Upload README.md with huggingface_hub

README.md

@@ -24,7 +24,12 @@ tags:
 
 <table>
   <tr>
-    <td align="center"><img src="./assets/effects/output_004_bigger.webp" alt="LTX2.3-ICEdit-Insight preview" width="700"/></td>
+    <td align="center"><img src="./assets/effects/output_004.webp" alt="Video restoration preview" width="420"/></td>
+    <td align="center"><img src="./assets/effects/视频高清对比效果.webp" alt="Video HD enhancement preview" width="420"/></td>
+  </tr>
+  <tr>
+    <td align="center"><img src="./assets/effects/去水印对比效果.webp" alt="Watermark removal preview" width="420"/></td>
+    <td align="center"><img src="./assets/effects/去字幕对比效果.webp" alt="Subtitle removal preview" width="420"/></td>
   </tr>
 </table>
 
@@ -41,7 +46,7 @@ Unlike conventional frame-level enhancement pipelines, this model family operate
 
 Project links: [GitHub project](https://github.com/Valiant-Cat/LTX2-ICEdit-Insight) | [JoyFox on Hugging Face](https://huggingface.co/joyfox)
 
-## Model Files
+## 📦 Model Files
 
 | File | Purpose |
 | --- | --- |
@@ -51,7 +56,7 @@ Project links: [GitHub project](https://github.com/Valiant-Cat/LTX2-ICEdit-Insig
 | `ltx2.3-ic-watermark-remove-general.safetensors` | Watermark removal and occlusion-aware reconstruction |
 | `ltx2.3-ic-subtitles-remove-general.safetensors` | Subtitle removal and text overlay cleanup |
 
-## Showcase
+## 🎬 Showcase
 
 <table>
   <tr>
@@ -72,7 +77,7 @@ Project links: [GitHub project](https://github.com/Valiant-Cat/LTX2-ICEdit-Insig
   </tr>
 </table>
 
-## Script Usage
+## 🚀 Script Usage
 
 Run all scripts from the project root.
 
@@ -86,7 +91,7 @@ bash run_subtitle_rm.sh
 bash run_subtitle_rm.sh /path/to/input.mp4
 ```
 
-## Command Examples
+## 💻 Command Examples
 
 ### Video Restoration
 
@@ -156,7 +161,7 @@ python run_pipeline.py \
   --lora ./models/loras/ltx2.3-train/ltx2.3-ic-subtitles-remove-general.safetensors
 ```
 
-## Key Improvements
+## ✨ Key Improvements
 
 ### Task-Aware IC-Edit Framework
 
@@ -226,7 +231,7 @@ This is especially helpful for recovering:
 - typography-like fine structures
 - natural image clarity
 
-## Inference Notes
+## 🧠 Inference Notes
 
 - Single-stage inference is recommended for most editing tasks.
 - Two-stage refinement can improve visual polish but may weaken task-specific LoRA constraints.
@@ -237,7 +242,7 @@ This is especially helpful for recovering:
 - Frame count should follow the `8k + 1` rule.
 - Output height and width should be multiples of `32` in single-stage inference.
 
-## Training
+## 🏗️ Training
 
 This model family was trained and optimized by **JoyFox Lab** (**Chengdu Xuanhu Technology Co., Ltd.**).
 
@@ -251,13 +256,13 @@ The training pipeline includes:
 - high-frequency perceptual enhancement
 - instruction-guided video editing optimization
 
-## Contact
+## 📬 Contact
 
 For research collaboration, commercial licensing, or workflow integration, contact:
 
 - `z@vvicat.com`
 
-## License
+## 📜 License
 
 Licensed under **Apache 2.0**.