Fujibayashi Sheena - [Tales of Symphonia]

Iterative LoRA Training Model - My first attempt at creating a LoRA

Suggested tags for consistent results:

fujibayashi sheena, black hair, brown eyes, messy hair, pink bow, pink sash, black pants, purple boots

If you include Pony Tail, the generation will have a tail. I messed up with the training data--If you want to include it, just use Ponytail.
I found that the LoRA works fine if you omit it entirely, though.

Recommended settings:

• DPM++ 2M or Euler a

• Karras

• Sampling Steps: 25-35

• CFG Scale: 4-6

• Model Weight: 0.75-1 (I've had good success with weight 1)

As a personal preference, I generally use ADetailer for the faces of any generation, although this model is fairly consistent with it's accuracy.

I have had success with multiple Checkpoints, but any checkpoint that has a large amount of anthro images included in the training data seems to affect the results for this LoRA.

I would love to see what you create with this LoRA.

I'll be updating this LoRA as I better learn the process of creating them.