RealityPrime

Semantic Modeling

Parametrics are just one way of encapsulating more useful information (and if your eyes are glazing over, skip to page 3). What we really need is a new language of object representation that encapsulates and preserves form and function, aesthetics, style, meaning, and behavior, all tightly coupled and never discarded in the "art pipeline" until the object is finally rendered on your screen. And the big problem here is that things like semantics are so far from concrete math that any program, even if it supports the concept, can have its own varying interpretations. So this language needs to be fully expressed, down to a fairly programmatic level, so that these assumptions are clear and enforced. It should contain the instructions on how to render the 3D object, but also how to create it, use it, kick it, break it, change it, and even say what it is.

If we can do that (and I’m actually working towards this as part of something else), then the marketplace of 3D objects makes much more sense. We might avoid having 3,000 similar versions of a door. If I want to make a "Door," what I’d really like is a Meta-Door, a door-generating unit that I can take and tweak. I should be able to tune the resulting door to wood or stone, thick or thin, modern or art deco. The Door should also know it’s a door, and if plugged into a wall, should know how to open and close and let people through. It should accept a new Knob if one is given and even know how to get kicked in or just grow old. And anyone looking for a particular kind of Door should be able to search through the thousands of specific instantiations that people create, or they should be able to make their own unique combination, use it and sell it. Certainly the Meta-Door is worth more, but that’s because it’s actual technology (i.e., "know how") beyond the normal art. It’s a canvas that knows how (not what) to paint.

The language should retain authorship information on Meta as well as instanced objects. Now, I’m not a fan of DRM by any means. But it seems like there are two common options: sell basic points and polygons that can be easily copied and thereby lose all sense of history; or make such a closed DRM-laden system that you shoot your customers in the foot.

There’s a third option, which is to embed creator information in each fully-expressed object, as above, so that when I use your door in my house, my house knows that a part of it was created by you. Will it stop me from using it? In my opinion, it shouldn’t even try, not once you made your door available to others. But if I sell my house, you should get a proportional cut for your doors. If I give it away for free, you should at least get some credit. The system should be able to handle this kind of transaction with ease.

The theory is that the fully-specified object is more likely to retain authorship information over generations because it never goes through a translator. That’s because the object is not just data, but procedure, self describing and hopefully importable into any system that can load a software library. Now, certainly, any dedicated enough hacker can always alter those procedures to strip out authorship, watermarks, or what have you. But history shows that if you produce an easy path for your customers with reasonable prices, most people will go along. Most people will even want to pay a reasonable price if they know that the actual authors, not just the aggregators or middle-men, get payment and/or credit for their work. That’s true for the big 3D blockbuster movie you’ll make and the doorknobs you’ll download off the web. That’s the theory, anyway.

Pages: 1 2 3

Tags: Articles, Featured //