so that is what it is

This LoRA appears, at best, to have been trained on poorly captioned images, since the keyword alone does not actually generate images akin to the results shown.

The LoRA's abilities seem restricted to close-up shots, and you only get those if you include additional prompting that is not explained in the LoRA's description (but is included in the caption for the close-up example image). The image showing the broader view (which doesn't even show a bussy) has no caption.

Anything other than a generating close-up using the exact prompting language in the closeup example either fails or yields mutant parts.

I appreciate the creator's effort, but this feels very half-baked.