Using a Camera to Capture and Correct Spatial Photometric Variation in Multi-Projector Displays

Abstract

Photometric variation in multi-projector displays is arguably the most vexing problem that needs to be addressed to achieve seamless tiled multi-projector displays. In this paper, we present a scalable real-time solution to correct the spatial photometric variation in multi-projector displays. A digital camera is used to capture the intensity variation across the display in a scalable fashion. This information is then used to generate a per-pixel attenuation and offset map for each projector. The former achieves intensity scaling while the later compensates for differing black levels at different regions of the display. These maps are then used to modify the input to the projectors to correct the photometric variation. The contributions of our photometric capture and the correction method are the follwing. First, despite the limited resolution of the digital camera, this method is scalable to very high resolution displays. Second, unlike previous methods, this does not require an explicit knowledge of the location of the boundaries of the projectors for generating the attenuation and offset maps. Third, this compensates for the varying black level of the display. Finally, the correction is implemented in real time for OpenGL applications using the distributed rendering framework of Chromium. 1