Outlook 2018: Capitalizing on render farm elasticity
Chris Bond
Issue: January 1, 2018

Outlook 2018: Capitalizing on render farm elasticity

Chris Bond is the Founder of Thinkbox Software (www.thinkboxsoftware.com) and Director of Product Management, Amazon EC2 at AWS. Here, he looks at the power, flexibility and cost savings that cloud rendering can offer.

Rendering is a crucial component of content creation workflows across VFX, animation, design, and AEC, among other industries. How artists leverage compute is important to a studio’s ability to deliver work efficiently and at scale. Having an elastic infrastructure provides the flexibility to dictate budget and turnaround time on a granular level, and then snap back down to zero when a job is complete. 

Most studios have an on-premises render farm, which was designed to absorb capacity when it was purchased. However, with the desire for more dynamic and complex imagery, combined with content standards changing from HD to UHD and higher, as well as constantly shifting and accelerating delivery timelines, this investment may no longer be enough. When I started a facility in the ‘90s, the on-going issues of local resources to house the render farm — space, cooling, electricity, etc. — were accepted as coming with the territory. If I was building a studio today, I would go all-in on cloud because of the flexibility, agility and control the cloud provides me. My dream as an artist was to have the power of a mega-studio at my fingertips, and Amazon Web Services (AWS) brings that scale to anyone right now.  

VFX for The Tick were rendered using elastic resources.

Simply having the ability to quickly deploy hundreds or thousands of instances in minutes around the globe changes the way you can approach the work. When limited to physical resources, artists tend to submit render-intensive jobs at the end of the day. The very nature of these types of jobs — the complexity, resolution and scale — means they should get more iterations, not less. Moving to the cloud and the advantage of having nearly unlimited scale means artists can iterate more quickly and don't have to wait overnight to get render results. With an elastic farm, there’s no reason a studio should be stuck with thousands of a job in the queue due to limited infrastructure. Jobs can be done in parallel and turned around as quickly as desired. 

When looking at data from some of our customers, it becomes clear that there are frequent times in a project where artists can be waiting multiple hours or even days to see the result of their efforts. This pain point shouldn’t be underestimated. Not only can more iterations improve the art, but scaling to the AWS Cloud can also reduce the cost. For example, one of our customers would have had to purchase over 12,000 machines to ensure that artists had their renders within a few hours rather than days. Not only is this a massive investment in capital outlay, space, cooling and power, but they would also be idle most of the time and only used for peak production periods. Instead, using AWS EC2 Spot allows this customer to render through these problems for less than what one of those machines would cost per day. This improves the quality of the work, improves the artist experience and helps deliver a better product to their customer.