Some issues in Release 2 are addressed~!

The training images are the same, but captions isn't. Also moved one image to a lower repeat folder.

Improvements (vs. Release 2):

Tail generation is more controllable with prompts

Slight improvements to outfit accuracy

Slightly more consistent background

Controllable Limitations:

She may have short hair. I don't know why this happens, but adding (long hair:1.25) mitigates this.

Stockings still may be plain black instead of blue + pink. Adding (mismatched stockings:1.15) to your positive prompt mitigates this.

Tails may appear uncontrollably. Adding (tail, mechanical tail:1.15) to negative prompt mitigates this.

Her weapon may appear uncontrollably. Adding (weapon, sword, holding, holding weapon, holding sword:1.15) to negative prompt mitigates this.

Uncontrollable Limitations:

From the viewer's perspective, the right side of her jacket may not be black, and instead gets colored white at times. This is gacha and not controllable.

Some limitations with H content due to training dataset all being SFW.

Training Details (w/ bmaltais/kohya_ss; see Training Data File for full LoRA training parameters extracted with kohya-ss/sd-webui-additional-networks)

Trained on: nai.ckpt (925997e9)

Training Batch Size: 2

LR: 0.0007, UNet LR: 0.0007, TE LR: 0.000175, Scheduler: Cosine w/ Restarts

Epochs: 3, Dim: 32, Alpha: 16, Resolution: 512x512, Clip Skip: 2

*EasyNegative in the example images is an embedding from here.