Thorra Anime Lora Lite (flux edition)
as seen in SDXL: https://civitai.com/models/460522/thorra-sdxl-public
I decided to split the dataset into 3 separate Lora's.
Main = 90's Anime Thorra Character Lora
(Character Lora)
Scene = 90's Anime Thorra
(Megacity, Fortress Corridor, Airlock)Actions = 90's Anime "Thorra" Action Compositions
(Rushing, Shooting)
Note: You can run the Main Lora alone, however it is designed to be reinforced by Scene and/or Actions
example, using Thorra Scenes only:
versus, loading the same prompt & seed with Thorra Main + Thorra Scenes at 0.6
Thorra Lite V2 is trained in rankdim 4-16
captions are in the training data for each release