Expressive 3D Components for Building Simulation and BIM

The transfer of sophisticated Non-Photo Realistic (NPR) rendering technology from the entertainment industry (animation and video games) to the AEC industry will have far-reaching effects on building simulation and on data-driven Building Information Modeling (BIM). The so called "NPR shaders" are small graphics programs that plug into software or reside natively on powerful graphics cards. They enable the same hard-edged 3D geometry to be flexibly re-presented in a myriad of artistic styles, for example, sketchy, watercolor, oil painting, or any of several other traditional analog media styles (see Figures 1 and 2). While these real-time rendering styles are currently applied to all objects in the 3D scenegraph (a scenegraph is a general data structure commonly used by vector-based graphics editing applications and modern computer games), ongoing developments in NPR now permit shading at per object and per pixel scales. Soon, each 3D object in a scenegraph (or even any part of it) can be rendered in its own unique graphical style.

Expressive Graphics

Because NPR shaders enable multiple artistic styles to be applied to traditional "CAD" geometry, I refer to them as expressive graphics, and will use that term in favor of the less descriptive "NPR" acronym. Expressive graphics will become key technology in building simulation and BIM at a number of levels. Most obviously, they will enable 3D designers to express their design intent and relative levels of design maturity graphically. For the first time, 3D geometry can be used for something as subtle and important as personal and interpersonal design communications. Expressive graphics will enable designers to quickly explore initial concepts in "soft" 3D, rather than render them out in hard-line or photorealistic images indistinguishable from as-built photographs. This same expressive geometry can then subtly be re-purposed for client presentation—without conveying that unmistakable "hard-edged" feeling of design immutability. While such uses may sound subtle, romantic, or even peripheral to something as staid as information modeling, expressive graphics will simply make 3D more relevant to the architectural design process, enabling 3D building designs to be "born" digitally.

Expressive graphics will also become an essential enabler of Building Information Modeling (BIM). As a highly collaborative process, BIM requires what can without exaggeration be characterized as "extreme collaboration." To be included in this highly pluralistic process, designers will need more expressive graphics for shared mark-ups of evolving 3D building designs. One can imagine for example, an architect in Bombay wanting to "redline"' part of a 3D building design for review by another consultant and architect based in London. At the moment, while some file-based 3D markup software exist (see Adobe Acrobat 3D), direct 3D markup functionality or true "in-world" editing will be needed for 3D model servers in high-bandwidth environments. Designers would like to use interactive building simulations to explore and document emergent building design(s) from the earliest outset. To realize this potential, however, designers will need to share sophisticated new visible languages that expressive graphics enable, in addition to the necessary technologies of model servers, high bandwidth, interoperability, 3D "layer standards," and others.

Expressive graphics are also the unexpected key to rationalizing investment in BIM for the AEC industry as a whole. Building simulation and BIM are currently but one aspect of the larger delivery pipeline that Seletsky calls digital design (see AEC Viewpoints #19). Restated with different emphasis, BIM cannot yet span the full length of the AEC project pipeline: from conceptual to schematic design, client presentation, cost estimation and construction documentation, fabrication, transportation, building, and facilities management. This limitation stimulated contention and visioning as to "How-to-BIM?," questions of "Where, When, Who, Why, Return-on-investment (ROI)," a debate that is polarizing the AEC industry. A major impediment to industry-wide BIM penetration is the de facto use of BIM as an errors and omissions (E&O) check on already "constructable" building designs. This downstream application further separates those who design from those who construct buildings, which is both unfortunate and counterproductive. The essential search for a "first use" and ROI for building simulation and BIM are being badly misdirected.

By exploiting such new technologies as expressive graphics, it is conceptually feasible to span the entire AEC services pipeline using the same highly expressive 3D building components. While some might argue that the expressive visualization of building components could not hamper something as important as BIM, I contend that many of the 2D drafting languages and drawing conventions currently in use were developed over decades, if not centuries, and are now being taken for granted to the detriment of the industry. These existing visible languages must now be reinscribed in modern technologies and 3D idioms. Were it not for the availability of new technology itself to enable this re-invention, the prospect of dimensionally enhancing our collective visible languages in something close to real-time would be overwhelming, rendering the clear promise of BIM impotent. Fortunately, however, object-oriented technology is clearly capable of adding expressive visualization to its existing feature set.

