First, thanks to group member @Ganzy for providing the images so I didn't need to search for them.

FP8 (e4m3fn scaled) precision, merged with 4-step distilled LoRA, optimized for low-VRAM deployment. Model download link: https://huggingface.co/lightx2v/Qwen-Image-Edit-2511-Lightning

Qwen-Image-Edit-2511-Lightning is a series of models optimized for image editing tasks, leveraging step distillation and quantization techniques to achieve efficient inference performance.

Usage Instructions This model suite supports two mainstream frameworks. Detailed guides are provided below:

1. Qwen-Image-Lightning Framework For complete documentation on model usage within the Qwen-Image-Lightning ecosystem (including environment setup, inference pipeline, and customization), please refer to: Qwen-Image-Lightning GitHub Repository https://github.com/ModelTC/Qwen-Image-Lightning/

2. LightX2V Framework These models are fully compatible with LightX2V, a lightweight video/image generation inference framework. For step-by-step usage examples, configuration templates, and performance optimization tips, please refer to: LightX2V Qwen Image Editing Documentation https://github.com/ModelTC/LightX2V/blob/main/examples/qwen_image/README.md

Key Optimizations Step Distillation: The LoRA model reduces the original inference steps to just 4 steps, achieving a significant speedup (≈10x faster than standard 40-step inference) while maintaining image editing quality. FP8 Quantization: Quantizing the base model balances performance and resource efficiency, reducing GPU VRAM usage by approximately 50% compared to FP32 while maintaining editing fidelity.

Support For technical issues, feature requests, or integration issues:

Please submit an issue in the Qwen-Image-Lightning repository (for Qwen framework-related issues): https://github.com/ModelTC/Qwen-Image-Lightning/issues

Submit an issue in the LightX2V repository (for LightX2V integration issues): https://github.com/ModelTC/LightX2V/issues