I'm a youngish software developer, and I'm thinking of writing a commercial WebGL application for preparing models for Shapeways (fixing problems, altering wall thicknesses, etc). I have experience in this area (I wrote much of Blender's modeler).
The basic idea is to write a bare-bones modeler in Javascript and WebGL, somewhat akin to SketchUp (though less heavy duty, since this is Javascript). Users could create models from scratch, or import them, and there would be a rich set of tools for resolving problems with model geometry and topology.
Toolset
There would be a basic modeling toolset, which I think is necessary for prepping the models. People could model stuff from scratch, too, though I suspect the main use would be prepping existing, imported models for printing.
Speaking of model prep, here are my ideas for some basic tools (I don't really know what the common problems are with Shapeways models, so feedback would be greatly appreciated):
1. A thickness tool, that would interactively show you parts of your model that are too thin or too thick. The user would then choose from either an algorithm that would try to modify the thickness automatically, or they would modify the model manually (or perhaps a user-controllable auto algorithm, if I can think one up).
2. An interactive wizard for finding and resolving non-manifold topology. The idea would be to make it as easy as possible to fix non-manifold problems, with varying levels of user control.
Curved Surface Issues
The modeler would have some support for subdivision surfaces. That being said, I am well aware that subdivision surfaces are not ideal for manufacturing. How common is it for ShapeWays users to use non-polygonal modeling solutions (e.g. NURBS)? I'd rather not go there if I don't have too, since I have no experience coding patch modelers.
Pricing
I've not worked out pricing yet; I'm thinking of combining a minimum fee with a percentage of the ShapeWays base price. The charge would happen when sending the model to ShapeWays for printing. I'd also like to charge a subscription fee, say, $10 a month, to make revenues more predictable. There would be no ads; in my opinion, advertising revenue is simply too volatile to be worth the hassle for a small start-up. Plus, I'm not keen on giving away yet another modeler for free; if people want free stuff, they can use my work in Blender.
The Poll
I'd appreciate if everyone answered the poll, as it will give me an idea of how much processing will have to be done on the server side as opposed to the client side in JavaScript.
Conclusion
Thank you for your time; I hope the ShapeWays community can help me.