This model excels in producing illustrations with strong social attributes, making it highly suitable for creating stickers, cards, and similar needs;
Currently, the model is at its 1.0 phase. It requires specific formats for keyword input. I suggest starting with simple keywords and gradually adding more complex ones to incrementally enrich the illustration's details;
· Recommended sizes: 768x768
· Sampling steps: 30-35
· Sampler: DPM++ 2M SDE Karras
· CFG suggestion: 5-7
· Plugin recommendation: Enable the ADetailer plugin to enhance the precision of faces and hands;
Start with simple formats and progressively move to more complex structures.
Subject: 1 girl, 1 boy
Styles: Hairstyle, hair color, eyes, lips, etc.
Clothing: Style, upper body, lower body, material, etc.
Props: Flowers, weapons, bubble tea cup, witch hat, etc.
Background: Solid color or scene (window, table, etc.)
Other: Special decorations (sparkles, face stickers, moon, etc.)
The model has a good understanding of colors, and there are several key terms that are recommended for users to try first:
(gradient hair, multicolored hair)
(letterman jacket, jacket)
(window, desk, bubble tea)
For other keywords, refer to the sample images for combinations and applications.
——
Tips: You can copy the prompt to produce a similar image. For specific applications, you can refer to the cover.
这是一款能产出强社交属性插画的模型,对于做一些贴纸及卡片等需求,有较好的应用性;
模型处1.0阶段,对关键词的书写格式有一定要求,建议从简到复杂不断添加,以逐步提升画面的丰富程度;
——
·尺寸建议:768*768,768*1024
·采样步数:30-35左右即可
·采样器:DPM++ 2M SDE Karras
·CFG建议:5-7
·插件建议:建议开启ADetailer插件,提升脸+手的精细度;
——
格式建议,建议还是从简单到复杂
主体:1girl、1boy
造型:发型,发色,眼睛,嘴唇,等;
服饰:款式,上身,下身,材质,等;
道具:花,武器,奶茶杯,巫女帽,等
背景:纯色 or 场景 (窗户,桌子等;)
其他:特殊装饰(闪亮,脸贴,月亮等;)
模型对于颜色的理解不错,有几个特点词汇建议大家优先尝试:
(gradient hair,multicolored hair)
(letterman jacket,jacket)
(window,desk,bubble tea)
其他关键词参考示例图片,组合应用
——
Tips:您可复制prompt出近似的图片,具体的应用你可以参考封面