Recommended settings: heunpp2 + linear_quadrtic, 12 steps, 1 cfg

推荐设置：heunpp2 + linear_quadrtic，12 步，1 cfg

Built on Moss Image SFT base, this model merges 6 LoRA/LoKR sets. It elevates aesthetic performance, fixes human anatomy issues, and balances generation speed and output quality. It features enhanced photorealism, cinematic framing and atmosphere. Benefiting from Moss’s strong CLIP capability, it delivers stunning results with original IL & XL prompts. Regional prompting is supported, and it has excellent text rendering ability.

本模型基于 Moss Image SFT 基座打造，融合合并 6 组 LoRA/LoKR。优化画面美学表现，修正人体肢体结构，兼顾生成速度与出图质量；着重强化写实质感、镜头语言与氛围感。依托 Moss 优秀的 CLIP 能力，使用 IL、XL 体系的原有提示词会收获惊艳效果。支持分区提示词，文本渲染能力同样出色。

Have fun creating! 祝大家玩得愉快！