The LoRA model focuses on creating a anime illustration style of "sweet cute+cyber complex+highly saturated color contrast", which integrates "cute character form" and "high-density decorative elements". Based on the highly saturated candy color, the LoRA model matches complex and delicate decorative details (such as bows, badges, theme small items), and is suitable for festivals, theme clothing, cyber cute and other multiple scenes, so that the work has both the healing sense of cute and cyber style visual tension, and is a highly recognizable tool for anime creative illustration and role setting.
Description
The recommended weight for using this model alone is around 0.6-0.8, and when combined with other models, the weight may need to be lowered. Adjustments can be made according to the actual situation.
Sampler recommendation: Euler, with a recommended sampling step size of around 25-30.
High definition repair sampling recommendation: R-ESRGAN_4x+Anime6B, with a redraw sampling step of about 20 and a redraw amplitude of about 0.3.
Just keep CFG at 7.
The reference for positive prompt words is as follows: 1 girl, solo, skirt, shirt, pleated skirt, white shirt, full body, socks, pink hair, black footwear, bow, collared shirt, black skirt, red eyes, horns, shoes, red bow, , white socks, cardigan, dress shirt, loafers, looking at viewer, long sleeves, loose socks, pink cardigan, open mouth, ribbed legwear, sleeves past wrists, open cardigan, hair between eyes, puffy long sleeves, school uniform, sitting