't exist

If by meaningful you mean portray the artists message then i think it can, mechanics can often do a better job at giving a message than level design.

nah you're just the right type of autist to enjoy that mindless shit

yeah definitely. it's silly to think not. give it some years.

EVERYTHING BECOMES THE SAME AT A LARGE ENOUGH SCALE DUDE

Have you seen any of Aui's streams? The dude just rambles about nothing for the entire game. Imagine having to put up with that shit for 8 GAMES in the final day of a tournament where millions of dollars are on the line.

>EVERYTHING BECOMES THE SAME AT A LARGE ENOUGH SCALE DUDE
lol ok. that doesn't stop procedural generation from being incapable of creating anything that isn't repetitive copy pasted garbage

Well i'd hardly call myself autistic, as i said spelunky does it rather well, it creates more traditonal levels using preset chunks made by the developer, it also randomly places clutter like rocks and stuff around the level, it gives an interesting layer of resource management combined with the traditonal mastery of levels you get in older 2d platformers.

also aui was only a good support during the 4th core meta

he's garbage now

As discussed in a thread earlier today, good levels and maps aren't difficult to come by with numerous games featuring good procedurally generated areas. What many games fail at is inadequate simulation and lack of robust mechanics that make this generated environment interesting to interact with.

If you don't just want good content but some "meaning" to that content then it gets more difficult. Expert systems have already performed well in the arts (composing programs can create compelling music, a computer even did well in a novel competition: sciencealert.com/a-novel-written-by-ai-passes-the-first-round-in-a-japanese-literary-competition) but creating good story content to a game doesn't require just an expert system but probably requires artificial general intelligence.

Some games have still had some success with carefully hand-tuned algorithms. For example, Dwarf Fortress has its generated history and one of the dwarves might create art depicting some of the events from it: it's all very formulaic text description so you can easily tell it's algorithmically generated but it does make the game world feel more compelling nevertheless. There's also stuff like NationGen for Dominions 4 that generates random nations under some constraints: For example, each generated nation tends to only feature one "species" and instead of fully randomizing magic for each individual mage of the nation, it sets up a theme ("this is a nation of earth and air mages who wear green robes and tend to ride giant frogs") so there's some consistency. Some seeds can end up with inconceivably OP nations, but with some vetting by the admin, these nations can be playable just fine in a real game (often times with players submitting their own fluff to the generated nations).

Adding to this, this is why Mincraft was so successful. 100% of Mincraft's gameplay is manipulating the environment, and therefore having a limitless/always different environment is why the game works. Procedural Generation's limited ability to create compelling or interesting scenarios, rather than purely environs, is sidestepped.

Never played Dominions, i assume your pic is a example of "OP nation".

How much is that OP compared to your average vanilla nation?

That's actually an overpriced although capable unit. I looked for silly units from a batch of generated (and vetted) nations and a swan-riding elf was the first one that caught my eye. There are some OP factors that are intuitive (for example, immortal units or a cheap unit that turns to a powerful fire elemental upon death) but some wouldn't be apparent without knowledge of game mechanics (for example, it's highly desirable to have a large amount of attacks per square, so small units with multiple attacks for example would often be highly desirable).

It's almost unprecedented for games to mix vanilla and mod nations (whether it was made by someone or generated). It's also worth noting that it's a free-for-all (only one player wins, a typical game has maybe 6-16 players) so there's balancing factor from players being inclined to gang up on nations consider strong or annoying to play against (a classic example is native nation Ermor, undead Romans, whose religious influence causes population to die, and since invading a barren wasteland requires some logistics and provinces you capture are worthless, it's generally speaking a good idea to kill them off immediately, especially considering Ermor itself is unaffected by cancerous global spells they're likely to cast if not removed).