A three-in-one package of superhero women in one LoRA for Krea 2. The package includes Supergirl, Powergirl, and Wonder Woman, using the same source data that fed my similar package for Wan.Important:<span style="background-color:rgb(26, 27, 30);color:rgb(141, 141, 143);font-family:-apple-system, BlinkMacSystemFont, &quot;Segoe UI&quot;, Roboto, Helvetica, Arial, sans-serif, &quot;Apple Color Emoji&quot;, &quot;Segoe UI Emoji&quot;;font-size:16px"> As always with my models, this LoRA does not represent any real person in any way. If this model produces a video with any resemblance at all to a real person, that is purely an accident of AI random generation.Each character has her own trigger word. The model does occasionally mix up the wardrobes a little bit, so it can help if you reinforce the specific character/costume you want with some extra description. In my samples, I used phrases like these in my prompts:<ul><li>a supergrl woman wearing a supergrl costume is pictured. her costume includes a blue bodysuit, a long red cape, a short red skirt, and tall red boots. there is a superman logo on her chest and she has long silky blonde hair.</li><li>a powergrl woman wearing a powergrl costume is pictured. her costume includes a white bodysuit with a chest cutout, a long red cape with a gold chain clasp, long blue gloves, and tall blue boots with a red belt around her waist. she has short blonde hair that falls partly in front of her face.</li><li>a wonderwmn woman wearing a wonderwmn costume is pictured. her costume includes a red metallic bustier bodice with gold accents over the chest and blue bottoms cut high over her hips. she wears tall red boots with gold trim. she has long wavy black hair and bright blue eyes.</li></ul>A strength of 1.0 should work well for all three models. You can even prompt them to be in scenes together since Krea's prompt understanding is really good. There are examples of that with prompts included in my sample set.I like the results I get with the LCM / beta sampler/scheduler and also res_multistep / sgm_uniform which gives you a slightly less perfect, grittier, more human result in my opinion.Enjoy!

Files with SHA256: 6e7d2cf9a5cb1341662faf7582943e28eed921f5765901b4fb05faffc8bbbd03