V2使图片输出更稳定一些，不容易出现肢体问题。V2 improves image generation stability and reduces the likelihood of anatomical issues.MOSS · 绮真｜Refined Realism绮真是一款围绕写实质感与精致人像审美打造的 MOSS 融合模型。以 MOSS-Image Base 与 MOSS-Image SFT 各占 50% 的权重融合模型为基础，进一步融入两支分别侧重写实人体、肌肤质感与精致妆容、柔光人像的 LoRA，探索自然肌理与细腻美感之间的平衡。素材风格涵盖室内写真、近景人像、人体细节与日常随拍，也包含妆容鲜明、色调柔和、带有修饰感的人像摄影。整体创作方向关注肌肤纹理、人物姿态与光影氛围，同时融入精致的面部审美，为人物写真、氛围肖像和写实人体题材提供创作空间。“绮”取精致之美，“真”取写实之感。让细腻的美感，落在可感知的肌理与光影之中。推荐采样步数：8 步或 15 步，可根据实际出图效果选择。<hr />MOSS · Qizhen — Refined Realism is a MOSS-based merge exploring the balance between photographic texture and refined portrait aesthetics. Built on a 50/50 weight blend of MOSS-Image Base and MOSS-Image SFT, it further incorporates two LoRAs: one focused on realistic human form and skin texture, and the other on detailed makeup and softly lit portraits.The source imagery spans indoor photography, close-up portraits, human-form studies, and casual snapshots, complemented by polished portrait styling and gentle tones. Its creative direction centers on skin texture, pose, lighting, and atmosphere, with an emphasis on refined facial aesthetics.“Qizhen” brings together refined beauty and realism—a meeting of delicate aesthetics, tangible texture, and light.Recommended sampling steps: 8 or 15. Choose whichever works best for your images.

MOSS · 绮真｜Refined Realism

mossRefinedRealism_v1050steps_int8.safetensors

Files with SHA256: fa779819c53b884ebcb2bf3b0c9d052ae263965aa9142db24e1f067670d529cc