Object-Agents

Expressive graphics are an example of what I call aesthetic parametrics. By making the visual display of a 3D object simply another parameter of the object, we can build an essential bridge between 3D visualization and 3D information modeling, enabling 3D software and designers alike to readily exploit expressive graphics using existing object-oriented representation. But what if simple graphical expression were taken one step further: reframing 3D building components not simply as chameleon-like visual objects with or without non-geometric data attached (IFCs, aecXML, etc.), but viewing them instead as highly integrated, expressive, and even organic virtual entities that have multiple dimensions of self-interest and expressive behavior? In this scenario, the 3D object is no longer a passive artifact and therefore, these highly expressive conceptual entities might be called object-agents. Hypothetically at least, the same highly flexible 3D object-agent could effectively transcend the full ecology of digital design, including the entire length and diameter of the AEC project pipeline.

Technologically speaking, the idea of object-agents is not new. Rodney Brooks at MIT has been working on physically embodied object-agents (otherwise known as robots, see Figure 3) for more than 30 years, and he has come to understand that much of what we as humans or robots know is a direct result of our having a physical body that mediates our understanding of the built environment. In other words, intelligence as we know it, is very likely impossible without a physical embodiment, and Brooks' research is a vast treasure trove awaiting those that would like to mine it.

However, the idea of applying object-agent technology virtually to the integrated delivery of AEC projects is relatively new, and it has immediate and serious economic consequences. If the industry could reframe its current conception of 3D building components from passive objects to proactive object-agents—virtually embodied robots capable of flexibly spanning and supporting the entire project delivery pipeline (see Figure 4)—it could immediately stop seeking an ROI from one specific stage in the project pipeline. Instead the digital design, as an organic collection of 3D object-agents, could mature dynamically with the ongoing project delivery, making the BIM a relatively effortless and integral byproduct of professional services. After all, the basic economic premise of BIM itself is that construction documents are a byproduct of 3D simulation. Why then shouldn't building simulation itself simply be an extraction of integrated services delivery, as opposed to its current point load on the AEC project pipeline?

While the idea of such "automagical" AEC object-agents may at first sound far fetched, their development would nevertheless be a huge step forward. Designers could immediately engage in more productive conversations and efforts among industry professionals; CAD vendors could stop finessing the ROI and interoperability issues; each segment of the AEC delivery pipeline could stop sniping at one another and begin working on consensual object-agents capable of supporting existing delivery processes. This holistic view of building simulation, building components, and service delivery appeals to designers' innate sense of aesthetics and a collective wisdom. The decision to adopt new technology is, at root, a gut-level act of faith. Early adopters bet that many small obvious benefits (and hopefully a few large non-obvious benefits) will justify the considerable additional sunk costs of investing. Others adopt reactively because competitors have already committed and are gaining a competitive advantage. Practically, however, industry migration to 3D digital design and BIM will come when the AEC industry as a whole believes there is sufficient "shared fate" to ensure against any existing and future unknowns.

Flexible AEC object-agents would need additional behaviour that is fundamentally different from traditional parametric technologies. They would need to be capable of "evolving" from the notional objects used in early conceptual design into schematic objects, systems-level objects, verified objects, and ultimately into highly complex and detailed constructable objects—the product and engineering models necessary for component selection, ordering, fabrication, transportation, construction, as-built objects—and finally, gracefully regressing back into lighter-weight facilities management objects. Such virtuoso levels of integrated component expressiveness and dynamic fidelity are not yet available. They would exacerbate the interoperability challenge if they were. But other communication technologies (i.e., high bandwidth and 3D model servers) are already fundamentally mitigating these challenges. As development proceeds, AEC object-agents will need strong polymorphism, enabling shared 3D object-agents to be collaboratively defined at design time, and thereafter organically and opportunistically re-defined via continuous levels-of-detail based on graphical, geometric, and non-geometric attributes. In this respect, AEC object-agents would almost literally evolve as they mature inside the AEC services ecosystem.

Generative Components

Director of Research at Bentley Systems, Dr. Robert Aish, and the Smart Geometry Group have recently been showcasing what Bentley calls generative components (GC), which are essentially algorithmically generated 3D building components and require a high degree of programmatic skill. By encoding different design constraints and then using code to generate components, highly unique and organic 3D responses to variable building constraints can be achieved (see Figures 5 and 6).

