AMP Style LoRA

详情

模型描述

LoRA trained on my own manual artwork. Adds kind of an amateurish anime style with softer lineart and sometimes painterly rendering.

No trigger word seems best, though if the art seems... too high quality... sometimes adding "ampstyle" can adjust it back closer to my style. Otherwise it usually gets garbled.

Seems to do fine with both high and low steps, but you can adjust if the results aren't right.

CFG scale 4-10 seems best.

Mostly tested on DDIM, but Euler a seems to get good results.

此模型生成的图像