KREA 2 VERSION UPDATED WITH CORRECT FILE PLEASE REDOWNLOAD
Suggested strength: 1.0
Suggested tags: wearing <color> and <color> embroidered floral lingerie, bra, panties, garter belt, lace thighhighs
Krea 2 is very impressive! Listens very well, you can use more than two colors as that is present also in the dataset. It can do lingerie or other clothing, even random objects! Captioned using natural language.
Illustrious 1.0 Release:
Trained on Illustrious 1.0!
God I love Illustrious still. Same dataset as Anima.
Anima Release:
Suggested strength: 1.0
Suggested tags: embroidered floral lingerie, bra, panties, garter belt, lace thighhighs
First Anima LoRa! Trained on Anima-base 1.0 with 25 images. Pruned the tags well but forgot to add color to the first token so all of them were tagged "embroidered floral lingerie" without color identification, seems to work well enough. May train an IL one as well but I spent too much time experimenting with Anima on these examples. Prompting for color works, also tends to match the character from what I've gathered which is what I was hoping.
Description
Suggested strength: 1.0
Suggested tags: embroidered floral lingerie, bra, panties, garter belt, lace thighhighs
First Anima LoRa! Trained on Anima-base 1.0 with 25 images. Pruned the tags well but forgot to add color to the first token so all of them were tagged "embroidered floral lingerie" without color identification, seems to work well enough. May train an IL one as well but I spent too much time experimenting with Anima on these examples. Prompting for color works, also tends to match the character from what I've gathered which is what I was hoping.
FAQ
Comments (2)
Amazing work on this and all of your other Loras! I love this bra set and other Wacoal Salute ones! Any plans to have this on Illustrious or other models? I prefer to do photo realistic and have been using your other bra Loras almost exclusively. Please keep up the great work! 🎉
Thanks for the kind words and glad to hear they're getting good use. I can definitely do an Illustrious one. It costs a bit more time and resources to train on models like Flux or Zimage but I'll think about it