Needs work, but you can see what I'm going for here. prompting A &amp; B: Do not use the score prompts, they will break it. For the lora to activate the prompt needs 'packetstyle' and words like 'glowing eyes', 'indoors', 'solo' B was trained with more images that also included monsters. It produces different results. I feel that most of the time A is better. prompting C: Same training set as B. No trigger words required, but it's also not as strong.Despite the added difficulty in prompting, I feel that A produces the best results.I'm really interested in developing this one further, and after all the testing and training I've done the past month between Pony and FLUX I can understand why people are begging for buzz now. Speaking of, once I'm happy with this, I may try to make a FLUX version. My recent experiments with a single-image FLUX lora have shown that it takes to loras really well, so this should be entirely doable.

Toon Horror

More images including monsters in dataset, trusted auto-tag more this time.

Files with SHA256: f0f7399ed5838ed653798fc50392b324fb166e2571da5eb75a6ee72237ce85b4