NAB
Arguably, the studio system in Hollywood has never had the cards stacked in favor of risk taking. To fuel the machine it is by and large the tentpole movies of franchise IP which gets greenlit, reducing genre to superhero and horror movies, and squeezing out mid-budget indie fayre with more original stories for $100m budgets in the hope of recouping $1bn.
article here
Could AI save the day? On the face it that seems the worst
possible idea. A machine replacing a ‘dream’ factory, yet that is exactly what
could come to pass if you take the positive view of AI tools that are rapidly
gaining traction in all parts of the creative process.
AI and so-called AI generated motion pictures represent a
genunine threat to the Hollywood elite, reckons Pinar Seyhan Demirdag, the
co-founder of Seyhan Lee, a production company specializing in AI.
Just as the smartphone revolution put a high quality camera
in everyone’s hand, a swathe of ‘creative AI’ apps will bring the production
values of VFX, costume and exotic or fantastic locations into the grasp of any
creator.
“The top-down triangle power pyramid will come to an end
with creative A.I. tools,” she argues. “In the following years, creative AI will be taking all the limitations between
your imagination and the production in the movie-making pipeline away, turning
each filmmaker into their own personal production studio.”
AI motion pictures will be an artistic revolution
facilitated by GPUs and parallel processors, she says.
“Creative AI tools are helping us emancipate our creative
self-expression. You can now paint like a surrealist with Midjourney, mocap
like a VFX studio with A.I. + Ebsynth, and give motion to avatars like a
professional animator with Thin Plate Spline Motion Model. From an existential
point of view, creative AI is giving us all the tools, talents, and arguments
for us to make an idea conceptualized, funded, produced, and come alive.”
Of course, as it stands right now, AI is rubbish at
generating any meaningful motion picture let alone with a compelling narrative.
Most people think it is only a matter of time before that changes though.
It’s happening now, in fact. Demirdag traces how the
revolution will happen before our eyes.
First to use AI in the open, she thinks, will be credits and
title sequences because they “are the segments of the production pipeline with
the most artistic freedom.
“As the outcomes of the creative AI models are still wild,
wild west and are more artistic than realistic, this gives an excellent
opportunity to spice up the title sequence market,”
Expect much more artistic AI-generated openers and title
sequences in 2023, she predicts.
The second stage will be the integration of AI into Unreal
Engine and volume stages for virtual production. Demirdag is at an advantage
here since she teases that her company is currently doing just this on a stage
in LA.
Concepting, storyboarding, and animatics will be the third
phase. “Creative AI will liberate the storyboard artist from the
repetitive, ‘revise til you die’ burden and cut down costs/production
times.”
AI is already being applied to VFX and further impacts will
occur “when artificial neural networks learn real-world physics, temporal
cohesion, and filmmaking lingua franca.”
Text to video AI models are emerging including Facebook’s Make a
Video (which is 2D computer generated) and some early-stage text-to-3D
models including Dreamfusion, GET3D from Nvidia, CLIP Mesh, and CLIP Matrix by Nikolaj
Jetchev.
“[Soon], we will witness these open-source models and new
ones integrating light, coherent foreground and background generation,
video-driven motion, performance, camera movement, generated depth maps, and
all other aspects of the production pipeline.”
Demirdag predicts that some entity (Amazon perhaps?) will
combine all these scattered production pipeline tools “under one interoperable
roof, like a virtual film generation studio.”
She suggests that studios like her own specializing in AI
motion picture making will pop up, either serving traditional studios as
vendors or merging with them.
“Needless to say, there will be AI Netflixes carrying only
generated content.”
Going further, “a generated-films-only streaming service
will likely be disrupted by AI-generated dream states that can transition
seamlessly between the personalized reactive show, game, social experience, and
relaxing escape on a whim.”
It follows then that Hollywood is at a crossroads. Fail to
adapt to new technology and wind up like Kodak, Nokia, Yahoo, Xerox, and
Blockbuster or embrace AI for the benefits it can bring to cutting production
cost and boosting creativity.
That’s after a period of futile resistance.
“At the beginning of the AI motion picture revolution we can
expect some confusion, resistance, and lawsuits to arise from the studios at
the beginning, similar to the chaos around sampling in music. But when the dust
settles, the actors and directors will embrace AI, and they will get tomake double
the amount of money by working fewer hours as they will be licensing their
‘likeness’ and ‘style’ to generative films.”
Studios will develop novel ways to differentiate generated
filmmaking from conventional filmmaking.
“Studios will push directors to create a more original look
and feel for their movies. This effort to always try to be ahead of amateurs or
semi-professionals making professional films will give way to the emergence of
new trends in Hollywood.”
Her thesis is no outlier but part of a widespread belief
that AI will automate some jobs from the industry, but will create just as
many, and lead to potentially new forms of creative expression.
“Welcoming creative AI tools with resistance would be
similar to bashing digital cameras, cars, or other paradigm-shifting tools that
improve the human experience and expand freedom,” Demirdag says.
“I consider the limitation of human creativity a much bigger
crime than giving people a creative self-expression explosion tool, like
creative AI.”
No comments:
Post a Comment