Artist Style Lora for Kumiko Aleron. Trained on Pony.
Activation tag: kumiko_aleron_style
Additional character tag: Syndra (from League of Legends; due to her being in a lot of good training images)
Best weight seems to be 0.6-0.8. 0.7 works best for me personally.
Trained on a lot of nsfw images due to the artists preferences, but generates sfw without issue.
V4 is out! V4 removes patreon watermarks entirely and improves dataset variety and tagging.
Description
No more patreon logos! Removed all occurances
Improved dataset / Removed problematic images
Fixed some tagging issues
FAQ
Comments (6)
how did you generate that detailed eyes,it's always blurry in my character pupils when use this lora
You should use the Adetailer extension. I use that all the time with every model.
It basically detects faces and automatically enhances them with inpainting after generation. Can be extended for other detection models like hands, feet etc.
It also lets you change details afterwards, like changing eye color or adding facial details not specified in main prompt.
I use it for any generation where im not experimenting with basic image composition
@Azfr thanks for that,I will try.You made a wonderful lora ^_^
Thanks for the kind words :)
I assume the eyes being blurry could stem from either conflicting weights when using a child-model of pony or more likely the fact i couldnt find any close up shots for the dataset.
Can you tell me which base model you used, the strength of the lora and wether you prompted for a close-up or full body shot when generating? Would help fixing it up if i ever get to release a V5 in the future @Opasio
@Azfr I use a self-merge base model and it's merged by too many models that i cant even rememer the number,I tought it may be the reason caused the blurry.
adetailer works poorly on my model,now i use photoshop to fix eyes XD
Can you retrain for Illustrious?
Details
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.