Update README.md

README.md

@@ -11,3 +11,117 @@ base_model:
   - Wan-AI/Wan2.1-I2V-14B-720P
 library_name: diffusers
 ---
+# Wan2.1 Distilled Models
+
+This is a collection of distilled and accelerated versions of Wan2.1 video generation models, offering multiple precision and format options. All models are optimized for **4-step inference**, dramatically improving generation speed while maintaining high-quality outputs.
+
+## 📦 Model Overview
+
+This repository provides multiple distilled versions of Wan2.1 models, covering different tasks, resolutions, and precisions:
+
+### Model Types
+
+- **Image-to-Video (I2V)**: 480P / 720P resolutions
+- **Text-to-Video (T2V)**: 14B parameter version
+
+### Precision Variants
+
+Each model is available in the following precision options:
+
+| Precision | Suffix Identifier | Size | Framework | Description |
+|-----------|-------------------|------|-----------|-------------|
+| **BF16** | `lightx2v_4step` | ~28-32 GB | LightX2V | Original precision, highest quality |
+| **FP8** | `scaled_fp8_e4m3_lightx2v_4step` | ~15-17 GB | LightX2V | FP8 quantization, half size |
+| **INT8** | `int8_lightx2v_4step` | ~15-17 GB | LightX2V | INT8 quantization, half size |
+| **FP8 ComfyUI** | `scaled_fp8_e4m3_lightx2v_4step_comfyui` | ~15-17 GB | ComfyUI | ComfyUI compatible format |
+
+### Naming Convention Examples
+
+```
+wan2.1_{task}_{resolution}_{precision}.safetensors
+
+Examples:
+- wan2.1_i2v_720p_lightx2v_4step.safetensors              # 720P I2V original precision
+- wan2.1_i2v_720p_scaled_fp8_e4m3_lightx2v_4step.safetensors  # 720P I2V FP8 quantization
+- wan2.1_i2v_480p_int8_lightx2v_4step.safetensors         # 480P I2V INT8 quantization
+- wan2.1_t2v_14b_scaled_fp8_e4m3_lightx2v_4step_comfyui.safetensors  # T2V ComfyUI scale_fp8 format
+```
+
+> 💡 **Tip**: Browse [Files](https://huggingface.co/lightx2v/Wan2.1-Distill-Models/tree/main) to see all available models
+
+## 🚀 Usage
+
+**LightX2V is a high-performance inference framework optimized for these models, approximately 2x faster than ComfyUI with better quantization accuracy. Highly recommended!**
+
+#### Quick Start
+
+1. Download model (720P I2V FP8 example)
+```bash
+huggingface-cli download lightx2v/Wan2.1-Distill-Models \
+    --local-dir ./models/wan2.1_i2v_720p \
+    --include "wan2.1_i2v_720p_scaled_fp8_e4m3_lightx2v_4step.safetensors"
+```
+
+2. Clone LightX2V repository
+
+```bash
+git clone https://github.com/ModelTC/LightX2V.git
+cd LightX2V
+```
+
+3. Install dependencies
+
+```bash
+pip install -r requirements.txt
+```
+Or refer to [Quick Start Documentation](https://lightx2v.readthedocs.io/en/latest/getting_started/quickstart.html) to use docker
+
+4. Select and modify configuration file
+
+Choose the appropriate configuration based on your GPU memory:
+
+**For 80GB+ GPU (A100/H100)**
+- I2V: [wan_i2v_distill_4step_cfg.json](https://github.com/ModelTC/LightX2V/blob/main/configs/distill/wan_i2v_distill_4step_cfg.json)
+- T2V: [wan_t2v_distill_4step_cfg.json](https://github.com/ModelTC/LightX2V/blob/main/configs/distill/wan_t2v_distill_4step_cfg.json)
+
+**For 24GB+ GPU (RTX 4090/3090)**
+- I2V: [wan_i2v_distill_4step_cfg_4090.json](https://github.com/ModelTC/LightX2V/blob/main/configs/distill/wan_i2v_distill_4step_cfg_4090.json)
+- T2V: [wan_t2v_distill_4step_cfg_4090.json](https://github.com/ModelTC/LightX2V/blob/main/configs/distill/wan_t2v_distill_4step_cfg_4090.json)
+
+
+5. Run inference
+```
+cd scripts
+bash wan/run_wan_i2v_distill_4step_cfg.sh
+```
+
+#### Documentation
+- **Quick Start Guide**: [LightX2V Quick Start](https://lightx2v.readthedocs.io/en/latest/getting_started/quickstart.html)
+- **Complete Usage Guide**: [LightX2V Model Structure Documentation](https://lightx2v.readthedocs.io/en/latest/getting_started/model_structure.html)
+- **Configuration Guide**: [Configuration Files](https://github.com/ModelTC/LightX2V/tree/main/configs/distill)
+- **Quantization Usage**: [Quantization Documentation](https://lightx2v.readthedocs.io/en/latest/method_tutorials/quantization.html)
+- **Parameter Offload**: [Offload Documentation](https://lightx2v.readthedocs.io/en/latest/method_tutorials/offload.html)
+
+
+#### Performance Advantages
+
+- ⚡ **Fast**: Approximately **2x faster** than ComfyUI
+- 🎯 **Optimized**: Deeply optimized for distilled models
+- 💾 **Memory Efficient**: Supports CPU offload and other memory optimization techniques
+- 🛠️ **Flexible**: Supports multiple quantization formats and configuration options
+
+
+### Community
+- **Issues**: https://github.com/ModelTC/LightX2V/issues
+- **Discussions**: https://github.com/ModelTC/LightX2V/discussions
+
+## ⚠️ Important Notes
+
+1. **Additional Components**: These models only contain DIT weights. You also need:
+   - T5 text encoder
+   - CLIP vision encoder
+   - VAE encoder/decoder
+   - Tokenizers
+
+   Refer to [LightX2V Documentation](https://github.com/ModelTC/LightX2V/blob/main/docs/EN/source/deploy_guides/model_structure.md) for how to organize the complete model directory.
+