.jpeg){kind=icon}
So, after about a year or two i'm finally ready I think to start sharing some of my new Illustrious trained models. I've previously done most of my work before on SD 1.5 and haven't really touched SDXL at all because at the time it wasn't great and the hardware requirements seemed out of reach for my system.
This has been done with One Trainer for simplicity. Dataset was 44 images with no NSFW images included. Done for 100 epochs with no repeats, the 80th epoch was chosen for style and flexibility. DORA was included but probably not necessary? It's possible this was slightly overtrained, due to this being my first trained models in a long time.
Style seems overly consistant with the animation and character is generally good.
Prompt tag is "louiseanim"
Use "anime_coloring" or "sketch" to invoke the style although generally it should be present most of the time anyway
"brown_hair, brown_eyes, 1girl, green_dress, choker" tags should generally invoke the main character Louise?
Tested on WAI-NSFW-Ilustrious v14.
Hope this works for everyone else.