Obligatory Disclaimer: This LoRA is intended to be used alongside style LoRAs and/or artist style tags to bring out its full potential. All example images use WAI-Anima and the "haku89" artist style tag.
If you don't see all 4 example images, try changing Browsing Level to "R" or lower.
Comments and gallery posts are encouraged and appreciated regardless of their content! (Within ToS, of course.)
A character LoRA for the ASMRtist OC Bow Elepha. Trained on a dataset of 40 images that consists largely of cut up screenshots, images generated by ChatGPT, and a few official arts.
I've changed my LoRA training settings to hopefully allow for more flexibility. Total steps have been almost halved, I've started using Prodigy, and began training on WAI-Anima. That being said, it's still not perfect and the are definitely still issues. Most notably, there is some inconsistent generation in regards to the artstyle. The dataset contains images of her in the original more "cartoony" artstyle in addition to images with a more traditional anime style. As such, it can swap between them if you don't take steps to prevent it. This is especially prevalent in full-body generations where it seems to favor the anime styling more. The tag "simple lighting" appears to trigger the more cartoony style more often while omitting it or using the tag "detailed lighting" appears to trigger the more anime style. It also has a habit to tuck her sweater into her pants. If that's not something you want, be sure to prompt accordingly.
The below and attached triggers are general suggestions. You can almost certainly get better and more consistent results by more specific prompting. I'm just lazy. Play around with it to see what works for you. Remember, Anima loves natural language so converting the listed triggers into natural language can result in more consistent generation.
Detailers are highly recommended for the face and eyes. The eyes especially since they require the prompt "scribble eyes" to generate properly.
Manual inpainting or image refinement through ChatGPT are recommended if you're looking to generate small details.
Probably won't be updated.
All example images were generated without a refiner step. You can probably make your generations look much better with improved workflows.
Triggers:
BowElepha, white skin, medium hair, black hair, black eyes, medium breasts,
lab coat, long black sweater, white pants, smooth white shoes,