Trained from mostly anime screencaps. If the generated image appears stiff, consider adjusting the weight downwards and utilizing img2img or ADetailer for enhanced results. Recommended weight range: 0.7 to 0.9 (depends on checkpoint you use).
Update:
v2.0: Same dataset but trained with improved quality and flexibility, recommended weight range: 0.8 to 1.
Base:
sasha_necron, long hair, blonde hair, blue eyes, twintails, hair between eyes, bangs, hair ribbonOutfit:
black necktie, fingerless gloves, long sleeves, black jacket, red stripes, white shirt, garter straps, black dress, high heels, thigh boots