GoKawiilResearchers unveil physics-based method to relight photos after image decomposition
Chris Careaga and Yağız Aksoy of Simon Fraser University published a method that splits a single photograph into diffuse albedo, colorful diffuse shading, and specular residual components, moving beyond older models that assumed single-color light and purely matte surfaces. The work, detailed in their SIGGRAPH Asia 2024 paper, received an ACM Transactions on Graphics Best Paper Award Honorable Mention. They report the technique enables applications like removing specular highlights and performing per-pixel white balancing on real-world images.
GoKawiil's interpretation of the reporting above, not reported fact.
By breaking intrinsic image decomposition into simpler sub-problems, the researchers say they can estimate complex, colorful lighting effects even with limited labeled training data, a longstanding bottleneck in this field. This could give photo and video editing tools more physically accurate control over relighting and color correction, rather than relying on simplified lighting assumptions that often fail on real photographs. The recognition from SIGGRAPH Asia suggests the approach is seen by peers as a meaningful advance in computational photography research.
- New method separates photos into diffuse albedo, colorful diffuse shading, and specular residual components
- Approach drops prior assumptions of single-color light and purely Lambertian (matte) surfaces
- Enables practical edits like specularity removal and per-pixel white balancing, per the authors
Source: yaksoy.github.io — Sfu Computing Science, 2026-11-11
Published there as: “Physically Controllable Relighting of Photographs”
Read the original report → The summary and analysis above are GoKawiil's own, written from reporting by the source above. Facts and quotes belong to the original publisher.