Getting What You See, To See What You Get.
Rendering, yes, rendering, is the hot topic of 2009. It’s possible you have been following CAD and engineering long enough to remember that rendering programs were offered as an essential add-on more than 20 years ago. You probably also know that the rendering capabilities within most CAD programs have improved considerably. So, why is rendering a hot-button issue now? Again?
In a way, the reason is the same now as it was then: because you can. Rendering is the first love of many graphics doctoral students, and every year, universities crank out new, brilliant PhDs with new, brilliant approaches to rendering. In the past five years, the emphasis has been on shader technology, which the OpenGL and Microsoft DirectX APIs made available to hardware processors. In this new golden age, rendering is jet-fueled with hardware acceleration and rendering algorithms that are better than ever. The broad availability of 64-bit processors with 64-bit operating systems means that ever-cheaper systems can handle ever-higher amounts of data.
Raytracing, the process of calculating the path of light to create photoreal effects, has re-emerged as a practical application for computer graphics now that powerful, new multiprocessor systems have arrived to go to work on the problem. Also, those talented doctoral candidates have devised clever compression algorithms. Given all that, it’s not surprising that there are several contenders jockeying for position in the rendering sweepstakes, and there are plenty of interested, materially interested, onlookers.
Rendering is new again, thanks to significant advances by a number of vendors. Here, David Burgess illustrates the power of Bunkspeed’s HyperShot for this Ford Taurus image on behalf of Ford Motor Co.
The Players
Among the companies that are lining up excitedly to gain approval in the CAD world are Bunkspeed, Onesia, StudioGPU, ArtVPS, Autodesk (with Showcase), Realtime Technology (RTT), and Luxology. Longtime major-league player Mental Images is redoubling its efforts in the CAD market, and even Pixar has revamped the venerable RenderMan—a product that’s been doing its job for more than 20 years in the movie industry—to take advantage of multiple processors and 64-bit technology.
What’s interesting is that companies with rendering technology are taking different approaches, and they’re looking at different points in the workflow. In fact, many of these rendering companies are hoping to insert themselves into points in the workflow where they have never been before.
Luxology has worked directly with SolidWorks (a subsidiary of Dassault) and Bentley to create modules that work within the SolidWorks and MicroStation products, respectively, allowing engineers, architects, and designers to see and communicate design ideas with the push of a button. Luxology comes from the entertainment world. The use of its products have grown quickly in a few short years because the product is attractively priced, it’s powerful, and the Luxology development team has attracted a loyal following over the years. (The company’s founders, Stuart Ferguson, Allen Hastings, and Brad Peebler, were also developers of NewTek’s LightWave.)
Peebler, Luxology’s CEO, reports success as he branches out into new fields. The Luxology rendering modules are included for free in the professional versions of MicroStation and Luxology, and Peebler reports hearing that CAD users are giving these tools a try because they’re available, they’re fast, and they’re easy to use.
Peebler says he’s been told repeatedly that the ability to create a quick render with materials and lights is something people are trying out because it’s there, and as a result of being able to render quickly, they’re better able to sell an idea or identify a problem. The point is that these are people who wouldn’t normally render a model; it’s just not in their workflow.
Bunkspeed was among the first to offer the world a fast, relatively low-cost renderer with its HyperShot raytracing visualization tool. Like Luxology, Bunkspeed also reports fast growth over the few years it has offered HyperShot, and at last year’s SIGGRAPH, the company introduced HyperMove, an animation tool that adds physics and lets users add realistic motion to a 3D scene. And of course, Autodesk offers Showcase, for preparing, processing, and presenting 3D CAD data. More recently, Autodesk has beefed up the product with raytracing technology from Swedish company Opticore following the purchase of the company’s assets two years ago.
We’re seeing the rise of products that enable users to quickly put a model in a scene, adjust a few lights, play with materials and colors a bit, and, voila, a photorealistic scene is born. Germany’s RTT has a variety of products to do this at various points in the pipeline. Also, Nvidia last year acquired Utah raytracing start-up RayScale for its Mental Images division, and is also building tools for fast-rendered scenes.
There are new customers for rendering—or, at least, that’s what all these firms hope. The idea is that there might be users out in the design pipeline. A product has been designed, and the proud designers want to show off what they’ve got. They are not the ones who will create the TV ads or the full-spread magazine ads, but they can communicate their latest ideas to the full team and maybe to marketing to get everybody onboard.
To this end, Autodesk has just released Showcase 2010 in three versions: Viewer, Presenter, and Showcase, and all three now have raytracing and integrated global illumination. In this instance, Showcase takes advantage of the computers’ CPU—all of them. The more cores, the better. Autodesk has expanded the materials capability in Showcase with a unified and calibrated materials library, including hundreds of texture maps and fully editable materials so users can create their own and share them.
