English Version
This LoRA is a style-enhancement model. Its training data primarily references NAI5-style imagery, with emphasis on improving brushwork fineness, local texture, and detail layering while maintaining overall image coherence. It is trained on moss_image_sft_int8_convrot.safetensors. No trigger word is required by default, and it is not bound to a specific character or concept. Style strength is mainly controlled through LoRA weight and prompts.
Recommended weight: 0.6–1.0. Weights of 0.6–0.85 are more stable and suitable for general generation; 0.85–1.0 provides more noticeable detail enhancement, suitable for detail enhancement in illustrations, characters, and scenes. Best compatibility is with the same base model or closely derived models.
Version: v1.0. No trigger word is required by default.
Note: This is a style-enhancement LoRA, not a fixed character/concept model.
中文版
本LoRA为风格增强型模型,训练素材以NAI5风格图像为主要参考,侧重提升笔触细腻度、局部纹理与画面细节层次,同时尽量保持画面整体协调。模型基于 moss_image_sft_int8_convrot.safetensors 训练,默认无需触发词,不绑定固定角色或概念,风格强度主要通过LoRA权重与提示词调节。
推荐权重 0.6–1.0:0.6–0.85 更稳健,适合常规出图;0.85–1.0 细节增强更明显,适用于插画、角色与场景的细节增强;与同底模或同系派生模型兼容性最佳。
版本:v1.0。默认无需触发词。
说明:本模型为风格增强LoRA,非固定角色/概念模型。

