My first shared character LoRA.
Anima
0.8-1.0
Trained with arts and in-game screenshots
🐀
SD1.5
trained with 25 images, AnyLora Model, and SD VAE. on Kohya LoRA Dreambooth in Colab
Plays around with weight for better results. (0.6-0.85)
Put mustache/beard in negative prompt to generate him without facial hair.
This character is trained with in-game screenshots, so most results look 3D.
Try different LoRA anime styles to bring the 2D style.
🐀
Description
🐀
FAQ
Comments (3)
I wonder if this would work well with my RDR Art LoRA: https://civitai.com/models/1397355/promo-art-style-read-dead-redemption
I had planned to make illustrious version of him. but have to gather new dataset since the old one is 512*512.
@A_alius let me know if you do and if you do it for other RDR characters! Also, you can make outputs more 2D focused by mixing in art of the character as opposed to screenshots from the game into the traning data