For our VFX team, our architecture on AWS acts as a safety net, so they can put more resources toward rendering as big deadlines approach. That means we improve overall artistic quality while keeping clients happy.
Chris Spry Operations Manager and VFX Supervisor

Fin Design + Effects is a boutique design and visual effects (VFX) company based in Sydney, Australia, and Shanghai, China. Established in 2001, Fin creates multimedia advertisements and other design projects for such well-known global brands as General Electric, Coca-Cola, and Jim Beam. Recently, the company has branched out into visual effects for feature films and has just completed more than 200 shots on Alex Proyas’s film Gods of Egypt for Lionsgate. Because the company spans cultures and time zones, its 30 VFX artists, designers, and producers must work flexibly to meet project demands and collaborate with other artists throughout the world.

Fin artists frequently find themselves working on projects that require them to perform unexpectedly complex tasks using rendering software. As a result, these artists often need additional compute resources to complete those projects on time. “Our need for compute resources fluctuates all the time, based on the project,” says Chris Spry, operations manager and VFX supervisor at Fin Design + Effects. “Our rendering work requires a lot of processing, and there are a lot of compute peaks and troughs throughout a project’s lifecycle. The best-quality work comes from being able to see as many iterations as possible before a deadline, and each iteration requires computation.”

Traditionally, Fin has dealt with that issue by working with outside vendors to acquire more servers. For example, because the company’s on-premises render servers have a finite capacity, Fin sent specific projects to outside rendering farm services. “We have always expanded our render farm using external sources and tools, but that gets expensive and it’s not always fast enough for our needs,” Spry says. “We needed more agility, so we could access large amounts of compute capacity when big deadlines come around.” Fin also wanted to reduce the costs of relying on those outside resources. “We were never able to make sure our local rendering farm didn’t go idle,” he says. “Internally, our on-premises farm might run out of jobs to process at 2:00 A.M., while we’d still be paying for the external service to continue rendering for the rest of the night.”

To meet its needs, Fin decided to take advantage of cloud technology. However, the organization needed to operate in a hybrid environment to take advantage of its on-premises hardware investments and centralized data storage in its Sydney headquarters, where most of its artists work. “We wanted the flexibility to schedule compute tasks both on-premises and in the cloud,” Spry says.

The company also needed to ensure that the cloud solution could deliver fast performance for its artists, as well as comply with the security requirements of the Motion Picture Association of America (MPAA). “As we began working on more global film projects, that became a huge concern,” Spry says.

Fin evaluated several cloud technology providers before choosing Amazon Web Services (AWS). “We wanted to save money and gain more agility in how we added compute resources, and AWS looked like it could enable us to do both of those things quite easily,” says Spry. Fin started by using AWS Direct Connect to establish a dedicated Gigabit network connection from the Fin data center to the AWS Sydney region. “Using AWS Direct Connect allows us to treat AWS as part of our own data center, so the virtual machines we spin up act as if they’re on our own network.” Fin then started using Amazon Elastic Compute Cloud (Amazon EC2) on-demand instances, while continuing to use its on-premises servers. The organization runs shorter and less computationally intensive render frames on premises, and it pushes the longer, more complex frames to AWS.

Fin is also taking advantage of Amazon EC2 Spot instances, which are EC2 instances that allow users to bid on spare Amazon EC2 computing capacity. Spot instances run when a bid exceeds the Spot price, and they are often available at lower prices than traditional EC2 instances, so Fin uses them to cost-effectively grow its compute capacity. “We have an internal web interface that displays the EC2 Spot pricing in live graphs for different machine types and Availability Zones,” says Spry. “We then have the flexibility of adding a certain number of Spot instances to a resource pool as well as a group of different types of instances to the same pool.”

Additionally, Fin created scripts that enable EC2 instances to automatically scale down when they’re not being used in the cluster. “We use a countdown timer, so when a selected time comes and those machines aren’t doing any work, they are automatically shut down,” says Spry.

Fin now has the agility to add compute resources on the fly to meet last-minute project demands. “We can easily double or triple our compute capacity in just a few minutes,” says Spry. “That means our artists always have the capacity they need to meet big deadlines and handle all the changes that clients make to projects we’re working on.” Overall, the artists at Fin have now a better ability to meet critical project deadlines. “For our VFX team, our architecture on AWS acts as a safety net, so they can put more resources toward rendering as big deadlines approach. That means we improve overall artistic quality while keeping clients happy. We don’t have to make excuses about why their jobs aren’t going to air when they’re supposed to.”

Fin is also seeing significant savings by using Amazon EC2 Spot instances. “For us, using the cloud is definitely about cost savings,” says Spry. “We are reducing our operational costs by 50 percent by using Amazon EC2 Spot instances. We get additional compute capacity at the cheapest prices, which means we can purchase more nodes. Overall, it enables us to be more cost-competitive for our clients, while giving our artists more creative potential.”

The company now has the flexibility to schedule both on-premises servers and EC2 Spot instances through its render farm manager. “We can schedule rendering jobs to run overnight, and if something doesn’t get through on our internal network, we can put more EC2 instances on it in the morning to clear the backlog,” says Spry. Fin also has better performance on its internal network because of AWS Direct Connect. “Having a fast network link from the cloud to our data center through AWS Direct Connect makes a huge performance difference. And because Direct Connect does the heavy lifting, we didn’t have to develop our own syncing tools and possibly introduce lag time in our rendering.”

Additionally, Fin is confident in the security of its new solution. “AWS is MPAA-compliant, which is extremely important to us,” says Spry. “We also gain additional security because our instances are on a private network through AWS Direct Connect. Overall, we’ve very impressed with the security in AWS.”

In the near future, Fin plans to use additional AWS services for disaster recovery and other functions. “We will be doing a lot more in AWS going forward,” says Spry. “We’re definitely buying into the cloud. It’s the right model for our business.”

To learn more about how AWS can help you with your development and test environments, visit the EC2 Instance types page.