Portrait Relighting

Portrait relighting aims to realistically change the illumination of a person in a photograph or video, offering precise control over shadows and overall lighting conditions. Recent research focuses on developing methods that achieve photorealistic results, often employing neural networks combined with physically-based rendering models or leveraging 3D face geometry for improved consistency and generalization across different viewpoints. These advancements are driven by applications in augmented and virtual reality, video conferencing, and image editing, where dynamic and realistic lighting control is crucial. The field is also exploring more intuitive user interfaces, such as scribble-based controls, to make advanced relighting techniques accessible to a wider range of users.