Pony v2Well... here it is. Don't know what happened during the training process. It turned out good in terms of actions, poses, etc, but somehow even worse with styles than a v1. Some style Loras work, but you have to crank them up to the maximum for that, as a result... it tends to overcook the image and certain details become smudged. Guess, it calles for an eventual v3.Pony v1My first Lora that turned out good enough to publish. I honestly have no idea how well it works and how flexible it is, so please provide feedback so that I can do better next time!

[Underworld] William Corvinus, the first werewolf (Pony)

Reworked the dataset. Not overtrained, more stable, more flexible in terms of actions and poses, but, sadly, less flexible in terms of style.

WillCorv_v2.safetensors

[Underworld] William Corvinus, the first werewolf

Pony v2

William_Corvinus

willcorv_v2, 1boy, werewolf, no pupils, shaggy white long fur, sharp teeth, white fur, white eyes

Files with SHA256: 8109af8ddb218a59ebcf1f7323c76854e9277000859cc18eaeed5067b6f82a53