Alora

Thanks for playing this challenge. Beautiful creations. See you soon 🖤 🖤

Wow, that's a pretty good result for full model wts, no negative prompt, and pretty complex prompt! If I may offer a few suggestions... where you have the [LoRA 1] [LoRA 2] etc... remove those brackets with text and insert the actual LoRA token codes (the portion that consists of the "<" ">" marks) in those places instead. Also, adding the default negative (or tweaking it with your own custom words) can really help clear things up. Finally, 4 LoRAs at default wt of 1.0 almost always turns out "overcooked". The mix of models usually fight for the spotlight and one or two models will sometimes shine, but often, it's just a blurred mess. Adjusting the number in the token code to something less than 1 on all the models will help them to work together a bit better. With a mix of 4, I almost always drop everything 0.5 and see how that goes (which is usually pretty well) and if one model outshines the others, I will make that one even lower, like 0.25, or I'll increase something to 0.75