SIGGRAPH method adds physical light control to photos
Original title: Physically Controllable Relighting of Photographs
SFU researchers train a neural renderer with differentiable rendering so users can edit lights like in Blender.
Chris Careaga and Yağız Aksoy describe a self-supervised photograph relighting pipeline in Proc. SIGGRAPH 2025. It estimates a colored mesh from monocular geometry and intrinsics, lets users place lights in 3D, path-traces the scene, and passes that render through a neural renderer. The method is patent pending, with licensing through Simon Fraser University.
- Users can add spot, point, or environmental lights.
- Training uses real photograph collections without paired ground truth.
- The authors say the method is under intellectual-property protection.
From the original article
4 min read at yaksoy.github.io
We present a photograph relighting method that enables explicit control over light sources akin to CG pipelines. Users can insert different types of light sources, such as spot lights, point lights, or environmental illumination into the scene. We achieve this in a pipeline involving mid-level computer vision, physically-based rendering, and neural rendering.
By summarity · yaksoy.github.io
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.