Nile Rodgers defends his AI Christmas video and prices a traditional shoot at $100,000

Sylvain Chomet says his Dior film was drawn by hand, but software made some of its frames

Dior sold Chomet's short as all drawn by hand; his producer now says an automated tool made some in between frames. Neither has named the software or said how many frames it touched.

Dior launched Sylvain Chomet's new animated short as all drawn by hand, and now Chomet's own producer says software made some of its frames. When a label like that turns out to need a footnote, people stop trusting what they are told about how anything was made.

Christian Dior's Night in Color is a six minute holiday film narrated by Robert Pattinson. Dior's creative director Jonathan Anderson told WWD it was "all hand-drawn using techniques that go back to the 1940s." Animators online soon posted frames with a six fingered hand and other glitches they read as signs of generative AI, and Cartoon Brew removed "hand-drawn" from its headline and apologized for not watching more carefully.

Chomet and producer Aton Soumache told Variety the film took seven months and about 20 people, six of them animators, drawing on tablets. Soumache added: "We used an automated interpolation system for a very small part of the process." Interpolation means software filling in the frames between an animator's key drawings. Chomet says the fingers are an old animation trick, a hand fanned out in motion.

Neither man has named the tool, said whether it uses machine learning, or said how many frames it touched. Older in-between software only blends the drawings it is given, while some newer tools use AI to invent the missing frames, so that detail decides what "drawn by hand" means here.

Our take: drawn by hand is now a claim a brand has to show, not just say. Naming the in-between software would settle this faster than any interview.

Source: Variety, Nov 7, 2026; Cartoon Brew, Nov 6 and 7, 2026.

Nolan says The Odyssey's credits forbid AI training, but only Universal can enforce the line

Christopher Nolan says The Odyssey's end credits state the film cannot be used to train AI, and called it a first. Universal has used a similar warning since June 2025, and as he says, only the studio that owns the copyright can enforce it.

Christopher Nolan is pleased that the end credits of The Odyssey say the film cannot be ingested, scraped or used to train generative AI, and he called it a first. A line like that is only as strong as whoever owns the film, and on almost every Hollywood movie that is the studio, not the director.

Nolan, president of the Directors Guild of America since fall 2025, made the point on the Hollywood Insiders podcast on Nov 7. "With very few exceptions in the guild, we are not the copyright holders of the films that we've worked on," he said. "It's down to the studios to do that." The guild's next push, in his words, is holding the studios to account for protecting those copyrights.

The first is less new than it sounds. Universal has put a "may not be used to train AI" warning at the end of its films' credits since How to Train Your Dragon in June 2025, followed by Jurassic World Rebirth and The Bad Guys 2, The Hollywood Reporter found. In some countries that warning cites a 2019 European Union copyright law that lets rights holders opt out of data mining. Nolan may mean a first for his own films or for the "ingested" wording; the exact text in The Odyssey's credits has not been published.

He also argued that AI companies leaning on fair use will lose in the long run, comparing training to music sampling, which the courts settled decades ago. That is his prediction, and the training cases are still being fought.

Our take: a credit line is a notice, not a lock. Whoever holds the copyright decides whether it is ever enforced, which is why the guild is now pressing the studios as well as the AI companies.

Source: Hollywood Insiders podcast via Deadline and TheWrap, Nov 7 and 8, 2026; The Hollywood Reporter, Aug 2025.

Studio Meta-K's Kim Kwang-jib warns AI feature films could end up costing more than the 35mm era

Kim says usage fees for a 100-minute film on the priciest AI tools can top $366,000, and a tenfold price rise would reach the top of a typical Korean feature budget. He did not say which tools or resolution the figure assumes.

Every change of mind on an AI film is a new bill. Kim Kwang-jib, founder of the Korean AI studio Studio Meta-K, told Taiwan Creative Content Fest in Taipei that the usage fees (what AI services charge for every image and clip they generate) for a 100-minute film on the most expensive commercial tools can top KRW500 million, about $366,000, so the savings people expect from AI could vanish into revisions and price rises.

His bigger number rests on prices that have not moved yet. AI companies are losing money and keeping prices low as marketing, he said; a tenfold rise would take that bill to KRW5 billion ($3.66 million), the top of what a typical Korean or Taiwanese live-action feature with visual effects costs today.

