DIGITAL EMBRYO WORKSHOP
EXAMPLE EMBRYOS
GENERAL DESCRIPTION
Digital embryos are computer generated shapes. They are grown by simulating certain mechanisms of embryological development. In particular, morphogen based cell proliferation, programmed cell death, and the forces of attraction and repulsion between cells are simulated.
APPLICATIONS
Anything you might want to do with a computer graphics polyhedron, you can do with a digital embryo. They can be textured, embedded in scenes, ray traced, etc.
Digital embryos were originally developed to fulfill a need for truly novel objects in psychophysics of vision research (Hegde, Thompson, & Kersten, 2006; Brady & Kersten, 2003; Shams, Brady, & Schaal, 2001). However, they also have applications wherever novelty of shape is required. Examples include computer art, video games and films. With future enhancements, digital embryos may also be useful in product and mechanical design.
CONTRIBUTORS
Digital embryos were originally conceived as part of Mark Brady’s Ph.D. thesis on object learning at the University of Minnesota. The original version was written by Mark Brady and it ran on Silicon Graphics computers using the Inventor library. It was written in C/C++. Jay Hegde (University of Minnesota) has suggested ways in which to make digital embryos more useful in psychophysics research. These include the simulation of programmed cell death and evolution. Dr. Hegde also has added his own enhancements to the software. Nathan Gossett (Univeristy of Minnesota) ported the Inventor based program to Open GL so that it would run on Macs and Windows. Dan Gu (North Dakota State University) has overhauled the program by porting it entirely to C# with .NET framework. He has redesigned the graphical user interface, added new texturing algorithms, real time growth and morphogen visualizations, real time scene creations, movie output, and standard file format output. See “What’s New” for more details. Continuing software development is being done primarily by Dan Gu and Mark Brady (at NDSU). The program and documentation are available on this NDSU website. If you have any installation problems, contact huanzhong.gu@ndsu.edu.
USE AND COPYRIGHT
The executable software is available for free via download. Users should cite the following (Brady & Kersten, 2003).
DOWNLOAD
To download the latest version of the software and documentation, click here: DEW Installer (ZIP)
To view a sample movie of the embryo growth process, click here: embryo sample (AVI - 249 MB)
Hegde, J., Thompson, S., & Kersten, D. (2006). Object recognition in cluttered visual scenes: Is it better to learn objects in the presence or the absence
of clutter? Society for Neuroscience Abstracts, 438.19.
Brady, M. J., & Kersten, D. (2003). Bootstrapped learning of novel objects. Journal of Vision, 3(6), 413-422.
Shams, L., Brady, M. J., & Schaal, S. K. (2001). Graph matching vs. mutual information maximization for object detection. Neural Networks, 14(3), 345-354.
This page meets the standards for University, national, and international W3C compliance regulations.
|
|
|
|