High Performance Computing Boosts Maya Performance

Have you ever thought that IT could be the key behind the life-like animation that captures eyeballs and keeps people hooked on to their seats? Yes, IT does have an important role to play in giving life to the animation, which so easily captures our attention. Let’s get to know the role of IT in the animation industry by taking a look at Maya Entertainment Ltd, one of the most reputed Indian animation studios.

A Graphics-Intensive Industry

Maya Entertainment is one of the most well-known names in Indian animation. The company has been around for more than a decade and has a host of credible projects to its name.

Jai Natarajan, EVP and business head, Maya Entertainment, said, “The company has been in business for ten years now. It was founded by filmmaker Ketan Mehta and has been involved in the business of animation both for the Indian and international markets.”

He further added, “The process of animation is an extremely demanding one, especially 3D animation. Maya has a very advanced IT infrastructure. Internally, all our users work on high-end workstations. They all use dedicated graphics cards for graphic-intensive applications. We have over 20 terabytes of centralised storage in the facility that is required to be accessed at very high speeds. Also, everybody is connected to a gigabyte ethernet network through some very powerful core switches. Apart from this, we have a server farm, which we call a render farm, that does all the offline computations in batches for the graphics work. Our back up systems are also very fast as they have to deal with such huge amounts of data. We generate almost 50-100 GB of data every week.”

Rendering Defined

“Rendering is an integral part of the animation process. An animated film is made of 24 frames per second and the eye is tricked into believing that it is a continuous stream of film. During the day, our artists work on the main two or three frames for each second of work, during which they refine the look and the animation and then we go through a process of rendering where all the work built up from the previous departments on a particular scene is computed for each and every frame that is required for that scene. This is a batch computing process, which is floating point intensive and results in the final set of images, which actually go into the making of the animation film,” says Natarajan.

High Performance Computing Speeds Up Rendering

Pipeline and rendering are the two most mission critical processes in animation. With pulsating loads mounting, Maya chose to go in for High Performance Computing (HPC) to speed up the process of rendering.

“The way that our rendering works, we have several thousand frames that are rendered every day and each frame consists of several lakh pixels, which are computed. Each scene consists of 8-15 layers, which are finally joined together. These layers are broken down initially to give more control over individual parts of the image. If you look at the overall amount of data, which is being transferred across our network, stored on our storage or broken up for computation among different servers, it is enormous. All of these are typical candidates for HPC solutions,” explains Natarajan.

Talking from a purely technical angle, Amit Srivastava, chairman and CEO, DUX Soft, who helped Maya audit HPC, says, “If you can organise CPUs into clusters and then simultaneously fire render jobs, it will increase your computational time and this is where HPC plays an important role.”

HPC and DUX Soft - The right combination for increasing efficiencies

With an animation studio as large as Maya, the rendering needs were huge. Let’s see how DUX Soft helped Maya use High Performance Computing for optimum rendering.

Talking about how HPC was customised for the Maya animation studio, Krishna Prasad, director and CTO, Dux Soft, says, “Maya Entertainment Limited works on multiple projects simultaneously. Due to this, they require high levels of inventory. If the inventory is continuously increased in terms of the infrastructure and the manpower for rendering, there will be no end to it. The best locations to optimise include the creative side and the rendering side. For rendering optimisation, a lot of servers have to be stacked up and this also reduces the cost of infrastructure”.

“What we did from our side is that we clubbed all the computers, which we call as render servers and created a cluster of them. With this, the performance of the cluster increased. The different levels of rendering include a regular scheduler where you fire a job using a simple software and a grid computing mechanism, which is a cluster component that enhances the performance of 50-100 computers put together. Grid computing gives higher performance than the regular scheduler,” he added.

HPC: The challenges and benefits

According to Natarajan, in a facility like Maya where projects are constantly being executed and the technology is constantly being upgraded, making changes is similar to laying tracks in front of a moving train as it comes along. He says, “At Maya, we had to be very careful with our proof-of-concept or what we call POC implementation. We first tried out the solutions on a small set of computers. Secondly, we worked with Microsoft’s office in Singapore. The engineer came down and assisted us with the testing of all applications on the new OS. We had to be convinced about the working of not only our applications but also the plug-ins and the in-built software that we were using. All of it had to work properly on WCCS. Once we were satisfied with the testing, we started rolling out the applications slowly. The performance boost post implementation was so good that people were demanding things faster and were willing to adopt the new technology quickly to realise its benefits. This is the way we confronted the challenge without de-stabilising existing projects.”

However, Maya has not yet completed the process of getting the entire advantage out of HPC. “Moving to a faster OS and choosing some of the components has already given us a 15 percent performance boost from the existing infrastructure without putting in any more money, which is a very big deal for a facility like ours, which has a huge capital expenditure every year. Beyond this, we will have to adopt the next important phase of HPC. One must remember that HPC is not just about the components; it’s also about the workflow, mapping the workflow and the hardware components. We have embarked upon the next phase with our HPC consultants and are in the process of auditing our existing infrastructure and finding out where the inefficiencies lie. We are also in the process of finding out how we can make our process workflow faster and take intuitive advantage of the facility that we have built.”

High Performance Computing is definitely the way ahead for the animation industry and it is all set to fuel crowd-enthralling animation content.