If you enjoy my AI creations and wanna show some support, you can send some headpats my way through the link provided: https://buymeacoffee.com/fenrir_lupus
Video Game: Elsword, Developer: Kog Studios aka Kog Games,
This is a Lora Based on a Videogame Character Named Luciela R Sourcream from Elsword,
She is an Adorable Feisty 100+ Year Old Demon Lord from the Underworld! She saved a Man called Ciel by making a Contract with Him on his Deathbed from Losing a Fight with a Demon and thus Their Relationship was Created.
Checkpoint: Illustrious XL
Lora Strength: Any Strength Works So Far
Config: Usually Between 2.5 - 7.5
Steps: Usually 25 - 50
Resolution: Any Should Work Fine So Far
Issues:
All Forms Generate well Except for Nihilida, Wander Marchen, Dawn Sovereign Due to Limited Dataset of that Specific Version or Form
Tags
Prefix is Luciela and Suffix is the version or form wanted such as: luciela(standard) or luciela standard
Also The Dataset Was Trained with Broad Tagging to Try and Get the Forms to Work So Adding Specific Details to use those weights from that Tag can Help an Example of this is Blue Horns, Demon Horns, OR Blue Tail, Demon Tail,
I hope you enjoy this LoRA as much as I do, I’d love to see what you create with it! Be sure to post below what you think and made. ^~^
Description
Base Model: Illustrious XL 1.0
Version: 1.0
Dropout Training: Not Implemented
FAQ
Comments (2)
something is really wrong with ur lora
certain negative tags seem to break it completely above 0.5
it is also incompatible with artstyle loras
Thank You for Telling me this ^~^
IDK why the other loras don't work with it as only thing i can say is for a Lora to be compatible across the entire database, each lora must use the base model for it to be most compatible such as My lora's are trained on Illustrious XL 1.0 and 2.0 which is specified or i try to specify it...
What Lora's Specifically are You Using with it? and are their Base Models Using IllustriousXL 1.0 or 2.0? as From What I see on CivitAI is a LOT of Models that Exist But Break Due to Everyone Training on their Own Checkpoints Instead of the Base Models such as example Users Train on Illustrious XL BUT it could be Mistoon, or Furry Nova Which are NOT Base Models as they are Their Own Versions of the Base Model which Break Often...
This Lora was Trained on Aspects Of Her as this was My 10+ Version of Her which I only Tagged the Aspects that Exist and Not Everything Else Because it's Difficult with the Drastic Versions BUT Using the Base Models OR a single lora or 2 are fine but more than that it breaks so other than the loras you use how many do you use?
Again Thank You For the Feedback!!! It's Been a While Since I trained Anything BUT I am Learning Best I Can! ^~^