Training-free diffusion for controlling illumination conditions in images

Open Webpage

Xiaoyan Xing, Vincent Tao Hu, Jan Hendrik Metzen, Konrad Groh, Sezer Karaoglu, Theo Gevers

Published: 01 Jan 2025, Last Modified: 05 Nov 2025Comput. Vis. Image Underst. 2025EveryoneRevisionsBibTeXCC BY-SA 4.0
Abstract: Highlights•Diffusion models can be treated as black-box renderers.•Diffusion energy function can affect lighting change in image synthesis.•Training-free relighting can be achieved with proper physics-based constraints.