Capturing relightable images using computer monitors

Image based relighting techniques are a popular choice for generating photo-realistic images of objects under any lighting condition. A typical process for creating such a model involves photographing the object lit from different directions using a hemispherical lighting array. While generating high-quality results, this requires a customized hardware construction that puts the technology beyond average consumer reach. We investigate the use of monitor illumination as a light source for capturing relightable models, and show that comparable accuracy can be achieved using commodity hardware. In addition, monitors allow increased flexibility in terms of adaptive sampling of light positions. We also present a method of correcting extrapolation artifacts by augmenting hemispherical harmonics with a low order model when reconstructing novel lighting conditions outside the originally sampled region.

• Type: Journal
• Year: 2012
• Where: CVPR
• Authors: Prabath Gunawardane, Steven Scher, James Davis
• Keywords: Adaptive Sampling, Commodity Hardware, Comparable Accuracy, Computer Vision, Cvpr 2012