Creating Raster Omnimax Images from Multiple Perspective Views Using the Elliptical Weighted Average Filter

Abstract

Creation of Omnimax animation by computer opens up fantastic new visual possibilities. Unfortunately, the fish-eye distortion of Omnimax film images complicates synthesis by computer, since most image-synthesis programs can create only perspective views. As an alternative to modifying existing image-synthesis programs to produce Omnimax projections directly, we present a method for creating them from multiple perspective views. Four perspective views of the environment are created, each a projection onto a face of a cube centered at the camera, and then a mapping program creates an Omnimax projection from them. To minimize aliasing during resampling, the mapping program uses the elliptical weighted average filter, a space-variant filter we developed for this application that computes a weighted average over an arbitrarily oriented elliptical area. This filter can also be used for texture mapping 3D surfaces.