Artificial intelligence and its impact on the film and television industry was one of the main topics of discussion at the SMPTE DC Chapter’s “BITS BY BAY” conference held in Chesapeake Beach, Maryland from May 21st to 22nd.
The meeting was organized by the late Peter Wharton, former Chief Strategy Officer of Tag Video Systems, but returned after a long break, and the event is dedicated to his memory.
It’s going to be round
In a presentation dedicated to fractional film frame rates, James Snyder, formerly the National Audiovisual Protection Centre (NAVCC) at the Library of Congress and now an industry consultant, discussed the evolution of frame rates in parallel with film and television development, and the hope that the film industry will move digitally, so frame rates can be revised to address the flow market.
You might like it
The battle that fought across interlaces and progressive image standards during the transition from analog to DTV decades ago shows the upcoming battle between broadcasters, mostly supporting interlaces, and computer industry, which supported progressive decades ago.
In a world where 24 fps for traditional film is actually 23.98 frames (FPS) for television and digital video and 60 fps is technically 59.94 FPS, Snyder said it’s time to close to the next number (a uniform integer) as technological advances have not made fractions less relevant in today’s media distribution ecosystem.
With streaming and gaming still in place over the past decades, progressives have won that fight, and streamers simply don’t care about frame rates, Snyder added.
“Interlaces and fractions are two aspects of the same coin. These are both the problems that are necessary to get rid of,” he said. “(Interlacing) is the easiest to get rid of it now, as all displays are natively progressive. We don’t have CRTS anymore, and progressives are currently burned into technical distribution, or at least the final viewing.
Snyder denounced the current US broadcast television standard.
“The only reason we can’t even have an integer right now is that ATSC 1.0 is in the way. ATSC3.0 can handle that, and the display can handle that,” he said. “All vendors that I’ve spoken to people who have equipment that produce or post-production are even in service to integers. It seems that the only thing currently in the way is ATSC. 1.0 is a technical blocker.”
Snyder said he came to this conclusion in his work at the film library, where he worked in the digitalization and recovery of films. The demands of adherence to global media standards and the storage and management of millions of hours of film and video content have led us to consider new ways to approach cost and efficiency issues.
Streaming doesn’t care what your framework is. Your computer simply adjusts it to present content. ”
James Snyder
“We had to deal with the interlacing issue. We had to deal with the fractional frame rate issue. And we had to deal with the technical issues that come with creating new versions for reuse. “So I was able to see the extra costs firsthand. And in reality, just producing new content increases costs, and when it comes to reusing old content, the production costs increase.
“So, that’s where a lot of us came together and asked ourselves, ‘Do we really need to do this?'” Snyder added. “It’s a true technical reason why we’re still doing this, and all we can come up with is ATSC 1.0. If we take ATSC1.0 out of the mix, ATSC 3.0 doesn’t have that problem. Streaming certainly doesn’t have that problem.
Snyder added that this approach should be changed when considering live broadcasts.
“Areas still need some discussion and some work, “How do you proceed from a live feed of fractions to a live feed that is an integer?” we need to talk about it.
Democratization of AI and the Media
Deloitte’s Managing Director John Huten discussed the democratization of media production over the past 40 years and how AI is affecting the future, noting that for just $500 today’s mobile devices can produce specialized content that costs thousands of dollars in the 1980s. (footen also pens the Media Matrix column on TV Tech.)
“You don’t even need to educate yourself to do that. People are just doing it… that’s the overall trend of democratization in the production and distribution of content,” he said. “Have we finished democratizing the media? Are they all democratized? The answer is no.”
The average Joe’s ability to produce AI-powered Disney-level content is coming sooner than we can thank, Footen added.
“That level of production would be possible in theory for the good or bad people, that would happen,” he said. “And you’re going to democratize the content types that anyone can create. You’re not good at using cameras because you don’t have to touch them to use them. You don’t need to get a lot of training and framing and stuff like that because the machine is framed for you.”
