Enabling Level-of-Detail Selection for Exterior Scene Synthesis

This work presents a method to enable matching of level-of-detail(LOD) models to image-plane resolution over large variations inviewing distances often present in exterior images. A relationshipis developed between image sampling rate, viewing distance,object projection, and expected image error due to LOD approximations.This is employed in an error metric to compute error pro-filesfor LOD models. Multirate filtering in the frequency space ofa reference object image is utilized to approximate multiple distantviews over a range of orientations. An importance samplingmethod is described to better characterize perspective projectionover view distance. A contrast sensitivity function (CSF) isemployed to approximate the response of the vision system. Examplesare presented for multiresolution spheres and a terrain heightfield feature. Future directions for extending this method aredescribed.