Update README.md

README.md

@@ -1,3 +1,40 @@
----
-license: cc-by-nc-sa-4.0
----
+---
+license: cc-by-nc-sa-4.0
+base_model: black-forest-labs/FLUX.1-Kontext-dev
+tags:
+- lora
+- flux
+- image-restoration
+- photographic-simulation
+- five5k
+- reverse-retouching
+---
+
+# Kontext-Five5k-Revert LoRA
+
+## Model Description / 模型描述
+**Kontext-Five5k-Revert** is a specialized LoRA trained on the **MIT-Adobe FiveK** dataset, designed for **inverse image processing**. While most AI tools aim for "perfect" aesthetics, this model reverses the professional retouching pipeline.
+
+**Kontext-Five5k-Revert** 是一个基于 **MIT-Adobe FiveK** 数据集训练的逆向图像处理 LoRA 模型。与大多数追求“完美画质”的模型不同，本模型专注于**逆向还原**。它能够将经过专业后期修饰、高动态范围的精修图像，还原为具有“原始感”的标准画质，并能模拟常见的摄影瑕疵。
+
+### Key Capabilities / 核心能力
+- **Quality Reversion (画质逆向):** Strips away high-end post-processing to restore a standard, unedited aesthetic. / 去除专业的后期处理痕迹，还原图像的标准化未编辑外观。
+- **Exposure Simulation (曝光模拟):** Accurately mimics digital artifacts associated with underexposure and overexposure. / 精确模拟欠曝（Underexposure）和过曝（Overexposure）带来的视觉特征。
+- **Hardware Simulation (设备模拟):** Replicates the visual profile, color curves, and limited dynamic range of mobile phone photography. / 复制手机摄影的色彩曲线、噪点特征及有限的动态范围。
+
+---
+
+## 🧩 Trigger Prompts / 触发指令
+
+| Task / 任务 | Prompt / 提示词 |
+| :--- | :--- |
+| **Simulate Underexposure / 模拟欠曝** | Downsample refined high-quality images to standard quality and simulate underexposure |
+| **Mobile Photography / 手机感** | Downsample refined high-quality images to standard quality and simulate mobile phone photography |
+| **Simulate Overexposure / 模拟过曝** | Downsample refined high-quality images to standard quality and simulate overexposure |
+
+---
+
+## ⚙️ Recommended Settings / 建议参数
+- **LoRA Weight:** 0.6 - 1.0
+- **Base Model:** FLUX.1-dev
+- **Sampling:** Match standard FLUX settings