Footten, detailed in Agent AI, is the concept of using AI “agents” to perform certain tasks that can ultimately replace today’s “app” environment.
“You can think of it in the same way as the word ‘app’ we use today,” he said. “An agent is a system… it could be a simple system, or it could be a complex system that performs functions. In reality, it does what you would expect from many types of applications.”
As these agents become more proficient in each individual’s media preferences, this trend leads to a breakdown of brands that still act as a kind of “walled garden.”
“Curator agents are agents who work on your behalf to get your content for you today,” Hutten said. Using the Plex video playback app as an example, footen said origination is not important to the more intelligent the user adds to it, the more intelligent the more information they are.
“What we need to add to that is to provide this AI feature that tracks who you are and what your preferences are, accesses your data and curates content,” he said. “In the end, that curation suspends media companies because why do you care about why it’s on which channel it’s on, what app it’s on… If it just comes to you through my agent, then my agent is my app. I don’t care where the content comes from.”
Content verification
Deloitte’s Footten colleague Matt Garek discussed the issues of authenticated content in the age of AI and detailed the latest developments on the Content Authenticity Initiative (CAI), a six-year-old project designed to promote transparency and reliability of digital content. C2PA is also a consortium of high-tech companies responsible for developing technical standards for content qualification, among other things.
C2PA is the “primary standard” for verifying and maintaining trust in digital content, and is on the path to becoming an ISO standard this year.
“All of this together provides a solid foundation for authenticity,” Garek said.
The C2PA standard doesn’t address AI-changing media detection, but it starts with the assumption that it is now, Galek said. “The C2PA assumes that users are changing parts of the media, and in some of the research I’ve seen, most studies state that over 50% of people who post content on social media platforms are manipulating it in some way, such as editing that content or adding filters. “So, that’s a fair amount of change in some way, and you’re unconsciously consuming that media.”
Galek pointed out the importance of metadata in tracking the evolution of content through a complete production chain, and how that important information is lost when content moves to different platforms. Adopting a C2PA will help protect that information, he added.
“There are a lot of technology used today where metadata is essentially stripped when media is transferred from one system to another,” he said. “For vendors who are adopting this standard, there is a way to preserve that metadata throughout the supply chain.”
The C2PA working group focuses on areas such as audio and video watermarking, but Galek said that cooperation within the industry is needed.
“To provide a forum for broadcast-based and other industries to collaborate, we need an additional spotlight on implementations,” he said. “We know how fragmented the broadcast vendor landscape is, so it’s far more important to work together around content reliability.”
Detect deep fakes
Anyone who dabbles in generative AI knows how much better it will get over time, but changing videos and images (aka “deepfake”) has been around for decades.
(Grogan discussed his work on TV Tech in 2018 on Video Forensics).
“When we started this it took a bit of Hollywood studios to make a really good deepfake. They couldn’t do much,” Grogan said. “We’re in the sense that some of these levels that were there now that little Hollywood studios were being done by middle schoolers.”
(Using historical photographs of Winston Churchill, Franklin Roosevelt and Joseph Stalin at the 1945 Yalta Conference and Groucho Marx instead of Soviet dictators, Grogan emphasized the importance of discovering contradictions as soon as possible, as widely distributed supplementary images can rapidly change public perceptions. In other words, it can allow the truth to travel along the way around the world before they put on the boots.
“If you think it’s a fake, then someone in the future might lose the reference material, so put it in your documentation,” he said. “And if you don’t say you think it’s fake, then the audience didn’t have a historical reference, so they really should be Stalin, and they may not know that it’s the wrong Marx.
While no one sees media companies losing gatekeeper status anytime soon, Hutten summed up the promise and dangers of AI to remove media companies as “middles” in content consumption.
“I think if everyone can do what we do today, our industry will be hampered,” Hutten concluded. “We don’t know how much of the media industry we left behind when everyone is a media company. Every company, every company is a media company. Historically, our job has been to create content, bring in audiences, monetize that content. But what if we don’t need to do that?
