Returning a bit with this cityscape model from MJ creations, Obviously results can vary with different checkpoint, where realistic models gives minimalistic results and anime ones giving them most complexity. These results are from prompt reversing of some images used in dataset.