After spending countless hours tweaking the layering, colors, lighting, surfaces, shadows, shape and animation of 3D models, as well as every element in their electronic composition, 2D artists and 3D animators reach the moment of truth — rendering. Rendering is a complex, processor-intensive process whereby one set of data or pixels is converted into a new set of pixels, thereby producing the finished product — the added value the animation or VFX studio’s customers are paying for. In the case of 2D, rendering artists use raw materials, such as film scans or digital camera footage, and modify those elements and then render out the modified image. Typically, vendors of 2D visual effects and compositing packages, such as Adobe After Effects, Apple Shake and Autodesk Combustion, all offer their own renderer.
While 3D animation software also comes with its own unique renderer, it’s common for animators to turn to independent third-party products and use those for rendering instead. Because these vendors specialize in pushing the envelope on rendering, these software products promise to deliver the most stunning, spectacular and credible CG imagery from even the most complex 3D data.
RenderDrive 6400 by ARTVPS is a 64-bit dedicated network rendering solution. Developed using AMD Opteron processing power, along with the company’s raytracing processor, the AR350, the RenderDrive is a powerful, flexible, reliable stand-alone rendering solution. It can also be used as a renderfarm with multiple units.
RenderPipe is specialized raytracing software designed for use with the AR350 processors and is the only one that supports it. RenderPipe is supplied as part of the RenderDrive solution. ARTVPS technology currently supports Autodesk’s Maya, 3DS Max and Viz, as well as Catia V.5 from Dessault.
In mid-April ARTVPS announced the launch of its latest release, RenderPipe AV5.0, which can isolate rendering tasks as separate, standalone processes. This new version includes many new features designed to improve workflow and provide significant performance benefits for both 3DS Max and Maya users. For example, RenderPipe now allows scheduling and control of any number of large rendering jobs, without processes competing for limited resources. This eliminates the risk of failed jobs through computers locking up and makes it possible to produce very high-resolution 3D images.
“Most software solutions deliver a quick and acceptable solution for many rendering tasks using a technique called ‘scan-line’ rendering,” says Kate Marshall, head of marketing for ARTVPS (www.artvps.com) in Cambridge, England. “But, to produce photorealistic images that are indistinguishable from the real thing, a more accurate approach is required using raytracing.
“RenderDrive uses a specially designed hardware raytracer, so it produces images far faster than a software renderer with quality and realism far beyond anything achievable using standard raytracing solutions,” Marshall adds. “Advanced visual effects can be achieved simply with RenderPipe to capture the realism of a virtual scene. Also, we’ve added powerful user interface controls to make it simple to set-up complex lighting scenarios and use the sophisticated ‘real world’ camera features [offered by the 3D animation package], such as physically accurate depth of field and motion blur.”
Besides offering plug-ins that support the RenderMan shader language for materials, lights, cameras and atmospherics, ARTVPS has developed advanced, physically-accurate shaders for complex, shiny, glass or mirrored surfaces. “Our recent partnership [March ‘06] with Mental Images has meant significant developments to our rendering technology,” says Marshall.
“RenderDrives will now support the Mental Ray 64-bit standalone renderer with Mental Ray versions of the ARTVPS RenderPipe material library.”
[See the Mental Images section for more information on this partnership.]
“The way that one 3D animation package animates a 3D model is fundamentally different from another,” says Maurice Patel, head of Autodesk product marketing for Autodesk’s Media & Entertainment division (www.autodesk.com) in Montreal. “Renderers have to understand the 3D data they are being given, and they have to be tailored to work with that 3D animation software’s creative tools.”
As part of its acquisition of Alias, Autodesk recently added Maya to its product portfolio, which included 3DS Max, giving the company two of the leading 3D animation software packages in the industry. While both Maya and 3DS Max have their own renderers, Patel says that Autodesk works closely with third-party developers to ensure that the rendering software they develop tightly integrates with Maya and/or 3DS Max.
“The rendering capabilities within Maya and 3DS Max offer a basic level of reliability and functionality,” says Patel. “But our customers will most certainly use third-party solutions because they are designed and well-tuned for giving superior results. So people will gravitate to one renderer or another because they like the way it captures and represents their creative vision.”
Patel says there are four types of rendering: online, which ties up the processors on the artist’s workstation (sometimes bringing creative work to a halt); background rendering, which also uses the artist’s workstation so that it can render behind the scenes (and which can also slow things down); network rendering, which offloads the rendering onto a network of PCs so the creative work can continue unhampered; and inline rendering, which takes network rendering a step further, allowing processors to be grouped for realtime rendering performance.
