The Future of Technology in Media Production
Through the lens of AI and Cloud
Which is true?
- Dogs can order their favourite toys on the internet themselves, or
- Tom Hanks re-shot all his scenes in Forrest Gump in Japanese?
I took some time to visit the Media Production Show in London this year. Most of my time is spent with industries such as financial services, telco, airlines and governments. With a keen interest in the film, TV and the creative world, it is always a great opportunity to look at different industries and their adoption of the core technologies of the day.
I went to enjoy and look at this world through the lens of cloud and AI.
TL; DR;
- Cloud — the industry wants it and needs it, but its costs are too unpredictable
- AI — is radically changing what’s possible in the creative process, at an extraordinary rate
The Media Production and Technology Show has been running for many years, and 2022 was the first return to a face-to-face event after the pandemic. MPTS is the largest UK event for media and broadcast industries. It focuses on the creative collaboration of content production and exploring technological innovation across the lifecycle of pre-production, production, virtual production, post processing and distribution.
What I found was a lot of smiling faces of people who had not seen each other for quite some time. I saw an array of technology consuming media production companies and their technology providers. They were emerging bleary-eyed from the biggest jolt to their ways of working. It really was a mix of companies that were figuring out how they survived the change, and those that had expertly pivoted to use technology for remote collaboration and media production.
Cloud
Cloud was very apparent in technology offerings in many of the stands at the event. There were companies like Tyrell, an event sponsor, who were all in. Their goal to offer full lifecycle capabilities across broadcast, live, animation, pre- and post-production as various services to integrate into workflows — with or without their help — alongside fully managed service platforms.
Others like Twickenham Film Studios greatly increased their resilience and global collaborative capabilities by moving their work to use hybrid on-premise and cloud platform services. In another discussion about virtual production it was noted that in any of these studios there was a plethora of “single task” workstations (and screens, and keyboards) and the tide needs to shift so that these systems are virtualised and can be operated through one workstation for a reduced footprint and more agility in the studio.
One panel discussion surfaced the benefits and the concerns of cloud in the industry. On the panel were Nick Sopel (Tyrell), Jess Nottage (Clear Cut), Tracie Mitchell (Greenfish), and Nick Pearce (Object Matrix)
All the panel agreed on the benefit of cloud: how it is instant and scales with producers especially loving the cloud for its flexibility in enabling the workflows across the lifecycle of film. The theme of predictable cost emerged. Smaller companies in this industry shouldn’t be looking to the hyperscalers like AWS and Google Cloud direct since they have no buying power for discounts, even though they have big data workloads (e.g. rendering), their workload usage is so variable and not necessarily sustained usage. The attention should be on SaaS offerings.
The workload discussion went across bursting for UHD rendering , whether to do post-production online or offline, and how it brought positive social interaction during production.
The key hidden costs were led by cloud data egress charges. This can be overcome if a company can have defined workflows. There was recognition that cloud providers offer multi-region geo-redundancy, but in some hyperscalers the egress from one region to another was a problem cost. Tyrell commented on how their pre-integrated vertical platform was “use case” oriented and their charges are by the use case transaction rather than the piece parts of cloud consumption underneath.
Another area of cost concern was DevOps and system integration skills. With the power of using cloud provider technologies comes the need for core cloud technology skills often beyond the technical creative skills of production companies. The costs to hire and maintain such staff are prohibitive. FinOps was a known topic, but early days in maturity of its consumption by production companies.
On the topic of data, identity and security were of great concern. There was a view that 98% of companies suffer a security breach whilst in the cloud, and being a target was a factor in slow cloud adoption. The panel were watching data privacy laws of UK and France, reminding us that the production business is a global ecosystem of companies that need to interact across the lifecycle.
In conclusion the panel gave their thoughts as to when the industry will have shifted to cloud. Their answer was 10 years. Who knows, technology changes and adoption are always accelerating?
Artificial Intelligence
Whereas the cloud conversations focused on process enablement, efficiency and costs, the AI conversations focused on the excitement of the creative possibilities and its potential alarming impact on the industry.
Dr Alex Connock of Said Business School, University of Oxford opened everyone’s eyes with his talk on “Let’s use more AI tools in UK TV creativity”. He is an Oxford Fellow and focuses on business media and AI, with previous career experience at NFTS.
The talk discussed how AI can be used in the creative process of TV, and other uses in media industries.
Dr Connock performed a lightning tour of AI and technology companies, highlighting that the games industry was primarily driving AI innovation. AI is impacting the full lifecycle of development through to production and to distribution. Just one look at how the market capitalisation of media companies has been changing (tracked by Evan Shapiro) is fascinating — especially in recent years the growth of Nvidia.
AI is now painting. This year has seen the release of DALL-E-2 from OpenAI. It can create original, realistic art and images from text description.
To demonstrate the power of AI, Dr Connock showed a clip from the documentary about AlphaGo and the emotional impact AI had on its opponent Lee Sodel (at 49m). Even AlphaGo the system stated that move 37 had a 1 in 10,000 probability that a human would have made the same move — it was an original creative move (by AI) in the game of Go. The reaction from Lee when he returns from a break and sees move 37 is wonderful filmmaking.
The application of AI in the creative process of film, TV and media is now evident. Flawless advances dubbing to change lip movements in film so that they sync to alternate languages.
Charisma AI is helping create interactive stories and believable virtual characters. It also turns out that synthetic faces are more trustworthy than human ones.
Cinelytic is analysing scripts and generating predictive analyses of content for the entertainment industry. Other examples were Storyfit and even Petz commerce enabling dogs to do the shopping from their facial expression!
OpenAI’s GPT-3 model has the power to read a book, compose poetry, and auto-complete text — but with some notable biases. This is a known and important aspect of AI in today’s society. Who owns the data sets that trains the AI and how do we address biases?
“We’re really closer to a smart washing machine than Terminator”, Fei-Fei Li, Director, Stanford A.I. Lab (AlphaGo documentary)
The awe and impact of AI in the media and entertainment industry was apparent. It is generating excitement whilst feeling the trepidation of AI’s power. As Fei-Fei Li is indicating, we are only just at the beginning.
Reflection
Good strategy looks for inspiration from all walks of life. Industries such as media and entertainment are stretching the boundaries of creativity through AI. Despite the perceived cost of cloud, the industry has a fresh perspective — they look at the benefits of cloud through the lens of business use cases.
(Views in this article are my own.)
