A collection of random tests and experiments, across multiple gen types to figure out how different models handle input in tags, captions, and materials.

Anything good might eventually get shunted into solo models.