If you want a more traditional lora experience i recommend the 576 version of the lora.This is a model trained on bonk's style and most of bonk's esix portfolio of imagery. This model is intended to be used with fluffyrock derivative models such as Omega mix, which can be found in my user profile. This lora works best when combined with other artists to give visual structure to the image, such is the weakness/strength of style selective models such as fluffyrock.Note that i have added the term "eyeliner" to associate the heavy colored eyelashes of bonk's style. you may need to use "eyeliner eyelashes" inorder to get them to really popI might do a training on nai at some point.Remember that my representational images are non exhaustive and that creativity is the onus of the user of the lora.Enjoy.

Bonk

This is a model trained at 768x788 bucketing at batch 3 for around 4k steps

bonkv1.3 768 .safetensors

v1.0 fluffyrock 768

v1.0 fluffyrock 576

Bonk [style and characters]

slightly different training than 576 so youll get slightly different results3 batch at approx 4000 steps

Files with SHA256: 837602ec20bc66e6b77b91e0b95351fc6472ee84dc9316a8e2267180e3e20abf