This LORA was trained with the capacity of creating various types of armor suits and mechas. It is heavily inspired by nijijourney, but can also do realistic images.
Recommended LORA weight: 0.8-0.9.
No prompt activator needed, but it was designed to react to diverse trigger words.
Suit/robot type: android, cybersuit, exosuit, hardsuit, mechasuit, mecha
Armor spec: shoulder armor, upper armor, full armor
Accessory: helmet, headgear, headset, cape, cloak, visor, mechanical jaw
Aditional triggers: dark skin, bulky, mechanical arm, mechanical arms, mechanical parts
Important note: this lora was designed to create mechas without wings or any other kind of back volume - it will only happen if the checkpoint used overrides it.
Please, feel free to ask for help to use it or to adress any problems.
Description
Trained on 150 images from nijijourney. Most of them were edited (from unwanted artifacts removal to proper cropping).
Dataset resolution: 512x768.
FAQ
Comments (5)
Incredible job, absolutely love it!
Amazing work! Anyone trying to make power armor or any sort of military sci fi needs this.
Is there a good way to get consistent armour between generations?
A word of warning: it's really difficult to get poses done with this LoRA, most generations will have arms at sides and without expressions (even with ControlNet is hard).
Any reason why lower settings (below .85) force a male body and higher settings force a female body?