NAB
A randomized
documentary of the career of legendary techno-music pioneer Brian Eno, in which
every screening is potentially and infinitely different, is the latest project
to be served up by generative AI.
Eno is billed as the world’s
first generative cinematic documentary. “Like a musical performance that’s
different every night, the film creates a unique viewing experience for each
audience that takes it in,” explains Matt Grobar at Deadline.
article here
The 75-year old British music producer and visual artist who has worked with David Bowie, U2, Grace Jones and Talking Heads, and who birthed the ambient music genre and frequently mixes technology with art, is ripe for a video retrospective.
“I usually can’t stand
docu-bios of artists because they are so hagiographic,” Eno told Variety’s
Todd Gilchrist.
So, rather than charting a
chronological path through Eno’s career, documentarian Gary Hustwit proposed
using a generative system to create a film that would literally be different
for every audience that screened it.
“The use of
randomness to pattern the layout of the film seemed likely to override any
hagiographic impulses,” Eno said.
If that was
enough to pique Eno’s interest in the project, for Hustwit the approach was
about provoking new ways of creating and experiencing a film.
“I like movies
where you learn different things about the subject, but you, as the viewer,
make the connections… I always think that’s a lot more rewarding, as a viewer.
It’s a different kind of filmmaking, but it’s also a different kind of film
watching.”
It helps that
the first and last scenes of the 85-minute doc are always the same. Plus, there
are certain scenes pinned to the same timeslot in each version, including a
scene where Eno discusses generative art.
“We thought that was
probably a good scene that everybody should see,” Hustwit told Lauren
Forristal at TechCrunch.
Everything
else, however, can be different, depending on the material the generative
program decides to insert.
“It’s kind of a modular
approach,” Hustwit explained to Forbes’ David Bloom. “You can learn
different facts about that person at different times in the film. In the end,
you make the connections as a viewer.”
Like one of media artist
Refik Anadol’s AI creations, Eno is going to be different each
time it is screened. That poses a problem for film critics, Bloom points out.
To DeadlineHustwit
explained, “There are billions of different combinations that could possibly
exist of this movie, and every time you watch it, you’ll never see that version
again. So, it’s an interesting experiment. We can change the way that the form
of film works [so] let’s talk about the possibilities.’”
Hustwit had another reason
for making the film this way too. It’s a showcase for the generative tool
(cutely dubbed Brian One) that he has built along with digital artist Brendan
Dawes by their startup company Anamorph.
The tech was
trained to select scenes from over 500 hours of archival footage and new
interviews of Eno as well as animated visuals and music to produce the unique
iterations of the doc.
Anamorph spent
five years building the software, combining patent-pending techniques with the
team’s own knowledge of storytelling. The company says it’s not trained on
anyone else’s data, IP or other films.
“The main challenge was
creating a system that could process potentially hundreds of 4K video files,
each with its own 5.1 audio tracks, in real time,” Dawes tells TechCrunch.
“The platform selects and sequences edited scene files, but it also builds its
own pure generative scenes and transitions, creating video and original 5.1
audio elements dynamically. The platform also needed to be robust in a live
situation, it wasn’t an option to have it crash. So, we did a crazy amount of
testing. We can create a unique version of a film live in a theater, or we can
render out a ProRes file with its own 5.1 audio mix and make a DCP from that.”
He also
stresses, “This is a generative system, not generative AI. I just need to make
that clear, because pretty much everything that’s been said about Eno uses the
word AI.”
Advertising
agencies have apparently expressed their interest, Hustwit reveals to
TechCrunch, with one company wanting to make 10,000 versions of a one-minute
commercial.
Rather than
make its tools publicly accessible, the company wants to collaborate on
projects so it can “consider the source material and the overall story goals,”
says Hustwit.
“Our main goal
is to get the idea out about this new kind of cinema and hook up with great
collaborators to help explore this idea.”
Hustwit ponders what an
experimental form-pushing director like Jonathan Glazer (The Zone
of Interest) could do with something like this.
“You could make a movie
that’s always on, always evolving, always changing,” Hustwit told Forbes.
“I feel like Eno, it’s really kind of an opening conversation. What’s next?
What can we do with this?
A streaming
service such as Netflix — which has played with interactive forms of video —
could easily generate a different version of the documentary every day, Hustwit
added.
However, to TechCrunch he
poured cold water on the idea, saying that streaming networks aren’t equipped
to dynamically generate unique video files and stream them to thousands of
viewers so that each viewer is getting their own version of a movie.
“When we premiered Eno at
Sundance, all the big streaming companies loved it, but they also admitted that
their systems can’t handle the tech involved… These streamers need to
differentiate, and I think enabling the films and shows they’re releasing with
generative technology is a way to do that,” says Hustwit.
It’ll likely
take years before streaming services adapt to the technology. Until that
happens, Anamorph is sticking to live events and theatrical releases.
“Something
that the theater industry badly needs right now is a reason to get people to
come in, and if there is a uniqueness about the live cinema experience, that’s
one way that can be achieved,” he adds.
Six versions
of the film will be shown at Sundance, with additional screenings across 50
cities to be presented later this year.
No comments:
Post a Comment