Ortho-Pictures: 3D Objects from Independent 2D Data Sets

Abstract

This work portrays several ways in which a pair of completely independent 2D data sets, such as regular 2D color pictures or height fields, could be merged into one 3D object, creating an Ortho-picture. The Ortho-picture is an object that portrays the first input set from one view and will identify with the second independent set from an orthogonal view. While techniques to reconstruct 3D geometry from several 2D data sets of the same 2D models are well known in image processing, herein we strive to merge pairs (or even triplets) of completely independent 2D input data sets into one 3D object.

The end result of this effort is a regular 3D object that is automatically synthesized from two (or even three) completely independent pictures or 3D objects (converted into 2D height fields via a Z-buffer). This result is in line of the artwork of conceptual artists such as Shigeo Fukuda [Fukuda] and Markus Raetz [Raetz]. We present several, fully automated, ways to create Ortho-pictures, and show some examples.