|
@lorenschmidt | |||||
|
it could be worth it to spend some time thinking about what weird rules you could make for your world to fit it to the nature of the simulation harmoniously.
i always think about the way simcity 2013 had simulated people, but they weren't attached to individual houses.
|
||||||
|
||||||
|
loren schmidt
@lorenschmidt
|
31. sij |
|
thinking about an approach to live, simulated generated worlds which uses nested homeostatic units. a city can be run as a single homeostatic unit if you're zoomed out. if you zoom back in, it will "catch up", making new buildings, changing where fields of crops or buildings are.
|
||
|
|
||
|
loren schmidt
@lorenschmidt
|
31. sij |
|
the basic idea is that each container is zero sum. a planet has so much food, so much metal, so many people, etc.. and over time it can exchange with neighboring planets + update its simulation at planet level. there can be droughts, or plenty, or emigration, run at that scale.
|
||
|
|
||
|
loren schmidt
@lorenschmidt
|
31. sij |
|
then if you zoom in, looking at individual groups of people or cities or towns, it decides how the larger scale changes it has simulated should affect those finer homeostatic units. maybe the planet gained 30% people in the coarse simulation. it now has to actually place them.
|
||
|
|
||
|
loren schmidt
@lorenschmidt
|
31. sij |
|
and if you add people to a city, you have retroactively make dwelling places, farms, etc. which fit the new homeostatic state of the parent container.
|
||
|
|
||
|
loren schmidt
@lorenschmidt
|
31. sij |
|
you'd lose something without continuous granular simulation. some phenomena are granularity dependent and cannot be simplified at large scales. but if you played to the strengths of this you could have a huge simulated world which could truly change in large ways over time.
|
||
|
|
||
|
Steven Portzer
@stevenportzer
|
31. sij |
|
One way I could imagine this working is to simulate areas at smaller scales for a short period of time and extrapolate relevant metrics for larger scales and longer time periods. The simulation could spend more time more zoomed in if it detected anything interesting happening
|
||
|
|
||
|
loren schmidt
@lorenschmidt
|
1. velj |
|
oh interesting. yeah this could be really interesting and useful for certain features. for instance, if a town determines a new building needs to be made, not actually placing it could affect things like traversal time to the market, etc. if you don't step down and do so.
|
||
|
|
||
|
TangerIIIne
@CIIITRU
|
1. velj |
|
One could be speed: when zoomed in it runs slowly, and speeds up as you zoom out to the next layer. This would let you ignore short term small scale events as noise. And after zooming back in you would see a completely or partially resimulated landscape (as time has passed).
|
||
|
|
||