We demonstrate "3D-Mode," a software system that builds 3D models of objects and scenes by taking a few (minimally 2) photographs using a digital camera at possibly largely separated positions. It has recently been commercialized by 3D Media Co., Ltd. 3D-Mode has the following steps: (1) taking photos; (2) manually obtaining and matching a few feature points on the object; (3) automatically computing 3D structure and motion; (4) automatic Delaunay triangulation and manual modification; (5) automatic acquisition of texture for each triangular patch; (6) if necessary, assign a new coordinate system and scale of space. The system does not need input of camera parameters, but requires manual input of feature points in a special manner such that matching information is input simultaneously. Once epipolar lines are available, they are used to help locate corresponding points in other images.

In this paper we show that given two homography matrices for two planes in space, there is a linear algorithm for the rotation and translation between the two cameras, the focal lengths of the two cameras and the plane equations in the space. Using the estimates as an initial guess, we can further optimize the solution by minimizing the difference between observations and reprojections. Experimental results are shown. We also provide a discussion about the relationship between this approach and the Kruppa equation.