Space-time models (or 4D models) are descriptions of dynamic activities of the real world, that can be stored, analyzed by computers and visualized at distance. Digital imaging and computing technologies of today have matured to a level capable of recording human activities in rich three dimensional detail, and across a length of time. Such photographic 4D models shall in the future be the equivalents of photographic video of today, and produce equivalent artifacts of human culture and heritage. In this thesis, we are concerned with the problem of building such space-time models from the bottoms up through a multi-camera environment. Our contributions are primarily towards estimating motion, both at a coarse and at a dense level from surfaces reconstructed in this fashion.