i love all these little diorama creators that have popped up recently, they make it very easy to create a city that looks good. But they only hold my interest for so long. i'm looking for something with more meat on it. Any recommendations?
as an example, i remember the first time i managed to keep a city of over a million people going in Sim City 4. at this point money was tight, so the building aspect took a back seat to actually managing the city. balancing the budget, fixing congestion, and so on. it was great fun and a very different challenge than i thought i was in for.
most citybuilders these days seem more focused on the building than the older ones. for example, when i got to the point in Cities Skylines where i thought i was entering the "management" phase, i unlocked a building that just removed an aspect of the game. it was like the game thought that planning the electric grid or schools was a chore that got in the way of building a city, and as a reward it removed those chores.
basically, i'm looking for a game where rather than physically growing the city through placing individual buildings, i help the city grow. like transport tycoon, except the city is the focus rather than the interconnections.
a key part of this, i think, is time. a city that is frozen in time and where clicking with a tool just builds things, like C:S or SC2013, doesn't make for interesting growth. a city designed around historical limitations feels more like something that needs to be managed. a game where buildings and roads take time to complete and modify requires more forethought.
workers and resources comes pretty close but the central planning aspect means that i still need to micromanage the buildings. if it was all about zoning, with special buildings being unlocked by the request system in older sim cities ("x seeks permission to build a stink generator downwind of your residential area") i would enjoy it more.
not necessarily. image generation models work on a more fine-grained scale than that. they can seamlessly combine related concepts, like "photograph"+"person"+"small"+"pose" and generate plausible material due to the fact that all of those concepts have features in common.
you can also use small add-on models trained on very little data (tens to hundreds of images, as compared to millions to billions for a full model) to "steer" the output of a model towards a particular style.
you can make even a fully legal model output illegal data.
all that being said, the base dataset that most of the stable diffusion family of models started out with in 2021 is medical in nature so there could very well be bad shit in there. it's like 12 billion images so it's hard to check, and even back with stable diffusion 1.0 there was less than a single bit of data in the final model per image in the data.