Type | |
Stats | 58 0 |
Reviews | (6) |
Published | Feb 7, 2025 |
Base Model | |
Training | Steps: 1,224 Epochs: 30 |
Usage Tips | Clip Skip: 2 Strength: 0.9 |
Hash | AutoV2 A1C4A6720C |
This lora was trained on 51 images of Asura's Wrath cinematics, specifically 3D in-game art cinematic art. No gohma or really robotic characters in the dataset, just humanoid looking characters. Overall, it's pretty decent. I'm not used to making 3D style art so this might not be the best results this lora is capable of. I think I tried out mostly Pony Diffusion V6, PonyFission, Hassaku, AutismMix, and then some other checkpoints. Kinda think I like Hassaku, but Pony Diffusion V6 might be more accurate? If you use a character lora you possibly will need to lower the character lora weight. In general, I'd suggest Clip Skip 2, Lora Weight 0.6-0.9 for this style lora, I kinda preferred 0.9 from my testing.
Trigger Word (Maybe Not Necessary): asurawrath
Ripped Muscles: manly, muscular, masculine
Possible Details To Add: Skin details, 3d, CG Graphics, 8K, very high definition
Extra: Upper Body
Negatives: letterboxed, text
Suggested Negatives: jewelry, twin braids