As a companion to its 64-bit 2D systems, including Flame and Inferno visual effects compositing systems, and to Lustre for color grading, Autodesk offers Burn, a rendering solution that enables operators to “offload” their rendering onto a renderfarm or network of PCs dedicated to handling the rendering as a background process.
“With conventional ‘online rendering,’ when the artist hits ‘render,’ a ‘render bar’ moves across the screen. During this time, there is no opportunity to continue with creative tasks until rendering is complete,” says Patel. “With Burn, rather than having to go off for a coffee, creative work can proceed while rendering takes place on other PCs and the facility’s resources are used very efficiently.”
According to Jeff Campbell, visual effects supervisor at Toronto’s Spin, “Burn is great because when you’re working in HD or multiple layers, or both, it lets you render in the background, while allowing you to continue the work you’re doing. And for commercials, when it comes to multiple versions, it’s an incredible time saver.”
Spin contributed multilayered Maya 3D-animated visual effects for the Universal Studios release Land of the Dead.
“We’re focused on developing rendering plug-ins for leading animation packages that make the tools faster and easier to use while streamlining the production cycle,” says Michael McCarthy, president of Cebas USA in Bradford, MA.
The flagship product, FinalRender Stage-0, was released years ago as a plug-in renderer for 3DS Max. Since then, Cebas, headquartered in Germany, issued FinalRender Stage-1, a fully-integrated renderer for 3DS Max and FinalRender Stage-2 for Maxon’s Cinema 4D, a next-generation standalone render. Today, the company is issuing FinalRender Stage 1 SP3 (Service Pack 3), which is an extensive revision of the software and being issued to Stage-1 users for free. And soon the company will issue FinalRender Stage-2 for Maya.
“Our software never stagnates. We’re constantly updating it and rewriting the core software to add features and functionality and respond to our users’ changing needs,” says McCarthy. “We’re also in negotiations with Softimage to develop a version of FinalRender for Softimage|XSI, and hope eventually to support other popular 3D animation packages.”
Besides offering the basic features contained in competitive packages, Cebas goes further to add many advanced features for rendering both photorealistic and non-photorealistic animations. Among the features in FinalRender Stage-1 are advanced cartoon and illustration rendering features; intelligent and automatic raytracing acceleration; true geometry-based direct and indirect light sources; true Micro-Triangle Displacement; realtime volume light rendering; true caustic light rendering and fast area shadows.
“We also support render elements which give users the ability to output 20 different channels of elements, like alpha channels, shadows, reflections, global illumination, z-dept and more, and bring these into a compositing application for faster, more interactive manipulation,” says McCarthy.
With FinalRender, users can also have a single frame rendered by a “buckets of multiple CPUs” simultaneously, which greatly increases rendering speed. McCarthy says, “This is extremely advantageous when repeatedly modifying and testing the look of a rendered frame in a client supervised session.” FinalRender has been used on several movies, including Final Destination 1 and 2, Blade: Trinity and Harry Potter 3.
Chaos Group offers V-Ray, a fast, affordable rendering engine that can be seamlessly integrated within 3D applications, including 3DS Max and Maya, as a set of plug-ins. It’s also being offered as a stand-alone program.
According to Peter Mitev, software manager for Chaos Group (www.chaosgroup. com) based in Sofia, Bulgaria, “One of the biggest challenges to high-end rendering is the time constraint professionals face. The quality factor is directly linked to the amount of time the team can spend on finishing the job. Thus, having the fastest and most robust solution that best fits the need is the key to achieving best results.
“V-Ray has been designed and engineered from the very beginning with the thought that it is a high-end production renderer,” continues Mitev. “This involves developing algorithms that are faster; engineering technologies that work better; and designing the architecture in the most robust way so it can be expanded without hurting the current features.”
V-Ray’s features include a very fast raytracing engine; global illumination, as well as physically correct lights; fast 3D motion blur; caustic effects; and fast subsurface scattering texture map capability for quick, easy rendering of translucent objects, like a character’s skin or plastic materials. Also, render channels and a built-in visual frame buffer give users the ability to split and store image information into different channels and export them for post production.
“This can be a huge timesaver for fine tuning small things in post production applications,” says Mitev. “Moreover, we have our own raw image file format, where we store image data as floating point numbers, thus preserving the full color information, without causing banding and clipping of colors.”
Illuminate Labs offers Turtle 3 advanced rendering technology and sophisticated “baking” functionality for Maya — even involving very large datasets.
