[MUD-Dev] Morphable worlds, Reset based systems revisited

[John Robert Arras]

On 26 Oct 2002, Ola Fosheim Gr=F8stad wrote:

> Simply expanding the existing world has 3 unfortunate side
> effects:

My only problem with changing the world is that I may lose my
equipment and my house and everything else that I've created in the
world. I would want the familiar world left there.  If those things
can get left in, then it might be tolerable, but I don't know.

> No, I wasn't. Algorithmic content is only interesting if good
> artists make the selection (unless the designers are esthetical
> programmers which they most likely are not). So, yes I am of
> course implying good tools. Good tools, good basic elements.
	
> Procedural generation pose no problems though, that is just a
> parametric model with meaningful parameters (height, breadth,
> width, skin etc). So, yes, that should be included in the toolset.

Ahh. I don't make a distinction between procedural and algorithmic
generation. I consider content creation to be on a sliding scale
from handcrafted bit by bit to totally computer generated. If I were
working on a graphical game, I would look to have tools that let me
"generate sword <parameters>" and then look at what it makes.  Then,
pick something interesting. Similarly, I would generate areas using
parameters and then see what they look like. I would consider an
object or zone generator that makes results that are checked by hand
to be procedural and algorithmic. It's a question of how whether the
builders recognize "good" patterns or have to create every little
thing in the area.

>> I think I will get good results when I am able to come up with
>> enough of these details that each area can have a few interesting
>> things, but I won't have to be so repetitive across areas.
	
> Only if you have a human moderator of whatever your algorithm
> outputs. Or let the users rate the areas produced.

Probably, yes. :) I also envisioned people cleaning up the areas
after they were made. I am also restricting the code to making
wilderness areas only and within certain restrictive parameters, so
the quality won't deviate that much from one area to the next.

>> I also know that a lot of people don't like this idea, but I'm
>> terrible at building. The code I have now makes better areas than
>> I could make by hand, and it will be interesting to see how far I
>> can push this.

> Sure, you could let your users "nuke" bad areas. You can actually
> produce relatively good pictures using genetic programming (look
> up "genetic art" in a search engine). The good/bad ratio is likely
> to be 1/100 or worse though... (For NPCs, look at the works of
> Karl Sims) 

I don't think it will be as bad as that, since the areas have very
restrictive parameters involved in their creation. At the same time,
that means they will all be on the same level, which means they may
all end up boring. I haven't tried this out with real players yet,
so I'll see how it goes when people see it.

John


_______________________________________________
MUD-Dev mailing list
MUD-Dev at kanga.nu
https://www.kanga.nu/lists/listinfo/mud-dev