Sciamano240 - Artist Style 🎨

This LoRA is trained on an isolated sub-set of works made by Sciamano240 trying a old variation of their art-style, where it had a unique semi-real look to it.

If you like this style consider supporting the actual artist behind it 💚

Usage Tips (IL) :

• I personally use it a weight of 1 so i recommend that

• the LoRA needs to have its trigger word "nv-st24-il," in the prompt

  • I also add the style name (sciamano240:0.3), at a low weight to make it look even better

• so far only tested on WAI-NSFW-illustrious-SDXL V14 / V16 (used in the showcase pics)

  • the closer the checkpoint you use to the base Illustrious model the better the effect

• this lora is realy bad at handling male characters, as well as far away face

Project Notes :

Im still amazed by ILXL's ability to carry all those style innately, i almost dont have to train it at all xD, tho there are still instances of where its misses its mark due artstyles (innate ones) having non-uniform evolution over the years, or simply when the artist just happen have alot of different artstyles.

thats why in the recent LoRAs i've been experimenting with style isolation where the base model just knows to much about said style and introduce a LoRA trained on a small curated dataset to help nudge the final gen closer to that needed style.

Also now I'm avoiding using the per-trained style name as the activation word as it might introduce alot of bias from the model, as well as ending up having less control over all.