2 weeks ago
Three world filmmakers – moving successful nan U.S., China and South Korea – discussed really AI filmmaking tin complement alternatively than switch existing accumulation techniques connected a Cannes Marche sheet hosted by AI video procreation level Kling AI.
All 3 filmmakers are presently moving connected projects that impact Kling AI tools. Wonder Project co-founder Jon Erwin has already had House Of David streaming connected Prime Video and precocious dropped nan first section of three-parter The Old Stories: Moses, starring Ben Kingsley, connected nan platform.
Chinese animation head Wei Li, who has accepted credits including Annecy title Jiang Ziya: Legend Of Deification (2021), is presently moving connected AI-driven characteristic Born Of The Tides, while Korea’s Eekjun Yang, co-founder of Mateo AI Studio, is producing AI sci-fi action characteristic Raphael, group for theatrical merchandise successful Korea later this year.
In a week successful which sheet aft sheet astatine nan Cannes Marche attempted to assuage manufacture fears astir nan effect of AI technology, Erwin whitethorn person offered immoderate alleviation successful describing nan methods utilized by his accumulation outfit Wonder Project, which combines AI video procreation pinch existing technologies specified arsenic capacity capture, virtual accumulation and VFX.
Expressing his vexation astatine erstwhile projects being turned down by studios owed to constraints of clip and budget, Erwin said The Old Stories: Moses became imaginable erstwhile a short model opened successful Kingsley’s schedule astatine nan opening of nan year: “I had nan thought successful January, wrote it successful February, Amazon greenlit it and we filmed successful March, and nan first section conscionable came retired successful nan U.S.. So nan velocity of this accumulation is arsenic important arsenic nan savings successful cost.”
Wei described really Born Of The Tides is besides incorporating AI devices into accepted filmmaking methods, successful this lawsuit successful animation: “It’s not conscionable an AI-created film; it’s not pursuing nan AI workflow only, we’re still utilizing immoderate accepted animation creation. But successful utilizing AI, we’ve shortened nan accumulation schedule to 2 thirds of nan time, and likewise pinch nan budget.”
Yang is leaning moreover much heavy connected AI successful nan making of Raphael, saying a large chunk of nan fund was spent connected monthly subscriptions to astir a twelve Kling AI tools. He besides said nan task wouldn’t person happened without AI. “If we’d filmed successful existent life, it would person required 150-300 group and a fund of up to $2M. But we had a squad of conscionable 7 and overmuch little cost.”
The feature-length film, astir an android who escapes and embarks connected a belief journey, is produced by Mateo AI Studio and MBC C&I AI Content Lab, owned by Korean broadcaster MBC. A Korean theatrical merchandise is being lined up for later this year.
Later successful nan panel, Erwin explained that Moses was mostly changeable connected a VP shape and that Kingsley enjoyed nan acquisition because it was much realistic than shooting against a greenish screen. The accumulation utilized AI to create integer assets that could beryllium output successful a matter of hours alternatively than weeks: “We tin make caller worlds connected nan aforesaid time arsenic shooting and that enables nan actors to measurement into a very existent space. We were doing 1 segment pinch an ocean, and he joked that he thought he was going to get his feet bedewed moreover though it’s not real.”
Erwin besides encouraged existing movie technicians to study really to usage nan caller tools, saying that nan exertion will create faster if nan platforms person feedback from users, and that nan champion AI artists are existing section heads – editors, cinematographers, accumulation designers and costume designers – who person been retrained.
“Ideally, I want to activity pinch nan unit members I’ve worked pinch for a very agelong time,” said Erwin. “But I want to activity successful this benignant of real-time way, truthful each these phases of filmmaking that are bifurcated from each different – development, main photography, editorial, post-production, ocular effects – are nary longer very acold apart. You tin create a script wherever they’re each happening connected nan aforesaid day.”
Speaking of challenges, nan filmmakers agreed that scenes that are difficult to create successful unrecorded action, specified arsenic battles, are overmuch easier erstwhile utilizing AI, but that nan other is besides existent – AI tin still beryllium a work-in-progress erstwhile it comes to close-up affectional scenes.
Yang said: “In nan making of Raphael, 1 of nan easiest things to create was nan large-scale warfare subject scenes, while 1 of nan astir difficult scenes was nan friendly relationship of 2 characters. With affectional expression, it’s very important to guarantee there’s bully relationship betwixt nan scenes. However, nan AI accumulation outputs would only beryllium capable to create astir 15-second maximum length, truthful to link nan footages people and create and support that affectional relationship was very difficult.”
Owned by Chinese tech elephantine Kuaishou, Kling AI operates a text-to-video image procreation platform, which precocious launched a type supporting autochthonal 4K video output. Launched successful June 2024, and since upgraded much than 30 times, nan level has been utilized by astir 60 cardinal creators crossed 224 countries and regions.
Erwin’s Wonder Project has a multi-year narration pinch Amazon MGM Studios and operates a subscription work connected Prime Video, wherever it watercourse its originals. Wonder Project besides precocious teamed pinch AI startup Luma to motorboat a caller company Innovative Dreams.
In summation to Jiang Ziya, Wei’s credits see Big Fish & Begonia and Edge Of Time, a Golden Goblet nominee. Before moving connected Raphael, nan produced AI shorts Mateo and Witness, which won Grand Prizes astatine KAIFF and BIAIF successful 2024.
