was the code that ran www.shadowpush.com is composed of 3 parts: Meshlab (for server-side 3D compression), Papervision3D (for displaying 3D models in the browser), and the server-side code written in Python, using Django, for interfacing between the two to set up a software-as-a-service architecture for allowing users to upload and then visualize models online.

Much of the code from Meshlab (www.meshlab.sf.net) and Papervision3D (http://code.google.com/p/papervision3d/) has been copied in, but changes were made to some parts.

Meshlab was extensively bug-checked to prevent it from crashing on opening, reducing, and saving COLLADA 3D files. The time taken to open COLLADA 3D files was reduced by orders of magnitude, by use of 'static' calls and hash tables. A texture atlas generator was also coded, in order to reduce the number of textures necessary to load for rendering. Finally, the output file was optimized and compressed for quick loading by the Flash player and Papervision (by quantizing floating-point numbers, and using offsets instead of full integers). In addition, a binary space partitioning tree generating algorithm was coded, in addition to a parser.

the project has 3 folders: 'merlin' is for the meshlab code, which normally resides in a user's 'home' folder 'winebago' is for the python/django web server code 'foxtrot' is for the flash code