Just updated to V2. Added shiny version, more poses, and drawn images to the training.
To use shiny use "shiny umbreon". To use regular just use "umbreon".
Lora I trained using AnythingV5Ink_v5PrtRE. Used some blender animations and images to get better poses. To not get 3D-looking generated images that resemble a 3d blender model put "3d (artwork)" in negatives. You should now get better poses compared to V1.
If you can recommend a better model to train with I would love to hear it. I wanted to use animefull/NAI but I don't want to torrent it. AnythingV5Ink_v5PrtRE should be good enough from the look of it. I would love to see if you could generate any good umbreon art with this Lora model. I got good images with anythingV5.