Please note that this Lora is outdated, my last LoRA "Fluffy dragons #2" is the refined versionFurry NSFW, trained on a mix of furry art. V1.1 is splashed with ai-generated pics and it is absolutely better: while heavy, it changes the style of the gen softening the female figure (even not-furry ones), hopefully preserving the pose, while resolving issues due to the few epochs of V1

First custom style

Reduced quantity of training files, reduced dimension of the LoRa, splashed with personal generations and augmented the epochs. Hopefully the results should be less rough.If you put the strenght at negative values, somehow you get more realistic gens. Under -1 I found hyper-detailed images at 2.5D. Do not ask me how.

First_custom_style.safetensors

Soft and defined

Washing in hot water

Mix of furry art by selected artists, with 10% of images generated by AI. It should be better than the first try, hopefully it should stabilize the generations without altering badly the perspective. Furry-oriented, but strangely not exlusive.

Custom style

Reduced quantity of training files, reduced dimension of the LoRa, splashed with personal generations and augmented the epochs. Hopefully the results should be less rough.

Files with SHA256: d2be9afe793a8954953e834a8c22c6f4c982823f6f9aa229f43f248570c63ab5