13 images. Trained on Hassaku. This series doesn't really have any LoRa, so I'll provide. Works well, you might have to use prompts to generate her in "full body" view, I didn't have to use it, but most of the images I put in were of her upper body screenshots, so keep it in mind.
Works best with ThorneMix. The other models had good results, ThorneMix was cut above the rest in my opinion. All of the images I post are from that model, I didn't even realize it until I gathered my material for the release.
Works welk with MeinaPastel v6, MeinaPastel v4, RuviStyle_v2.1 and Mistoon_Anime v2.0 (kl-f8-anime2 VAE).
If you find other models that produce adequate results, then be sure to share them.
The basic settings I generate the LoRa images with.
I use Euler A
23 sampling steps ( I use this little because of testing, feel free to use more)
7 CFG
0.85 LoRa strength in the PixAI sidebar.
Thumbnail Model: ThorneMix