“The thing that differentiates Turtle from other renderers is our excellent baking support,” reports Nils Wirell, product manager for Illuminate Labs (www.illuminatelabs. com) in Gothenburg, Sweden. “This means that we treat baking the same way as standard rendering, which means that you can use adaptive super sampling, render passes, adaptive ambient occlusion, region renders and all the things you’re come to expect when rendering a standard image.
“This all boils down to an easier-to-use product that saves a lot of time for the artist and improves the control of the end result,” continues Wirell. “One of the biggest challenges involved with rendering is being able to handle the enormous amount of data that is thrown at the renderer. When rendering HD, you have a need for more detailed geometry and higher resolution textures. By treating geometry in smart ways and caching textures to disk, you can render with a low memory footprint in Turtle.”
Turtle is easy to understand and use. It supports Maya 6.5 and 7.0 on Windows, Linux and Mac. Recently, West Hollywood-based Engine Room used Turtle to render the new opening ID that runs before all Lions Gate Films. CG supervisor Andrew Honacker says, “We needed a renderer that could give us the quality we needed out of the box without having to waste days of tuning and problem solving. Because Illuminate Labs is a relatively fresh face, they were able to give us that level of attention that more senior render software companies cannot. The more we challenged them to improve an already incredible product, the better our work became and the more Turtle continues to become a production-proven render engine.”
Turtle 3 was also used for rendering by MFX, in Gothenburg, Sweden, in association with Percival Productions, for images in brochures for Volvo and Swedish sports car manufacturer Koenigsegg. The creative objective was to create images that would be indistinguishable from photographs. Turtle was chosen because its quality and speed were advantageous for these commercial campaigns’ tight deadlines. [See Illuminate Labs’ Web site for an online tutorial of the “baking” process, and how it benefits multilayered textured images.]
As part of a recent partnership, ARTVPS will become an OEM partner of Mental Images for the integration of Mental Ray in future 64-bit versions of ARTVPS’s RenderDrive product family. And ARTVPS will become a system integrator for RealityServer, a software platform for the development and deployment of 3D Web services and applications from Mental Images. In turn, Mental Images has agreed to license intellectual property from ARTVPS with the exclusive right to its sub-licensing.
“In cooperation with ARTVPS, we will explore the use of dedicated graphics processing hardware architectures for the acceleration of raytracing algorithms in Mental Ray and RealityServer without compromising on versatility and image quality,” says Rolf Herken, CEO/CTO of Mental Images (www. mentalimages.com) in Berlin, Germany.
Mental Images’ Academy Award-winning Mental Ray rendering software offers control over light scattering and the effects of indirect light in a 3D scene for greater photorealism and visual richness.
Headquartered in Madrid, Spain, with US offices in Washington, DC, Next Limit offers the Maxwell Render. Maxwell is a new render engine based on the physics of real light and the governing physical equations of light transport. Next Limit says that the algorithms and equations used by Maxwell reproduce the behavior of light in a completely-accurate way using very advanced lighting tools.
Maxwell is available as a stand-alone product for Windows, Mac OS X and Linux, with a set of plug-ins for most common 3D applications, including 3DS Max, Maya, Viz, Maxon Cinema 4D and NewTek LightWave. The plug-ins are freely interchangeable for every Maxwell license. Each license uses four CPUs simultaneously — so four single, two duals or a single quad processor.
All of the elements in Maxwell, such as light, material shaders and cameras, are based on physically-accurate models. According to Next Limit (www.nextlimit.com), Maxwell is “capable of producing incredibly realistic illumination results without resorting to illumination tricks that are common to other renderers for results that are accurate renditions of the real world.”
“In the rendering process, you’re synthesizing the 3D shapes that have been modeled, the motion that’s been animated and the cameras and lights,” says Dave Wilton, product manager for Gelato, at Santa Clara, CA’s Nvidia (www.nvidia.com). “You’re pulling assets from everywhere in the production pipeline and squeezing them down into one perfect image,”
While Nvidia Gelato 2.0 and Gelato Pro 2.0 are relative newcomers to the rendering market, Nvidia is well established as a hardware vendor of graphics accelerator boards. Gelato software is unique in that it is a GPU-accelerated final-frame renderer for use with any animation package. However, it ships with plug-ins for Maya and 3DS Max. Users and third parties can create plug-ins for other modeling packages using Gelato’s API.
Gelato offers fast rendering of complex scenes involving gigabytes of geometry and thousands of textures; fast, high-quality anti-aliasing and motion blur; NURBS, subdivision surfaces, polygons, particles, hair and fur; and raytracing (which represents light reflections and refractions) and global illumination (which simulates the physical process of light propagation).
