Using image-based modeling for exporting 3D content from Second Life

• Authors: Ajay Surendranath;Geetika Sharma
• Venue: Proceedings of the 8th International Conference on Virtual Reality Continuum and its Applications in Industry
• Year: 2009

There is a rapidly growing community of users of online virtual environments like Second Life [SL], Opensim [Os] and so on. These environments allow users to create content in the form of 3D graphical models which can be used to build up spaces in the virtual environment. Some of these virtual worlds, e.g. Second Life, do not allow users to export self-created content out of the environment. This is a grievance with many users who may have spent a lot of time and effort in creating a model and would like to reuse their content. In this paper, we present a method to recreate graphical 3D models in virtual environments from their images using multi-view reconstruction techniques. We employ a point-based approach which necessitates complete control over the model to obtain the information required for reconstruction. This precludes the use of our technique for piracy of 3D models. Further, we argue that image-based modeling techniques should perform well in this scenario since the inputs can be controlled and are virtually noise-free. The reconstructed model can then be written in a standard 3D format and reused as desired. We show the results of some initial experiments towards this goal.