With all the chatter about 3D model repositories lately, we thought we’d add another twist to the puzzle. Could a virtual world act as a 3D model repository? There are many virtual worlds available today, including Second Life and grids built on its open source cousin OpenSim, the upcoming BlueMars, There and others. Each of these services uses 3D technology in a very fundamental manner.
That said, could they act as a repository, where model-seekers visit and come away with a model in hand?
We suspect this is possible, but perhaps not optimal.
Today most 3D print users already have at least reasonable 3D modeling software on their machines, and they would not require a new way to view models. However, we suspect in the future there will be many more less sophisticated users seeking models, users who probably won’t ever have Solidworks or similar 3D artillery in hand.
But they might have access to a 3D virtual world.
At that point you might imagine a novice model-seeker visiting a virtual world and walking through displays of hopefully relevant 3D models. They’d get quite a good view of the object, since they could theoretically pick it, walk through it and view it from all angles without having to understand complex 3D modeling software.
One major barrier to this concept will be the proprietary nature of some virtual worlds, who wish to “keep” the content within their grid. Second Life, for example, offers no proper import/export tools at this time.
Of course, the problem of finding the right model remains. There may be work underway attempting to automatically generate keywords for 2D images, but we’re not aware (yet) of similar research for 3D models.
You might want to check http://www.probado.de as they have developed means to generate metadata from non-textual formats, including 3D models.
Some thoughts:
SL operates on a primitive-based modeler, for the reason that this is a bandwidth-efficient method of transferring model data which has the added benefit of being comparatively easy to "drive" by a user.
Mesh-based systems like STL files on the other hand are pretty bandwidth-intensive, especially if you want UV mapping for colors. That said there are certainly ways around this problem:
* Polygon-culled "preview versions" of mesh models can be constructed automatically.
* Small repositories (small being a very relative term) could be updated by CVS-like processes and rendered locally. (This is how World of Warcraft works.)
* Exporting Primative-based models to STL might not be prohibitive as long as a good boolean-union operator is available. It might even be preferable given the ease-of-use advantages.
As to the user's experience, this would have the advantage that a community of users would share incentive to organize things well, and with self-organizing frameworks such as a marketplace, a shape library community could have much of the benefits in terms of finding what one is after found in games like SecondLife…