V2版本： 触发词：kinrolstyle, 这是由132张图通过1536尺寸素材训练出来的一个lora，我自认为效果还可以了。但是不认识国漫的角色得你搭配其他国漫角色来使用。并且他其实和V1的差别还是比较大的。 可以试着加上这一段反而效果会更好：High-end CG illustration with delicate thick painting, visible brushstroke texture, and vibrant Octane rendering. Extremely high definition. 另外最好搭配一个lora：anima-turbo-lora-vo.2.safetensors 强度开0.35~1都可以，越高衣服材质和饰品的金属质感越好。 V2版本第一张图的提示词如下： kinrolstyle, Film photography, cinematic lens language, muted pastel tones, dark and moody, low-light atmosphere, a distinctive personal perspective, a wonderfully blended artistic conception. A beauty, holding a crystal ball that radiates mysterious light, photorealistic, with a devastatingly beautiful face, fair and impeccably smooth skin, BJD doll-like style, adorned with opulent head ornaments. A cool, detached aura, a sense of Lolita fragile brokenness, translucency, a glass-like texture, a Lolita mature and elegant presence. Dreamy color palette, exquisitely intricate clothing and accessories. Mysterious, dreamcore, high-end. Premium, hazy smoky beauty, bright, high facial attractiveness, high contrast, soft focus blur, déjà vu, emotional atmosphere fully charged. Dark haze, luminous haze, high particle noise, film grain texture. Rich layering, freehand expressiveness, hazy aesthetics, the aesthetics of light, Lomo effect, surreal, high-end feel, a masterpiece.<pre><code></code></pre>

Files with SHA256: 548ba5be10c924f45523e0c006f2a3ce2095532cfd48baa67a1236407bf0cefe