Creating computer-based 3D models is getting even faster thanks to a new wave of software.

The word "new" is relative in this case because, as it turns out, this sort of technology has actually been around for some time.

First off, here is a look at 'Video Trace', a piece of software developed by the good folks at the Australian Centre for Visual Technologies (The University of Adelaide).

Rather than write a small essay, have a look at this pretty self explanatory video from the Video Trace team:

I'm sure you can see the great potential here for modeling existing structures. For example, this sort of technology could be used in a maintenance sense. An occasion I know involved a maintenance company modeling a factory boiler so that they could plan their maintenance schedule as efficiently as possible and thus minimize downtime for the factory. With this sort of technology that modeling phase might involve quickly running a video camera over and through the boiler and the rest being done remotely, as opposed to hours of measuring and manual modeling.

The potential for video effects and structural surveying is also very apparent from the video.

You can read more about Video Trace on the ACVT page. This has links to a paper on the project, a Google Tech Talk, and much more.

In the same vein we have ProFORMA (Probabilistic Feature-based On-line Rapid Model Acquisition, in case you have a penchant for full names), developed by the Department of Engineering at The University of Cambridge.

The difference, as far as I can tell, between Video Trace and Pro FORMA is, in layman's terms, automation.
 Whilst Video Trace uses a data collection phase (someone going out and taking video), and a modeling phase (drawing on the video to contribute to creating the model), Pro FORMA carries this process out live, or "on-line".

This is done through recognition of certain points on the object and tracking them as the model is moved to create a point cloud (basically a cloud of dots that each represent a certain point on the object). A process called Delaunay Tetrahedralisation (I don't know what this means, email me if you do otherwise smile and nod) is then carried out, followed by probabilistic tetrahedron carving (once again, smile and nod) and finally texture mapping the photographic "skin" onto the resulting model.

So we have two, somewhat similar, methods for obtaining geometric information about the real world very quickly and very efficiently. There is a long way to go before manual modeling of the existing world becomes redundant but these first tentative steps are very exciting to witness

I for one will be keeping a close eye on how these projects progress and will be updating this blog with any new developments.

Drop me a line at ash@coastalconcepts.org if you want to discuss this, or any other relevant matters further.