README.md

metadata

base_model:
  - Wan-AI/Wan2.2-I2V-A14B
library_name: diffusers
license: apache-2.0
tags:
  - diffusion-single-file
  - comfyui
  - distillation
  - LoRA
  - video
  - video generation
  - SGMD
pipeline_tag: image-to-video

🎬 Wan2.2 Distilled Models (SGMD)

This repository contains distilled versions of the Wan2.2 models using SGMD (Score Gradient Matching Distillation), as presented in the paper SGMD: Score Gradient Matching Distillation for Few-Step Video Diffusion Distillation.

⚡ High-Performance Video Generation with 4-Step Inference

Distillation-accelerated version of Wan2.2 - Dramatically faster speed with excellent quality

---

---

🔥 News

• 2026.04.12: We are excited to release the Wan2.2-I2V-A14B-4step-720p-high and Wan2.2-I2V-A14B-4step-720p-low models. Compared to previous iterations, this version was trained on a high-quality 720p dataset and features an optimized low-noise training algorithm. These enhancements significantly boost the model's performance in fine-grained detail rendering and visual texture.

🚀 Quick Usage (Python)

To use these models with the LightX2V framework for 4-step inference:

from lightx2v import LightX2VPipeline

# Initialize pipeline for Wan2.2 I2V task
pipe = LightX2VPipeline(
    model_path="lightx2v/Wan2.2-Distill-Models",
    model_cls="wan2.2_moe",
    task="i2v",
)

# Enable offloading to reduce VRAM usage
pipe.enable_offload(
    cpu_offload=True,
    offload_granularity="block",
    text_encoder_offload=True,
)

# Create generator for 4-step inference
pipe.create_generator(
    attn_mode="sage_attn2",
    infer_steps=4,
    height=480,
    width=832,
    num_frames=81,
    guidance_scale=[1.0, 1.0],
)

# Generate video
pipe.generate(
    seed=42,
    image_path="path/to/your/image.jpg",
    prompt="A cinematic shot of a sunset over the ocean",
    save_result_path="output.mp4",
)

🌟 What's Special?

⚡ Ultra-Fast Generation 4-step inference (vs traditional 50+ steps) Approximately 2x faster using LightX2V than ComfyUI Near real-time video generation capability	🎯 Flexible Options Dual noise control: High/Low noise variants Multiple precision formats (BF16/FP8/INT8) Full 14B parameter models
💾 Memory Efficient FP8/INT8: ~50% size reduction CPU offload support Optimized for consumer GPUs	🔧 Easy Integration Compatible with LightX2V framework ComfyUI support Simple configuration files

---

📦 Model Catalog

🎥 Model Types

🖼️ Image-to-Video (I2V) - 14B Parameters Transform static images into dynamic videos with advanced quality control 🎨 High Noise: More creative, diverse outputs 🎯 Low Noise: More faithful to input, stable outputs

📝 Text-to-Video (T2V) - 14B Parameters Generate videos from text descriptions 🎨 High Noise: More creative, diverse outputs 🎯 Low Noise: More stable and controllable outputs 🚀 Full 14B parameter model

🎯 Precision Versions

Precision	Model Identifier	Model Size	Framework	Quality vs Speed
🏆 BF16	lightx2v_4step	~28.6 GB	LightX2V	⭐⭐⭐⭐⭐ Highest Quality
⚡ FP8	scaled_fp8_e4m3_lightx2v_4step	~15 GB	LightX2V	⭐⭐⭐⭐ Excellent Balance
🎯 INT8	int8_lightx2v_4step	~15 GB	LightX2V	⭐⭐⭐⭐ Fast & Efficient
🔷 FP8 ComfyUI	scaled_fp8_e4m3_lightx2v_4step_comfyui	~15 GB	ComfyUI	⭐⭐⭐ ComfyUI Ready

---

🚀 Alternative Usage Methods

Method 1: ComfyUI

Please refer to workflow

⚠️ Important Notes

Other Components: These models only contain DIT weights. Additional components needed at runtime:

• T5 text encoder
• CLIP vision encoder
• VAE encoder/decoder
• Tokenizer

Please refer to LightX2V Documentation for instructions on organizing the complete model directory.

🤝 Community

• GitHub Issues: https://github.com/ModelTC/LightX2V/issues
• HuggingFace: https://huggingface.co/lightx2v/Wan2.2-Distill-Models

If you find this project helpful, please give us a ⭐ on GitHub