这个LORA用的是majicFlus 麦橘超然 作为底模，生成的明星大幂幂拥有超高度的还原

喜欢我的LORA，请关注我，如果你想要定制你的LORA，可以私信给我，或者加QQ：317705699.备注定制LORA

麦橘超然 MajicFlus 是一款基于 flux.dev 微调融合的模型，专注于高质量人像生成，尤其擅长表现 亚洲女性 的细腻与美感。模型以 唯美、写实、易用 为核心特色，能够通过简单的提示词生成优质效果，同时对复杂提示词也有出色的响应能力。

模型特点

• 卓越的人像生成能力： 优化了在不同光影条件下的表现，确保人像在各种构图中的 面部细节 和 肢体完整性。

• 广泛的适用性： 除了人像生成外，模型在生成 非人生物 和 场景 时也有显著改进，适应更多创作需求。

• 简单易用： 用户无需复杂的提示词即可生成高质量作品，同时支持更长提示词的精细控制。

参数推荐 Parameter Recommendation 参数推荐

• 步骤: 20~30

• Distilled CFG Scale: 3.5

• CFG : 1

• 低精度扩散：float8-e4m3fn

• 采样： Euler + 简单/β（通用），DPM2M + SGM 均匀（皮肤纹理），DEIS + DDIM 均匀（休闲写实风格）

• Vae: flux vae

• Clip: clip_l.safetensors 和 t5xxl_fp8_e4m3fn.safetensors

This LORA uses MajicFlus (麦橘超然) as the base model, achieving ultra-high fidelity in generating images of celebrity Dami Mimi.

If you like my LORA, please follow me. For custom LORA commissions, feel free to DM me or add QQ: 317705699 (Note "Custom LORA" when adding).

MajicFlus (麦橘超然)

A model finetuned and fused based on flux.dev, specializing in high-quality portrait generation with exceptional performance in capturing the delicacy and beauty of Asian females. Its core features include:

• Aesthetic & Realistic Outputs: Produces professional-grade results with simple prompts while maintaining responsiveness to complex instructions.

• User-Friendly: Designed for effortless creation without requiring intricate prompt engineering.

Model Features

1️⃣ Superior Portrait Generation

• Optimized for diverse lighting conditions and compositions.

• Preserves facial details and ensures limb integrity.

2️⃣ Broad Adaptability

• Enhanced performance in generating non-human creatures and scenes, expanding creative possibilities.

3️⃣ Simplified Workflow

• Achieves high quality with minimal prompts.

• Supports refined control through extended prompts.

Recommended Parameters

• Steps: 20~30

• Distilled CFG Scale: 3.5

• CFG: 1

• Low-precision diffusion: float8-e4m3fn

• Samplers:

  • General: Euler + Simple/β

  • Skin texture: DPM2M + SGM Uniform

  • Casual realism: DEIS + DDIM Uniform

• VAE: flux_vae

• CLIP: clip_l.safetensors & t5xxl_fp8_e4m3fn.safetensors

The translation retains technical terminology while ensuring readability for international audiences. Let me know if you need adjustments to specific terms!