Just look at the Coca-Cola ad made with ChatGPT-4 and its image generation platform, DALL-E and how many Super Bowl spots turned the spotlight on AI.
And while studios have no shortage of creativity and ideas for these projects and technologies, it turns out, many do have a shortage in storage capacity.
How Generative AI Platforms Are Accelerating Creativity On-screen
Maybe you’ve heard YouTube sensation “There I Ruined It,” who uses AI to create song mashups and vocals of long-lost greats such as Elvis and Hank Williams covering modern classics. Or, maybe you saw Everything, Everywhere All At Once, which used Runway AI Magic editing tools for its moving rocks scene, among others.
They’re all indications of where we’re headed, but take note: These ideas start with human creativity, not the other way around. The key to generative AI (GenAI) in these use cases is supercharging human creativity and postproduction output, not replacing it.
“The studios do a lot of the postproduction work in films — particularly in animation. The postproduction companies have a software development culture, so they will embrace generative AI.” – Yves Berquist, to MIT Sloan Review
For production and animation, in particular, AI’s powers are nothing new. Video game studios and post-production companies have used AI to auto-fill backgrounds with realistic graphics for years. But GenAI and large language models (LLMs) will only make it easier to do it well and at scale. Sora, a text-to-video model, is the latest evolution, promising to make time-consuming tasks that much easier for animators, designers, and editors. Just enter a prompt and Sora can generate video up to a minute in length, with multiple shots within the same scene.
For those studios and companies not already invested in the tech to support this, however, there could be growing pains. A simple, modern data platform is the best way to walk into it—without having to retrace their steps later.
What Does It Take to Execute Creative Projects of This Scope?
The answer: immense GPU processing power, modern data storage, backup, and security capabilities. Creating large-scale digital film projects can take an army of artists and specialists, each working at a high-powered workstation, simultaneously generating massive amounts of data.
Cutting-edge special effects play a starring role in recent films like Godzilla: Minus One and Mission: Impossible—Dead Reckoning Part One. Last year, Avatar claimed landmark accomplishments such as mastering how to successfully digitally animate water and its use of an ultra-high frame rate.¹ It’s also one of the most data-heavy films produced. (For reference, the first Avatar reportedly required a petabyte of storage. The sequel blew this out of the water, pun intended, requiring 18.5 petabytes of storage.)
“Storage capacity as tech demands grow is one of the most common problems in the industry.” –Nicky Ladas, Chief Technologist, Cumulus VFX
For the second Avatar, underwater performance-capture data was collected daily and sent back to VFX artists, who used files to bring characters and scenes to life. The studio needed the capacity to seamlessly receive and process data to create around 1,600 simulations.
It’s a challenge Cumulus Visual Effects (VFX) knows well. Based in New South Wales, Cumulus is one of Australia’s premier boutique VFX providers. The company is a trusted collaboration partner for larger VFX studios, nationally and internationally. According to Nicky Ladas, Chief Technologist at Cumulus VFX, most of the organization’s IT challenges came down to storage capacity.
“There was no overhead left on our old storage subsystem,” explains Ladas. “We had hit the wall from a capacity and performance point of view.”
What’s in a Modern Digital Film Studio’s Tech Stack?
To support all this innovation, creativity, and data, serious computing power is needed. Cumulus VFX’s Ladas says that for effects studios, “Storage capacity as tech demands grow is one of the most common problems in the industry.” That’s because, to produce on time and at scale, studios rely on:
- Powerful computing cores—multiple racks with chassis containing numerous machines, adding up to thousands of processors and hundreds of terabytes of RAM. Graphics processing units (GPUs) like NVIDIA, in particular, are in high demand.
- Fast-fiber network attached storage (NAS) and multi-gigabit networking—with local flash storage to support visual effects workloads’ high-bandwidth requirements. (Around 100 terabytes can be in play in a production environment.)
- Transparent, easy-to-provision, fast object storage—storing complex, object-based files for rendering systems. Provisioning those data storage resources in real time has to be simple and transparent for users.
- Fast, robust backup architectures—able to back up and archive massive snapshots in a batch time window before the next day’s work begins.
Tech demands aren’t just intense; deadlines are, too. When one scene can take weeks to simulate and millions of processing hours to render, time is of the essence. Lagging tech can mean productivity headaches for hundreds of artists working in parallel.
A Data Center Movie Dreams Are Made Of
To meet the enterprise-level SLAs of global VFX studio clients, Cumulus needed a future-proof storage environment for growing data demands.
“We’re already talking about doubling simulations,” says Ladas. “Pure is solid. They’ve got the redundancy and performance that we need to free up storage bottlenecks and better support global clients.” This allowed the team to grow from 20 to 60 rendering artists, enabling more processing of complex VFX simulations and six-nines of availability during massive simulations.
In addition, Cumulus’ new, green Pure data center supports their commitment to ESG practices, minimizing environmental impact and reducing the physical footprint through higher density and performance.
Bringing Movies to the Silver Screen Requires Data, Too
Once upon a time, projectionists hoisted a film reel to play a film. Today’s theaters use software to play movies and advertisements and at-home streaming services require massive throughput to serve millions of concurrent users binging the latest content.
For digital cinema company Arts Alliance Media, storage is incredibly important. The company has about 30,000 cinema screens around the world from which it needs to pull data that allows the software it builds to function. Spinning disks couldn’t do the job—only flash has the performance needed to process all this information. Pure Storage provided the power needed to handle these massive customer data demands, saving the company around £1 million.
Powering the Film Studios of Tomorrow
As more studios outsource data production demands to bigger data centers, processing farms, and the cloud, modern data storage providers that stay in lockstep with their demands will be key. The more (and bigger) projects a studio wants to accept, the more critical this is.
Studios can benefit from as-a-service, pay-as-you-go models, removing the need for large up-front investments in render farms, workstations, and the storage they require. As many leverage the cloud for processing and rendering power, data mobility and accessibility will be paramount.
The Pure Storage portfolio offers cloud native, fast file and object storage solutions that uncomplicate data, with:
- Storage automation, to simplify provisioning and scaling so IT can seamlessly deliver resources
- Storage as a service, with subscription and pay-as-you-go models
- Cloud and container native storage
- Data archiving and backups that are immutable and require less space, without interfering with performance of production systems
Workloads that are capacity oriented and don’t require top-level performance can benefit from Pure’s QLC-based arrays that can replace hard-disk drives with a lower total cost of ownership and massive reductions in power, space, and cooling.
To learn more about how Pure can help your digital effects, animation, or film studio meet all your biggest data demands, contact us.
¹ https://www.nytimes.com/2022/12/16/movies/avatar-2-fx-cgi.html
