v2.0:
The trigger word is new "european badger" to expoit the previous knowledge of "badgers" by the checkpoints.
The LORA works now better with many checkpoints.
Still not good with FULLY_EXTREME_MODIFIED and RealVis.
v1.0:
SDXL knows badgers, but messes up the anatomy, especially the stripes on the face, hence this LORA.
The trigger word is "m3l3s". Different actions and views can be prompted by:
{front|three-quarter|side|back|overhead} view, {running|running airborne|standing|sitting|lying|erect}
Prompt "paws" if you want the legs to be completely visible.
"Wet fur" in the negative should make the fur consistently dry.
It's trained on the base model and I have tried it with these SDXL checkpoints:
Working: Base Model (best!), CyberRealistic, LEOSAM's HelloWorld, NightVision, Copax TimeLess
Struggeling: BriXL, FULLY_EXTREME_MODIFIED, Iris Lux, Juggernaut, STOIQO NewReality, Ratatoskr, RealVis, ZavyChroma, Colossus
This is my first LORA, please be patient, give me some feedback and expect improvements.