Light Mixture Estimation for Spatially Varying White Balance

Eugene Hsu, Tom Mertens, Sylvain Paris, Shai Avidan, Frédo Durand / SIGGRAPH 2008

Paper (24M PDF) / Paper (1.9M PDF) / Video (4.3M MP4) / Slides (77M KEY)


Abstract

White balance is a crucial step in the photographic pipeline. It ensures the proper rendition of images by eliminating color casts due to differing illuminants. Digital cameras and editing programs provide white balance tools that assume a single type of light per image, such as daylight. However, many photos are taken under mixed lighting. We propose a white balance technique for scenes with two light types that are specified by the user. This covers many typical situations involving indoor/outdoor or flash/ambient light mixtures. Since we work from a single image, the problem is highly underconstrained. Our method recovers a set of dominant material colors which allows us to estimate the local intensity mixture of the two light types. Using this mixture, we can neutralize the light colors and render visually pleasing images. Our method can also be used to achieve post-exposure relighting effects.

Paper Results

The following thumbnails link to full-resolution, lossless-compression, 16-bit, PNG-format images for all results in our paper. They are about 10 megabytes each.

Below, we compare the results of standard white balance (left), the Local Color Shifts model [Ebner 2004] (middle), and our technique (right).

Average White Balance Local Color Shifts Our Result





















Below, we use synthetic input images (left) to produce ground truth results (center) and compare them to our results (right). Difference images are shown in our paper.

Input Image Ground Truth Our Result






Light mixtures can be used to achieve interesting lighting effects. In this set of images, we modify the lighting to achieve a sunset effect.

Input Image Our Result New Lighting



In the following set of images, we dim the interior and exterior lights.

Our Result Dim Interior Dim Exterior



Stay tuned for additional results!

Acknowledgements

We would like to thank the Adobe Creative Technology Labs, the MIT Computer Graphics Group, and Bill Freeman for their helpful advice. Frédo Durand acknowledges a Microsoft Research New Faculty Fellowship and a Sloan Fellowship. Tom Mertens acknowledges a research fellowship from the Belgian American Educational Foundation. Part of the research at Expertise Centre for Digital Media is funded by the European Regional Development Fund and the Flemish government.