Better workflow comes from better communication.
In the world of mechanical design, that means you need visualization tools to share pictures of your products with the rest of the supply and design chains. Now software tools and super-fast graphics cards are combining with cheaper workstations and plentiful memory to bring that vision to life.
Improving that workflow can boost not only design speed, but also design quality, says Peter Mehlstaeubler, VP of product development for Alias Wavefront (Toronto, Ontario, Canada). He spoke in an interview at the Siggraph trade show here.
"New products have to make the transition from design into engineering. Making that smoother is an ongoing effort," he says. Many companies today have collapsed the design and engineering groups into a single department, but they still use distinct tools, so communication isn't perfect. "When a product is sketched in 2D then evolved into 3D, it leaves room for interpretation if you don't have good enough digital consistency," he says.
Alias' new StudioTools 10 is a "major milestone release" that's designed to solve that problem. Studying designers' habits, Alias found that people spend less time working on overall shapes than on details such as lines, blends, and fillets. So version 10 of this CAID (computer aided industrial design) platform lets you snap design tools (like a paintbrush) to precise engineering shapes (like curves), achieving unprecedented sharpness in creating images. It also lets the user work in either novice or expert mode, varying the length of menus. And it has added a native interface with PTC's Granite CAD format, in addition to Catia, UG, SDRC, and Autodesk. For hardware, these conceptual changes make it a good complement to the Tablet PC, Mehlstaeubler says. Set for release in November, that new computing platform will let users sketch drawings on its touch-screen with a pressure-sensitive stylus.
But none of this would be possible without graphics cards, the guts of a computer that allow it to handle the enormous computation necessary for creating high-quality pictures and animations.
"To evaluate your model from an aesthetic point of view, you need visualization. But rendering can be very time-consuming. So graphics cards are the next big thing that's coming," Mehlstaeubler says. Recent cards from ATI (Markham, Ontario, Canada) and Nvidia (Santa Clara, CA) mean that rendering on the board can now start competing with rendering in software. Likewise, Matrox' (Dorval, Quebec, Canada) Parhelia card gets rid of the "jaggies" and screen-door effect typical of animated wireframe models, and can display that smooth image on three monitors at once, says Alain Thiffault, Matrox Global Software product manager.
Rendering in hardware is significant because you cannot only see your design, but also change it on the spot, with real-time iterations. This enormous computing power is primarily marketed to the video game market, but is equally useful for engineers.
Driving those cards are workstations, which have traditionally been a cost bottleneck. But powerful Pentium chips and competitive pricing means that cost is no longer the limiting issue, Mehlstaeubler says.
At Siggraph, that trend was illustrated by Sun Microsystems' (Santa Clara, CA) XVR-4000 graphics accelerator, Sun Grid Engine compute-farm software, and the Sun Blade 2000 workstation. Together, they mark Sun's effort to move from the realm of digital content creation (DCC) to mechanical design, displacing Silicon Graphics (SGI) along the way.
The XVR-4000 offers high performance for one-third the cost of solutions from SGI, says Peter Foulkes, group manager of Sun's Workstation Products Group. That magnitude of price drop changes the dynamic of collaboration, since it allows companies to buy tens of computers, instead of a few.
In turn, SGI (Mountain View, CA) stressed performance, announcing its InfiniteReality4 graphics system. Running on SGI's Onyx 3000 workstation, IR4 is designed to handle massive image complexity through sheer volume of pixels. It handles the load with distributed computing in its "Project X" visualization platform and the University of Utah's *Ray ("star-ray") software.