The key aim of the generative component research is currently design exploration and experimentation. However some proponents envision eventually harnessing generative powers to design a building in a single day, for example, or to move sliders interactively on a generative component model, rapidly generating 20-30 different design solutions. Though generative components are still beyond the reach of all but the most avant-garde AEC firms, they are intriguing. In many respects, building design components grow algorithmically from virtual environmental pressures. Generative components are therefore more akin to research results in disciplines such as Artificial Life (A-Life) than to anything we currently characterize as 3D building components. Nonetheless generative components demonstrate the clear potential of design-driven parametrics.

Desired 3D Component Capabilities

Generative components and expressive graphics are both design-driven applications of aesthetic parametrics and object polymorphism—as opposed to the traditional "CAD" parametrics widely employed for pedestrian uses such as catalogue compression, or rapid refactoring (e.g., the proverbial "parametric staircase"). The dynamic fidelity of 3D object-agents, however, is far less complicated than that required of generative components and also is much closer at hand; especially if AEC object-agents are collectively embraced as a performance standard—a collection of desired 3D component capabilities, rather than prescriptive standards or file-based industry implementations.

The missing performance standard for these would-be AEC object-agents is a level of genetic expressiveness. This is a small suite of innate component capabilities that includes:

1. The ability of a 3D building component to opportunistically change its visual appearance as it matures.
   
2. The capacity for virtual embodiment, i.e., to refactor its basic geometry (parametrics) and object-type (polymorphism).
   
3. The capability to have self-interest and modify behaviours in response to environmental pressures.

While this type of digital DNA™ may again sound far-reaching, several of these genetic capabilities already exist in the marketplace as partial implementations. In SketchUp, for example, simple genetic expression is already possible in 3D components. SketchUp offers the capacity to achieve simple expressive graphical styles, to change simple 3D parameters, and for designers to paramorphically extend the form factor of 3D components interactively, using relatively sophisticated "in-world" component editing features.

These genetically expressive capabilities do not yet include more traditional CAD parametrics, or object-agent self interests, nor do they easily extend to the hidden data-definitions of 3D building components. However, even these capabilities may soon be within reach. More complete genetic expression is also likely to be forthcoming by various offerings in the near future, including of course, the already available export to more fully featured object-oriented modeling environments. Many traditional CAD and BIM providers are eager to build a bridge with Google's SketchUp. However it would be a strategic mistake for them to simply assume that SketchUp is a sexy "front end" that feeds into their "more mature" BIM packages. The fundamental component capabilities of SketchUp, given skilled and ongoing hybrid development, could quickly outpace old-line CAD parametrics, yielding highly expressive 3D building components with object-agent like capabilities that can first prime, and then optimize digital design workflow.

Conclusions

The AEC industry is quietly undergoing real technology transfer. Multi-million dollar investments in real-time rendering technology made by entertainment industries will soon benefit 3D building simulation and better enable virtual delivery of AEC professional services. Aesthetic parametrics in general, and expressive graphics in particular, will greatly enhance the use of 3D modeling from the earliest design stages, thereby enabling a more holistic view and graduated implementation of BIM—priming the AEC project pipeline for downstream reuse of flexible 3D assets. If implemented in a genetic framework, expressive 3D building components can be incrementally enhanced via ongoing project workflows and maintain dynamic fidelity by means of new object-agent capabilities. Several of these highly flexible genetic capabilities already exist in next generation 3D modelers. If collectively developed further by the AEC industry as a performance-based standard, genetic expressiveness can provide the flexibility and the information "bandwidth" that 3D building components will need to span digital design as a whole, thereby rationalizing economic investment in building simulation and BIM as an extraction of integrated AEC services delivery.

About the Author

Fred Abler is CEO of FormFonts.com, a subscription-based online 3D model library for virtual world making. FormFonts™ specializes in professionally developed building components and provides a multi-platform database for various 3D modelers and Building Information Modeling applications. Abler is also a design instructor and researcher in the Architecture Department of the College of Architecture and Environmental Design at CAL POLY San Luis Obispo, CA. He has 15 years experience in the software industry working on both research prototypes, and fielded Spatial Decision Support Systems (SDSS) for building design, maritime logistics, and the US DOD. He can be reached at fabler@gmail.com.

© 2006 Fred Abler. All rights reserved. No part of this article, in its entirety or separately, may be used, copied, or quoted, without the author's express written consent.

FormFonts™, Aesthetic Parametrics™, and, Digital DNA™ are trademarks of FormFonts Inc. All rights reserved.

Note: The views expressed in Viewpoint articles are those of the individual authors and do not necessarily reflect those of AECbytes.