There is Gelato Shading Language for surfaces, displacements, lights and volumetric effects. And, Gelato Pro includes Sorbetto, an interactive re-lighting tool that allows users to change any lighting parameter and re-render the frame in interactive time.
“Through a combination of fast re-rendering and workflow tools, Sorbetto allows the lighter to quickly re-render the image and make adjustments to the actual image, not a proxy that may not maintain fidelity with the final render,” says Larry Gritz, chief architect of Gelato. “Sorbetto uses the same geometry, shaders, procedurals and hardware as a regular Gelato render.”
When looking to buy a rendering package, Wilton says, “A renderer must be able to render extremely large scenes relatively quickly. It’s important to put a prospective renderer through its paces by rendering production scenes that involve the level of complexity and production techniques you use most often.”
Support for 64-bit Linux is now available and Gritz says they are working on 64-bit support for Windows. The company is also close to releasing the beta version of Gelato 2.1, which will support “texture baking” in Maya and will significantly speed up raytracing. (Baking is the pre-calculation of lighting.)
Nvidia recently decided to make Gelato 2.0 available for free to users, including students, freelancers, non-profit organizations, and others that cannot readily afford this caliber of rendering software. The Gelato Pro version, which adds scalability and productivity features for larger studios and a support package, is for sale. “Our intention is to see as many people as possible using Gelato to make beautiful images,” says Wilton. “By getting these users accustomed to working with Gelato, we’re hoping to establish our GPU as the tool for high-quality, non-realtime rendering.”
According to Pixar, its RenderMan product has been used on 30 films nominated for an Academy Award in the last 10 years. They also report that RenderMan has been a part of a number of key milestones in film rendering and animation history, including The Abyss, Toy Story and Jurassic Park.
“RenderMan empowers artists to push the envelope on visual imagery by being a highly configurable solution offering a huge breadth of lighting, shading and effects tools that can be combined in any way,” says Chris Ford, business director for RenderMan at Pixar Animation Studios (www.renderman. pixar.com) in Emeryville, CA.
“There is no specific RenderMan ‘look.’ Nemo doesn’t look like King Kong! And the user has the ability to make it do anything,” says Ford. “RenderMan has been tried and tested in so many production scenarios that it is very predictable. Though not a ‘feature,’ confidence in RenderMan’s ability to deliver on time and schedule is also high on the list of key capabilities.”
RenderMan, which is both an industry specification/format as well as a family of products, is now available in two versions: RenderMan for Maya, which is a new, easy-to-use plug-in for Maya; and RenderMan Pro Server and RenderMan Artists Tools (the classic RenderMan).
Designed for rendering animation and visual effects at film resolution, RenderMan features include micro polygon rendering, which simplifies the rendering of smooth curved surfaces; memory efficiency for rendering vast amounts of geometry; deep shadows, which add realism to fur, particles, and more; and displacements, providing faster creation of modeled detail through textures.
“RenderMan is one of the very few renderers that combines all the standard rendering techniques, such as scanline, global illumination, raytracing and particles, in one single render pass, so that everything can be seen within the final context and in-camera,” Ford adds. “RenderMan is written in the environment of a working production studio so we understand the demands of production as well as the technology.”
“Artist time is far more expensive than rendering software in a studio, and it is critical to the decision about rendering software,” says Scott Kirvan, CEO/senior software designer for Seattle's SplutterFish (www.splutterfish.com).
SplutterFish’s Brazil r/s rendering software promises reliability, scalability and performance, and numerous artist-inspired extras. It has a platform independent, plug-in-based rendering architecture that enables it to take advantage of new technologies as they develop.
Explains Kirvan, “If the artist is struggling with buggy, crashy tools, has to constantly re-render shots, can’t predictably work toward the image they want, is continually wrestling with a workflow that is too complicated, or hits unnecessary limits set arbitrarily by software programmers, then that tool, in production, becomes too expensive. [At that point] it becomes a production blocker and impacts the final results, regardless of how many features the renderer boasts.”
The Orphanage, a visual effects and production company in San Francisco, used Brazil r/s V.2 with 3DS Max for The Great Race, a Ruby Tuesday commercial that debuted during the Super Bowl.
“For The Great Race, we needed to create completely photoreal Mini Coopers and seamlessly integrate them into live action photography,” says Stu Maschwitz, director/VFX supervisor at The Orphanage. “It was critical that we had perfect motion blur with that curvilinear, sub-frame detail that makes silver wheels and streaking sheet metal look like the real thing.”
Kirvan notes, “In the end, all the client cares about is the final product…the image. If the renderer can’t deliver for the client, you could lose that job or the client.”