The spaal/tihospaal style brought to you for Pony Diffusion 6 !

Use the trigger "by spaal".

Trained in 240 minutes using 202 high-quality images from e621 and rule34.

Version 2 is finally here !

It was high time this LoRA got a revamp, and a much deserved one. This Version improves every aspect of generations, making this LoRA MUCH MORE FAITHFUL to the Spaal style.

• Dataset freshened with the latest Spaal artworks.

• Total change of the training method, ditching the Lycoris method for a classic LoRA with improved scheduler settings and way more training steps.

• Fix of an issue in the tagging that would sometimes ignore the "solo" tag and produce multiple characters unprompted.

• This version supports all weightings, all the way to 1.0 and even above! Good results have been observed in testing up to 1.25.

• Just better everything overall!

NB : This isn't published on the same page as the V1 because the V1 is a Lycoris while the V2 is a LoRA. You can find the V1 by clicking here.