A lora to try and improve model's understanding of mud. No more brown rippling water or thin bodypaint, now it sticks and clumps. From personal testing it should be pretty stable and tolerate other lora's and checkpoints. It's all trained on real life images of mud so it pushes style towards realism, but can be tamed with style loras.
Description
Version 1.0, trained entirely on real life images of mud. I think future versions will include generated images that look more illustrated so it becomes more style agnostic.





