Superheroine Pack

This model is a first for me. I wanted to try training multiple characters in one model with Wan 2.2 to see if I could get it to work and this is the result. This High and Low noise combo includes Powergirl, Supergirl, and Wonder Woman, all in one two-part LoRA for Wan 2.2. Unlike the other Wan LoRAs I've released recently, you can't really get away with using just the high or low noise model. You really need both to get the characters to look right. See below for important tips on prompts and strength.

Important: This model is 100% fictitious and is not based on images of any real person. She originated with images generated from still image checkpoints and has evolved from there using only AI rendering. No images (neither actual nor AI-generated) of any real person were ever involved in the making of this model.

One thing I was hoping for with this model that did NOT work was prompting for more than one of these characters together in the same scene. Nope. Wan just blended the two women and their costumes together. Maybe someday we'll have reliable regional prompting for Wan 2.2 and then I can try again.

This model is created using only images, in fact the same images that were used for the older Wan 2.1 models (Powergirl, Supergirl, and Wonder Woman). The model strength needs to be pretty high to capture the faces, the bodies, AND the costumes all at once, but without messing with Wan's native motion. So it was a compromise. For whatever random reason, each of the three characters seemed to train slightly differently in spite of having nearly the same number of training images between them. So the usage tips here can help if you want to make a good scene.

Usage Tips:

Easy Mode: Just set the model to 1.00 in both high and low and prompt for the character you want and it'll generally work most of the time.

Basic Prompts:

• powergrl. a woman wearing a powergrl costume is ...

• supergrl. a woman wearing a supergrl costume is ...

• wonderwmn. a woman wearing a wonderwmn costume is ...

If the model doesn't make your character look correct or seems to be blending the characters a bit, you can reinforce the look with some additional descriptions. The most likely thing to go wrong is the hair, so it's easy to say:

• powergrl. a woman with short blonde hair wearing a powergrl costume is ...

• supergrl. a woman with long wavy golden blonde hair wearing a supergrl costume is ...

• wonderwmn. a woman with long wavy black hair wearing a wonderwmn costume is ...

If you know in advance which character you plan to include in your scene, then you can adjust the model strength slightly for each heroine.

• Powergirl: High 1.00 / Low 1.00

• Supergirl: High 0.75 / Low 1.00

• Wonder Woman: High 0.75 / Low 0.75

Those numbers are not set in stone, they just seemed to work well for me in my tests. Feel free to play with settings and remember the best values may be different if you blend in additional LoRAs with this character LoRA.

If the costume is off, you can reinforce the costume descriptions to help Wan keep things straight. This isn't needed as often, but it can come in handy if you really like the scene and Wan decided to give Wonder Woman a skirt or to give Supergirl a chest cutout in her bodysuit.

• Powergirl: "Her costume includes a long-sleeved white bodysuit with a chest cutout. She also wears a red cape, long blue gloves, and tall blue boots. She has a red belt with a gold buckle around her waist. Her cape is attached at her shoulders with gold clasps and a gold chain."

• Supergirl: "Her costume includes a blue long-sleeved bodysuit, a red cape, a short red skirt, and tall red boots. She has a Supergirl logo on her chest."

• Wonder Woman: "Her costume includes a red strapless bodice with a sweetheart neckline and gold trim. her blue bottoms are decorated with white stars. she has silver cuffs on her wrists and she wears tall red boots."

Beyond that, you can provide additional prompting to make them more muscular or slender or busty or whatever and Wan will do its best to help.

I hope you enjoy using this model and make some cool videos with it. Please be sure to tag this model when you post so I can see what the ladies are up to these days.