From Factories to Film Sets, AI Has a Simple Goal

[ad_1]

Making movie magic in a Hollywood film studio has more in common with churning out cars at a Bavarian factory than most of us would like to admit. We see cinema as some creative symphony of sight and sound compared to the mechanical coldness of automobiles. Yet there’s a reason why both industries use the same word: production.

Artificial intelligence is seeping into these businesses for largely the same goal: to boost efficiency and increase productivity. The current actors’ strike in the US has shone a light on the reality of a profession which is at the same time glamorous and brutal. Among the biggest fears is the notion that performers will be replaced by their own avatars — digital twins — which preserve all that is human while rendering the actual person entirely redundant. It’s a valid concern, but likely overblown.

To examine how the use of AI may actually play out in Hollywood we can look at the car industry. BMW AG assembles more than 40 models at 31 factories around the world. With dozens of customizable options, there are at least 2,100 possible configurations of an automobile that bears the German giant’s logo. Some of those changes are easy, such as swapping out the paint color. Others require tweaks to the assembly line that have ripple effects on the entire process, including on those workers tasked with getting a car out the door as quickly as possible.

Customization is the enemy of productivity. A factory manager wants to minimize downtime between changes to tooling while ensuring each factory setup is as efficient as possible. That’s where artificial intelligence comes in and, unsurprisingly, Nvidia Corp. is a major player.

The US chip designer has become synonymous with the AI boom. It not only sells the hardware used to crunch numbers and spit out text, photos and videos, but it’s developed a broad suite of software that helps companies run more smoothly. Its Omniverse platform, for example, is used by BMW to create digital twins of a car plant, allowing management to manipulate 3D models of an assembly line in real time and optimize productivity.

Industrial giants Honeywell International Inc. and General Electric Co. are also keen to roll out their own digital-twin products for the oil and gas, power-generation and mining industries as a way to predict equipment failure, maximize uptime and lower maintenance costs. GE claims AI can boost reliability by more than 93% within two years. BMW says this technology makes the factory-planning process 30% more efficient.

Offer the line producer of a Hollywood movie a one-third boost in efficiency on a film set and they’ll likely jump for joy. It’s that person’s job to go through each line item of a film’s budget to maximize return on investment — hence the name. And it’s not just about saving money. Beyond counting pennies, they need to ensure working hours are adhered to and a minimum number of crew are employed on set.

Unions have very strict rules on everything from working hours to job roles. And they also define wages according to the budget of a movie. A higher-value production must pay its staff more, which means if costs go over by even one dollar, pay scales could jump into the next tier and blow out costs. A first assistant camera operator, for example, gets 65% more if the budget spills over the tier-one cap of $7.5 million.

Even modest Hollywood productions costs at least $100,000 per day of filming. While there’s great benefit to digitally replicating background actors to cut costs, their day rate, at around 17% of a feature actor’s wage, is less a factor than making efficient use of the film’s stars. If a producer could squeeze out more from a lead actor’s day, or get the same amount of scenes shot from fewer days of filming, then the financial benefits are exponential. It would also reduce time on set, which means reduced cast and crew costs, completion of an equivalent production within a lower budget tier, and smaller wage rates.

That’s where AI and digital twins enter the picture, and dozens of companies including Nvidia and Walt Disney Co. are building this future. The risk isn’t just stealing an actor’s likeness — you can be sure the unions will prevent that — but to cut their time on set, which is the yardstick for most performers’ wages. Rather than replacing a human for the entirety of a film, we could see increased use of footage from one scene being used as a template for other scenes while preserving the imperceptible human quirks which distinguish live-action from animation.

In fact, it’s already happening. Actors have told me of incidents they where they appeared on screen in parts of a production they weren’t even aware of. Others have been asked, and refused, to have their faces scanned by high-resolution sensors including lasers. Future deals between producers and performers could include rights to digital copies limited for use only in that production, with commensurate premiums to the base rate of pay.

It may not be romantic, but digital twins are here to stay. Movies, like cars, are a product — but they also reflect the progress humans make over time. And one of those changes is the never-ending pursuit of technology and productivity.