Three developments have crossed my radar screen that signal the arrival of a second generation of 3D graphics technology. I offer them as examples for your investigation.

RealVIZ, a San Francisco-based firm, has created a modeling application named ImageModeler which is being made available in versions for Maya, MAX, Lightwave and Softimage. I have seen some very interesting technologies to automate the modeling process, all of which involve some kind of special camera or similar setup to perform "3D scanning" of physical objects. ImageModeler works differently. It generates a fully textured 3D mesh using only photographs of the intended object, taken from a sufficient number of views. It is apparently mathematically possible to reconstruct the 3D structure of an object by reconciling different 2D projections of it, as represented by the photographs. The photographs naturally supply the source of the texture image, which is also mapped automatically. I have reviewed, though not yet performed, the downloadable tutorial the company offers, and the concept is extraordinary. It seems to me precisely as exciting as when the first researchers figured out how to generate 2D rendered images from 3D model data. But this is the exact opposite – interpreting 2D data to produce 3D models.

Realitisic animation of complex object interactions is incredibly difficult. What if you could assign physical qualities to the objects in your scene and allow the computer to figure out how they would naturally interact? The top tier programs, like Maya and MAX, already offer dynamic simluations, which typically take a long time and a lot of computer power to solve. But Havok, an obviously brilliant group in the Silicon Valley, has figured out how to perform dynamic simulations interactively, in real time. Once you set up your scene with the assigned physical properties you can simply play with the models interactively. It's scary to interactively drive a vehicle around in a scene and knock over obstacles in a realistic way. The obvious implementation of this technology is directly within interactive games engines, but you can also use it to record keyframe animation in 3D Studio MAX using the MAX Havok plugin. Download a demo version from the site and prepare to stretch your mind. You'll need a good quality OpenGL graphics card for this program.

The third example comes from my new employer, Eyematic Interfaces. Commercial character animation is, above all, facial animation. This is the area in which 3D has failed to live up to the promise of supplanting traditional cel techniques (in their current computer-assisted forms). You can draw a human facial expression far more easily and attractively than you can morph a 3D facial mesh. Commercial facial animation in 3D is incredibly tedious and necessarily expensive.

Eyematic has developed facial sensing technology that can record the facial movements of an actor as keyframed animation. I have seen it in action a couple of times. In one case, an animator sat in front of a 3D scene consisting of a Halloween Jack O'Lantern with characteristic eyes, nose and mouth. As the animator spoke into a tiny webcam, the pumpkin tracked his every expression and lip movement. Such technology promises to make character animation far cheaper and more available than it currently is. It points to a new direction in animation in which well-known actors drive an animated character in all respects. The remarkable achievement of Sean Connery's animated dragon character will become routine.

A word to the wise. Stay on top of this second generation technology or risk discovering that the skills you have worked so hard to refine have gone the way of the typewriter.