This is a character and style LoRA model for SD XL 1.0 for the characters and the style of the Powerpuff Girls, the original and reboot TV animation series.

To use as a character LoRA, apply it to the base model of SD XL 1.0 checkpoints as always, using some high strength value (e.g., 0.9-1.2) with the trigger word ppg and the name of the character you want to see (blossom, bubbles, buttercup or professor utonium.) A single-character use is just as other character LoRA.

This model, however, knows four characters, and it is possible to use more than one character in a single image, though the effect is not stable enough; you will need many iterations before getting a usable result.

To use as a style LoRA, apply it to the base model with some lower strength (0.6-0.8) with the trigger word ppg only (i.e., no Powerpuff Girls character names.)

Some random notes:

• This model tends to infer flat pink or otherwise unnecessarily pink-ish background.

• Inferred images may contain some unnecessary flowers or soap bubbles sometimes (for obvious reasons.) I have no good suggestion for the negative prompt words to suppress them. (For example, writing soap bubbles largely mutated Bubbles in my case.) Ideas?

• Some example images are produced by the AUTOMATIC1111 webui and some others are by ComfyUI. ComfyUI images are refined using the plain refiner model of SD XL 1.0 checkpoints. AUTOMATIC1111 images are not. Also, the seed numbers are incompatible between AUTOMATIC1111 and ComfyUI, as you may already know, and you can't reproduce AUTOMATIC1111 produced example by ComfyUI or vice versa.

• Although I have no will to restrict uses of this LoRA model in any way, each of the Powerpuff Girls characters appears an intellectual property of the appropriate right holder(s), and you are not allowed to infringe their rights.

• The training set of this model includes no images of Mojo Jojo. (I'm so sorry about that.)

• I consider Buttercup is the best among the three girls (and this is not about the LoRA quality.)

Have fun!