Fantastic results with a descriptive prompt!

If you just do a basic "make this look real" type of prompt, the results can range from boring to awful. As soon as you get into the prompt and describe what you expect to see, this LORA does an incredible job of helping the model get the details right.

Side note, the grammar of the trigger phrase is incorrect, so I wondered if that affected the "without LORA" comparisons. For the record, I tried generating "without LORA" options with the same seed and slightly modified prompt so the first sentence used correct grammar and... IT MADE NO DIFFERENCE. The style change without this LORA always looked awful, regardless of the prompting.