This paper proposes a novel zero-reference low-light enhancement framework that does not require low-light data or illumination-related hyperparameters. The framework is trained solely with normal light images and leverages a physical quadruple prior derived from the Kubelka-Munk theory of light transfer. This prior serves as an illumination-invariant bridge between normal and low-light images. The framework includes a prior-to-image mapping system trained on normal light images, which can automatically restore the illumination-invariant prior back to images, achieving low-light enhancement. The framework also incorporates a bypass decoder to handle detail distortion and offers a lightweight version for practical use. The model is trained using a pretrained generative diffusion model, and the lightweight version distills the complex multi-step optimization of the large diffusion model into a single forward pass. The framework demonstrates superior performance in various scenarios, with good interpretability, robustness, and efficiency. The model is evaluated on multiple low-light datasets, showing significant improvements over existing methods. The physical quadruple prior captures the essence of imaging under diverse lighting conditions, enabling the model to enhance low-light images without relying on reference samples or artificially set hyperparameters. The framework's ability to learn comprehensive illumination knowledge from the physical quadruple prior and normal light images contributes to its robustness across different scenarios. The model's lightweight version achieves comparable performance while significantly improving inference speed and computational efficiency. The framework's effectiveness is validated through extensive experiments on various low-light datasets, demonstrating its superiority in low-light enhancement tasks.This paper proposes a novel zero-reference low-light enhancement framework that does not require low-light data or illumination-related hyperparameters. The framework is trained solely with normal light images and leverages a physical quadruple prior derived from the Kubelka-Munk theory of light transfer. This prior serves as an illumination-invariant bridge between normal and low-light images. The framework includes a prior-to-image mapping system trained on normal light images, which can automatically restore the illumination-invariant prior back to images, achieving low-light enhancement. The framework also incorporates a bypass decoder to handle detail distortion and offers a lightweight version for practical use. The model is trained using a pretrained generative diffusion model, and the lightweight version distills the complex multi-step optimization of the large diffusion model into a single forward pass. The framework demonstrates superior performance in various scenarios, with good interpretability, robustness, and efficiency. The model is evaluated on multiple low-light datasets, showing significant improvements over existing methods. The physical quadruple prior captures the essence of imaging under diverse lighting conditions, enabling the model to enhance low-light images without relying on reference samples or artificially set hyperparameters. The framework's ability to learn comprehensive illumination knowledge from the physical quadruple prior and normal light images contributes to its robustness across different scenarios. The model's lightweight version achieves comparable performance while significantly improving inference speed and computational efficiency. The framework's effectiveness is validated through extensive experiments on various low-light datasets, demonstrating its superiority in low-light enhancement tasks.