LightIt is a method for explicit lighting control in text-guided image generation using diffusion models. The method uses shading and normal maps to control lighting, enabling high-quality image generation with consistent lighting. It introduces a shading estimation method to generate paired image and shading datasets, which are used to train a control network. The method also includes a relighting module that allows for identity-preserving relighting of images. LightIt is the first method to enable controllable and consistent lighting in diffusion-based image generation, performing on par with state-of-the-art relighting methods. The method uses a single-view shading estimation approach to generate a paired dataset of images and shading maps, which is used to train a control module for diffusion models. The method also includes a relighting application that allows for relighting images using a target shading map. The method is evaluated on various tasks, including image synthesis and relighting, and shows superior performance in terms of lighting consistency and image quality. The method is supported by a comprehensive set of experiments and ablation studies, demonstrating its effectiveness in controlling lighting in diffusion-based image generation.LightIt is a method for explicit lighting control in text-guided image generation using diffusion models. The method uses shading and normal maps to control lighting, enabling high-quality image generation with consistent lighting. It introduces a shading estimation method to generate paired image and shading datasets, which are used to train a control network. The method also includes a relighting module that allows for identity-preserving relighting of images. LightIt is the first method to enable controllable and consistent lighting in diffusion-based image generation, performing on par with state-of-the-art relighting methods. The method uses a single-view shading estimation approach to generate a paired dataset of images and shading maps, which is used to train a control module for diffusion models. The method also includes a relighting application that allows for relighting images using a target shading map. The method is evaluated on various tasks, including image synthesis and relighting, and shows superior performance in terms of lighting consistency and image quality. The method is supported by a comprehensive set of experiments and ablation studies, demonstrating its effectiveness in controlling lighting in diffusion-based image generation.