Same dataset as mega liminal, with the megaliminal class removed.
Prompt is courtesy of Eicosa because I've ran this set 15 times and can't be bothered to relist the prompting, these are the classes.
Prompting
liminal, <class>, <two to four plain sentences describing the image>
The classes are: landscape, suburban, simulacrum, vanishing point, empty mall, parking garage, cityscape, movie theatre.
The suburban images were sourced from a context ran on civitai a while back, thank you everyone who contributed.
The model will allow you to create complex landscapes due to the selection of images. Vanishing point works most of the time, on models that cannot do vanishing point. The prompts are complex, so most of what you prompt will be affected by the styled conceptualizations.
Many common words were used, and the prompting was specifically curated to be diverse plain English.
Have fun.

