LoRA权重推荐:0.9这是一高饱和风格的风格迁移模型 ,使用后会对底模风格整体覆盖。 可以通过增加权重来加强风格覆盖强度 ,但在加权的同时会影响稳定性。)

提示词推荐样式:触发词+主体描述+环境描述+其他补充描述。(本模型保持了墨幽风格模型一贯优秀的风格覆盖能力,触发词放在提示词最前,然后输入口语化的主体描述以及背景描述即可。)

例如: High saturation illustration,A vibrant, anime-style digital illustration features a young, fair-skinned girl with curly, neon green hair and blue eyes. She's rocking a yellow shirt with a blue tie and black vest, and she's got a pair of goggles perched on her head. She's sitting in a futuristic control room with a bunch of colorful, high-tech panels and screens around her. The room's got a playful, retro-futuristic vibe, with bright, neon colors and intricate patterns. The girl's leaning back in her seat, looking relaxed with one hand on the steering wheel, and the other hand on her chin. The background is filled with more high-tech gadgets and a bright, sunny sky with fluffy clouds, creating a fun and lively atmosphere. The composition uses a mix of leading lines and the rule of thirds, drawing attention to the girl's face and her surroundings. The image is safe for work (sfw) and has a cheerful, playful mood.(更多提示词示例请点开例图参考例图信息)

若撰写详细的自然语言提示词让你头疼,可以尝试使用AI文字工具进行格式参照后,根据需求进行改写。

注意事项:

若遇到手部、肢体问题,还请抽卡食用~

也可配合canny、depth做风格转绘等。

可配合fill做局部置换。

使用建议:

触发词: High saturation illustration

webui-forge中:

基础设置:

采样方法 (Sampler):euler

调度类型 (Schedule type):simple

迭代步数 (Steps):30

Hires Distilled CFG Scale:3.5

Hires CFG Scale:1

建议分辨率:768×1024、768×1152、1024×1152、1024×1280、1024×1536等

高分辨率修复 (Hires. fix)设置(如果性能够推荐开):

放大算法:4x-UltraSharp(也可以试试别的)

高分迭代步数:20

重绘幅度:0.3

放大倍数:2


comfyui中:(k采样器的参数)

采样:euler

调度:simple

步数:30

CFG:1

降噪:1

如果提示词使用joy caption two反推:

提示词类型建议:Descriptive (Informal)

提示词长度建议:very long


要是觉得不错还请多多返图支持,谢谢~


鹅群:

墨幽炼丹阁-太清阁:858495398

墨幽炼丹阁-上清阁:764481772

也请多多关注我的小红书和抖音账号:yy469370733