In this paper we consider the problem of computing the 3D shape of an unkno
wn, arbitrarily-shaped scene from multiple photographs taken at known but a
rbitrarily-distributed viewpoints. By studying the equivalence class of all
3D shapes that reproduce the input photographs, we prove the existence of
a special member of this class, the photo hull, that (1) can be computed di
rectly from photographs of the scene, and (2) subsumes all other members of
this class. We then give a provably-correct algorithm, called Space Carvin
g, for computing this shape and present experimental results on complex rea
l-world scenes. The approach is designed to (1) capture photorealistic shap
es that accurately model scene appearance from a wide range of viewpoints,
and (2) account for the complex interactions between occlusion, parallax, s
hading, and their view-dependent effects on scene-appearance.