Rendering newcomer Caustic Graphics has developed an API that takes advantage of a workstation’s CPU or GPU to combine raytracing and rasterization technologies.
Two Parts of the Brain
For the longest time, there has been a divide between the entertainment world and the CAD world, even though the technologies are very similar. All the work CAD users do goes into the creation of a digital version of something that is going to be manufactured or built. In general, it matters less if it is beautiful, but it must be accurate—lives, careers, and dollars depend on it. The entertainment world, of course, is all about appearances. It must live, but it doesn’t need to be real.
That divide is gradually getting filled in, and the tools used to create fantasy are also coming into play in the all-too-real world of CAD.
Consider Luxology, for example. The company’s roots are deep in the broadcast industry, but its sights are fixed on the broader CAD industry. As Peebler notes, “There’s hundreds of thousands of customers in the entertainment business; there’s millions in CAD.”
The low-cost Brazil renderer (formerly from Splutterfish) was a tool used in conjunction with Robert McNeel & Associates’ Rhino. Raytracing upstart Caustic Graphics recently acquired Splutterfish as a tool to reach out to new markets in CAD as well as entertainment. Another new entrant is StudioGPU, a company that has just arrived on the scene with enthusiasm and some crazy ideas that just might work.
StudioGPU has experience in game development, film, and video. Brothers David and Yanni Koenig and Robert Knaack decided to create better rendering tools after too long living the nightmare as creatives in entertainment and advertising. As Yanni Koenig puts it, the artist is still subject to a priesthood that dictates when they can really see their work. Modelers, animators, and artists have to pass their work off to be rendered, all the while waiting hours for a fast render or overnight for a complete view. Every change, every experiment has to wait for the rendering process, which happens elsewhere. Jobs go to the farm, and they’re delivered back.
The StudioGPU renderer is fast, really fast. It takes advantage of GPUs and shader technology, and can make changes at the push of a button. It’s also clever technology, taking advantage of rendering tricks to create reflections that seem as if they’ve been raytraced, but the technology is all rasterization. In the future, say the Koenigs, they’ll add raytracing technology, but the system they have now is capable of professional rendering that can free the creative process from the drudgery of the render cycle.
In their earlier incarnation in 3D design and development, the StudioGPU team worked on games, movies, advertising campaigns, and architectural renderings. They’re pushing MachStudio Pro first as a tool for pre-visualization, for small productions and advertisements, and for visualizations. For example, in the case of creating pre-visualizations for products, the group thinks it can offer a rendering tool that can help determine the look of the production throughout the process, instead of at the very end.
And, the StudioGPU guys are well aware that they have their work cut out for them when it comes to changing the way productions are organized.
Unfortunately, no matter how great a renderer is, it has to make sense for the user. In every great wave of enthusiasm for rendering within the technical and engineering fields, there’s been a corresponding tsunami of disappointment for the people who have put their heart and soul in these products only to find that the people they hoped would use them, don’t.
Rendering is a workflow issue. Who does it, and when? Come to think of it, another good question would be, Why?
For the most part, rendering has remained the territory of artists and placed outside the ordinary workflow of most CAD job descriptions. So the answer seems to be simple, everyone in the design workflow has a job to do and usually not enough time to do it. For most of these people, creating pretty pictures is a luxury, not a requirement. The time for a rendering traditionally has been early, in the original design phase to visualize and sell ideas, and then again as a product is being readied for production and sale as advertising and marketing materials are being prepared.
However, the new wave of renderers hope to offer products that are fast enough and easy enough to change the way people work and to make rendered images and animations a tool throughout the ideation and design processes.
Luxology has expanded its presence in the CAD industry, offering modules that work with SolidWorks and MicroStation, enabling engineers and the like to create images such as this one by Wiek Luijken.
Stakeholders: The Hardware
Selling 3D modeling and animation software is a lot about love and a little less about money. The developers of visualization software, maybe especially rendering software, are all enabling creativity. On the other hand, rendering software has the potential to sell a lot of processors, and that’s why companies like AMD, HP, Intel, and Nvidia are supporting and pushing the development of graphics software.
Nvidia has been a serial investor in rendering. The company bought Exluna, maker of Entropy, to develop its Gelato rendering tools. It bought RayScale for its fast raytracing technology, and it bought the very well established Mental Images to take it deeper into the process of rendering for entertainment and design visualization. These moves have paid off for the company. Nvidia’s expertise and development in software graphics has enabled it to develop powerful professional graphics systems. The company’s Quadro line has greatly benefitted from the relationships, and Nvidia is determined to keep leveraging its work in graphics software to sell hardware systems.
