Posted By John Rath On July 11, 2011 @ 8:55 am In HPC | No Comments
The visual effects in summer blockbuster sequels Kung Fu Panda 2 and Cars 2 were generated with powerful workstations and render farms. Here’s a closer look at some of the technology driving those projects.
Kung Fu 2 Panda Powered by HP
Long time partners HP and DreamWorks Animation united  once again to create Kung Fu Panda 2. DreamWorks artists used powerful HP Z800 workstations when designing whirling water and lush green lands, as well as battle scenes filled with thousands of animated martial artists. “The creative ambition for ‘Kung Fu Panda 2′ again pushed DreamWorks Animation to the cutting edge of technology,” said Ed Leonard, chief technology officer, DreamWorks Animation. “The film’s lead characters are some of the most intricate ever created and the environments are among the richest and most vibrant in CG filmmaking. HP technology continues to help us cross these new boundaries, develop new standards for the filmmaking industry and ultimately create enjoyable films for audiences around the world.” The original “Kung Fu Panda” film, released in 2008, used more than 50 terabytes of data and required nearly 25 million render hours. With the switch to stereoscopic 3-D and the ever-escalating creative aspirations of the filmmakers, “Kung Fu Panda 2″ used 100 terabytes of data and required more than 55 million render hours. The final battle sequence in the film took more than 7 million render hours to produce 14,000 frames.
Cars 2 Powered by Dell
Pixar introduced  some new technological breakthroughs in “Cars 2,” which required the company to triple the size and scale of its legendary render farm. New visual approaches to creating choppy ocean water and car reflections were met by technical teams with a powerful render farm located in the main Pixar  headquarters building. The compute power required to for “Cars 2″ averaged 11.5 hours to render each frame. Some of the more complex sequences that involved ray tracing took as much as 80 or 90 hours per frame to render. The expanded render farm features 12,500 cores on Dell render blades and Pixar also ‘boosted’ its file servers and the network backbone.
Whiskytree selects Avere and Arista
Visual effects provider Whiskytree  has implemented  a clustered solution of Avere  FXT 2300 appliances to scale its rendering performance by 700 percent, while reliably consolidating SAN and NAS data into a single storage pool and simplifying management. The storage fabric selected for the render farm is based on Arista Networks 10 GbE switches and Arista  EOS switch software that provides new features for provisioning, scaling and managing networks. Whiskytree offers visual effects, post-production, concept art and design services and has worked on such films as Thor, Tron:Legacy and Terminator Salvation. With a growing render farm and the need to scale data fast and easily, Whiskytree sought out a solution that could support its existing Xsan environment and allow for a seamless transition from SAN to NAS. “On the networking front, we needed to increase performance to optimize the render farm’s utilization. The Arista installation was straightforward and the performance exceeded our expectations,” said Jonathan Harb, CEO of Whiskytree. “We’re now looking to extend 10GbE to our production workstations to further improve artist productivity while using standard twisted pair cabling.”
Article printed from Data Center Knowledge: http://www.datacenterknowledge.com
URL to article: http://www.datacenterknowledge.com/archives/2011/07/11/turbo-charging-digital-render-farms/
URLs in this post:
 united: http://www.hp.com/hpinfo/newsroom/press/2011/110526xc.html
 introduced: http://news.cnet.com/8301-13772_3-20068109-52/new-technology-revs-up-pixars-cars-2/
 Pixar: http://www.pixar.com/
 Whiskytree: http://www.whiskytree.com/
 implemented: http://www.averesystems.com/News_PressReleases.aspx?ID=33
 Avere: http://www.averesystems.com/
 Arista: http://www.aristanetworks.com/
 John Rath: http://www.datacenterknowledge.com/archives/author/johnr/
Copyright © 2012 Data Center Knowledge. All rights reserved.