Trained with 59 images, using just the standalone keyword is sometimes enough, but using more specific descriptions yields better results.
Use lower weights for custom outfits.
Default outfit:
senyaichiya, cape, white dress,
Night outfit:
senyaichiyanight,
Casual outfit:
senyaichiyacasual, sweater,Description
Initial release
FAQ
Comments (2)
I am trying to get what the point is of trigger word training/usage like this, like you dont have characteristic tags, and the generation is fully dependent on a few tags, so any consistency is not quite there, who is the target audience?
There are not a lot of characteristic tags, to not bloat the prompt box if someone just wants the character with their original appearance. You can add any characteristic prompt to it to influence it further, and to customize it fully. I don't use many checkpoints, so I don't know how well my LORAs perform on a dozen other ones, but as long as it's not VPRED, it should be fine.
Target audience: whoever that uses Illustrious, finds the models and ends up liking them. I'm not looking for a specific audience.
Details
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.