Nvidia’s Mental Images has worked closely with companies to integrate its rendering software into Autodesk’s 3ds Max, Maya, and Softimage, as well as other companies’ 3D content creation software. Mental Images’ shader creation technology, Mental Mill, now is an integrated tool within 3ds Max 2010. It lets designers create custom shaders with adjustable parameters that are available to artists as they work. Complex effects can be created by combining shaders into phenomena—and, the Mental Mill promise is that these tools can be created without the user having to program. Mental Images has developed an overall format for its shader technology, called MetaSL. The firm is hoping that as companies adopt Mental Images’ technology, they’ll take advantage of the MetaSL format to share shaders and phenomena across product platforms.
Smart Design used Bunkspeed’s HyperShot raytracing visualization tool to illustrate this unique-looking juicer model.
AMD’s ATI had a leading position in the market with its FireGL graphics products, but let its position in the professional graphics market erode. For too long, the company failed to recognize the shift in the professional market that was once again tying hardware and software together. In the past two or three years, however, AMD has gotten control of the reins, and it is taking a much more active role in working with software developers.
Most recently, AMD has announced a strategic relationship with StudioGPU. There’s money involved, but neither side is saying how much. As a result of the relationship, StudioGPU’s MachStudio Pro visualization and rendering software will be sold as part of a system with ATI FireGL V8650 graphics boards. The two companies will work together to optimize the rendering software and graphics hardware to further improve StudioGPU’s already amazing render times.
Intel is also enthusiastic about the ability of rendering to sell processors—lots of them. In the case of Intel, of course, this already has occurred. For the most part, rendering has been done on CPUs, and as the company readies its Larrabee processor—a multi-multicore CPU—Intel sees no reason why rendering can’t sell a whole bunch more processors. Intel is avidly poking around companies with rendering software and taking a keen interest in the careers of those doctoral candidates mentioned earlier. It, too, sees opportunity in moving the rendering phase up earlier into the production process and in allowing it to happen more often. Unquestionably, a Larrabee processor with anywhere from 16 to 64 processing cores (the actual number, as of this writing, is unannounced) is going to change the equation for rendering on the desktop.
A creation of Mark Sullivan, this character was rendered using ArtVPS’s interactive raytracer, Shaderlight.
And, playing out in left field is another interesting contender, Caustic Graphics. Caustic’s CTO James McCombe has identified the computer’s basic hardware as a major culprit in creating bottlenecks for rendering, specifically raytracing. Caustic Graphics has developed an API—CausticGL based on OpenGL and a tiny bit of hardware IP (a raytracing controller, if you will)—that can accelerate the feed to memory and take advantage of the processors on a workstation graphics card. Obviously, Caustic Graphics sees the advantage in linking hardware and software, but McCombe isn’t religious about these things. He says his technology will work with the GPU or the CPU, and systems can be developed to combine raytracing and rasterization technologies. It’s all good.
Caustic Graphics acquired Splutterfish, developers of the well-regarded Brazil renderer, in order to understand how their customers might use rendering. It also gives them the ability to dig deep into the entrails of a software renderer and gives the Caustic developers a better understanding of how rendering software works and where the opportunities are for hardware optimization—pretty much the same reason Nvidia buys rendering technology.
The company continues to sell the Brazil renderer, but Caustic is primarily interested in forging partnerships with all the companies providing rendering. The company is maintaining that its hardware/software combination can speed rendering by as much as 20 times the speeds currently achievable. As a matter of fact, the company is a little concerned that it will have to convince potential partners that Caustic will not be a competitor with Brazil.
Caustic Graphics promises to reduce the time it takes to create raytraced images, such as this game controller.
This time, things really are different. There’s not much doubt about that. The ability to create high-quality renders faster is going to change the workflow in every business where computer graphics are involved. The ability to see what the finished product might look like as early as possible shortens time to market. The question for these companies is, when? Sometimes changing the model is easy; it’s changing the person that is hard. In order to succeed, each one of these companies is going to have to identify the people they want using their software. They’re going to have to understand exactly what these people are doing for a living, what pressures they’re working under, and how, specifically, they can take advantage of rendering without slowing down what they already do.
The potential, though, is enormous: It’s the Holy Grail, again. What will happen—and it will happen, because of the work being done by these companies and a whole generation of technologists and scientists—is that artists will be able to see what they’re imagining as they’re working. That opens the door to more artists and more content, more designers, and better designs. And, when the world changes, it’s going to look a lot better.
Kathleen Maher is a contributing editor to CGW, a senior analyst at Jon Peddie Research, a Tiburon, California-based consultancy specializing in graphics and multimedia, and editor in chief of JPR’s “TechWatch.” She can be reached at
Kathleen@jonpeddie.com.