LoCon trained on various hi-res works of Aroma Sensei.
Trained on a total of 123 pics + cropped closeups of the face style. NAI is the base model as its the most versatile base training.
Pics were generated using Holokuki
And all pics were generated using only txt2img + hires fix.
Trigger word is: aromasensei (no need for trigger word)
Works well at 0.65 weight.
Description
After many attempts, V7 is a improvement of V3, with a more defined style, better faces, better hands, less messy backgrounds, and more hair variety.
FAQ
Comments (3)
Utterly astounding results. Is there any chance you'd be willing to share your training process, or take an artist request? A TheMaestroNoob or DandonFuga LoCon by your hands would be god tier.
I think the reason why it's good is that Aroma's art suits well for learning. Her anatomy and composition is good, her shading is almost 3D render-like, even her semiface is a blessing.
I could give those a shot, my settings aren't anything special though