The mechanism is the useful part. Every re-render is paid for again, so ten client notes sent at once cost one extra pass, while the same ten sent one a day cost ten. Kim compared it to the 35mm era, when every take burned film stock and preproduction was meticulous; unless the plan is locked first, he said, AI could end up costing more than film did.

Advertising is a different story. A Pizza Hut spot that took about eight months and KRW2 billion to make was remade by his studio in two weeks for around KRW30 million ($22,000), about 2% of the cost.

On the same panel, Lee Sang-wook of MBC C&I's AI Content Lab said his lab's 139-minute feature Raphael (pictured) took about 16 months, with a single shot sometimes occupying 10 artists for a whole day. It opens in Korean cinemas on Nov 25.

Kim did not say which tools or resolution the $366,000 assumes, so read it as one studio's ceiling rather than a price list. Our take: the lesson travels anyway, because on AI work the expensive part is changing your mind one note at a time.

Kamen Rider studio Toei is making backgrounds with AI under rules it has not published

Toei's CEO says the studio wrote generative AI rules and began using AI in film and video this year, starting with backgrounds. The rules are not public and no production has been named.

Toei, the studio behind Kamen Rider and the Super Sentai shows that became Power Rangers, says it has written rules for generative AI and is already using it to make backgrounds for film and video. Nobody outside the company has seen those rules, and they matter most to the artists whose work they cover.

CEO Fumio Yoshimura said so in a keynote at the Taiwan Creative Content Fest in Taipei, reported by Variety on Nov 6 and Deadline on Nov 7. A dedicated team tests practical uses every day, Toei is building a tool to help junior producers estimate budgets, and back-office staff already use AI for routine work. He called AI a tool that expands what people can make rather than one that takes jobs, and said human creativity remains the core.

He named no titles. Posts on X have since tied the news to Kamen Rider, but in both outlets' accounts he did not single out any show. Toei has not said which productions used AI backgrounds, or whether the rules reach Toei Animation, a separate company in which it is a major shareholder.

Background painting is skilled work with its own artists on every production. Whether the rules require disclosure, credit or consent is what would show if the promise about jobs holds.

Our take: rules nobody can read protect the studio more than its artists. If AI backgrounds pass Toei's own test, publishing the rules and naming the work would prove it.

Source: Variety, Nov 6, 2026; Deadline, Nov 7, 2026.

Pokemon anime studio OLM built an AI coloring tool that learns a show from three hand painted frames

OLM says its coloring tool learns a show's palette from three hand painted frames and cut painting work by about 30% in early tests. The checkers who review every frame now have more to do, and OLM has not said if it runs on Pokemon episodes yet.

Three hand-painted frames are all OLM's new coloring tool needs to learn a show. OLM makes the Pokemon anime series and films, and its tool flips the usual AI bargain: instead of training on a mountain of other people's work, it learns only from the production in front of it.

OLM Digital board director Yotsukura Tatsuo described it at the Taiwan Creative Content Fest in Taipei. An artist colors three frames by hand, the system learns that show's line shapes and color palette in about 90 seconds, then colors each remaining frame in around 15 seconds. The more frames an artist paints, the more accurate it gets, and a new character, even a dragon, can be added straight away.

"Strictly speaking, it isn't generative AI," Yotsukura said. Because it trains only on the production's own frames, he argued, no other works are harmed. It came out of ANIMINS, a 15-month research project he led under GENIAC, a generative AI program overseen by Japan's Ministry of Economy, Trade and Industry, which wrapped in February.

The early numbers are modest. At proof of concept stage, Yotsukura estimated painting work fell by about 30%, but the checkers who review every colored frame now have somewhat more to do. OLM has not said whether the tool is used on Pokemon episodes yet, or whether other studios will be able to use it.

It is not the only attempt. Toon Boom, maker of the Harmony animation software, says the coloring beta in its Ember AI suite is wrapping up; one Manila studio that tested it cut painting on a 90-second to two-minute scene from 15 hours to two or three.

Our take: training on the show's own frames answers the consent question more directly than any policy statement. The open question is the checkers. If their extra hours eat the 30%, the saving is a reshuffle, not a cut.

Source: Variety, Nov 7, 2026; METI GENIAC.