Project to build a 3D fax machine (original) (raw)

There is a growing interest in the graphics, vision, and manufacturing communities in building an inexpensive device capable of digitizing the shape and external appearance of objects for transmission, display, and numerically controlled fabrication. Applications for such a device include product design, fast prototyping, reverse engineering, and digitizing of shapes for the visual simulation, animation, and entertainment industries.

Toward this end, we have been developing a laser-based scanning system capable of generating an occlusion-free low-level geometric description of the externally visible surfaces of an object. Here is a flow chart of the process. (This flow chart is somewhat out of date.) The crucial step in this process is the merging of multiple range images, each of which captures only one side of the object, in order to produce a seamless description of the entire object. The representations we have investigated for this step include dense polygon meshes and 3D voxel arrays with an occupancy function at each voxel. Both representations eventually lead to a dense polygon mesh. We have also been exploring methods for fitting curved surface patches to these dense meshes. Future investigations include automatically determining the next best view and acquiring surface reflectivity.

As an application of this technology, we have embarked on a multi-year project to create a high-quality 3D archive of the sculptures of Michelangelo for scholarly, educational, and commercial purposes. In other words, the 3D fax machine project has morphed into theDigital Michelangelo Project.

People:

Brian Curless < curless@cs.washington.edu >
Daniel Goldman < dg@cs.stanford.edu >
Hua Ge < huage@leland.stanford.edu >
Venkat Krishnamurthy < venkat@cs.stanford.edu >
Marc Levoy
Kari Pulli < kapu@graphics.stanford.edu >
Greg Turk < turk@redclay.stanford.edu >
Afra Zomorodian < afra@cs.stanford.edu >

Recent papers in this area:

Zippered Polygon Meshes from Range Images

Greg Turk and Marc Levoy

Proc. SIGGRAPH '94

Better Optical Triangulation Through Spacetime Analysis

Brian Curless and Marc Levoy

Proc. ICCV '95

A Volumetric Method for Building Complex Models from Range Images

Brian Curless andMarc Levoy,

Proc. SIGGRAPH '96

Fitting Smooth Surfaces to Dense Polygon Meshes

Venkat Krishnamurthy andMarc Levoy,

Proc. SIGGRAPH '96

New methods for surface reconstruction from range images

Brian Curless

Ph.D. Thesis, June, 1997

Fitting Smooth Surfaces to Dense Polygon Meshes

Venkat Krishnamurthy

Ph.D. dissertation, Stanford University, January, 1998.

Available software and data:

ZipPack Polygon Mesh Zippering Package
VripPack Volumetric Range Image Processing Package(now including source code)
Thescanalyze software package used to align and merge range data in theDigital Michelangelo Project.
Volfill, our diffusion-based hole filler for large polygon meshes.
The Stanford 3D Scanning Repository
An experimental repository of 3D models from the Digital Michelangelo Project.

Demos:

Movies of a zippered polygon mesh model of a plastic toy lion, with explanations.
Renderings and 3D fax of 2-million polygon model of ahappy buddha,
constructed from multiple range images using a volumetric method.

If images on this page look dark to you, see our note about gamma correction.

A list of technical papers, with abstracts and pointers to additional information, is also available. Or you can return to the research projects page or our home page.