Here I used a merge of this model with DieselpunkCity to show the versatility of these models is retained even when you merge several together, allows for a high degree of artistic variability that does not have to be in abstract.

If you look at the prompts, all the Nicholas Cage portraits are not using the Trigger Words. Of course you would need an embedding with 2.1 to reinforce a famous person.

The other images of the cities are to showcase the effect of merging these style models with the trigger words being used.

Re-ordering the Tokens and use of weighting/negative prompts will also affect your outputs as shown