{"id":678358,"date":"2026-11-08T05:56:10","date_gmt":"2026-11-08T05:56:10","guid":{"rendered":"https:\/\/www.newsbeep.com\/ie\/678358\/"},"modified":"2026-11-08T05:56:10","modified_gmt":"2026-11-08T05:56:10","slug":"pokemon-studio-olm-trains-ai-coloring-tool-on-just-three-frames","status":"publish","type":"post","link":"https:\/\/www.newsbeep.com\/ie\/678358\/","title":{"rendered":"‘Pokemon’ Studio OLM Trains AI Coloring Tool on Just Three Frames"},"content":{"rendered":"

\n\tOLM, the Japanese animation group behind the \u201cPokemon<\/a>\u201d anime series and films, has developed an AI coloring tool that picks up a show\u2019s look from just three hand-painted frames, OLM Digital board director Yotsukura Tatsuo said at the Taiwan Creative Content Fest<\/a> (TCCF) in Taipei.<\/p>\n

\n\tThe tool came out of ANIMINS, a 15-month research project Yotsukura led under GENIAC, a generative AI program overseen by Japan\u2019s Ministry of Economy, Trade and Industry. The project, which looked at how anime creators could put AI to use, wrapped in February.<\/p>\n

\n\tBecause AI models usually need vast amounts of training data that studios struggle to assemble, OLM took the opposite approach. An artist colors three frames by hand, and the system learns that show\u2019s line shapes and color palette on the spot in about 90 seconds, then colors each remaining frame in around 15 seconds. The more frames an artist paints by hand, the more accurate it becomes.<\/p>\n

\n\t\u201cStrictly speaking, it isn\u2019t generative AI,\u201d Yotsukura said.<\/p>\n

\n\tTraining only on a production\u2019s own frames means no other works are harmed, he added, and the tool can adapt immediately when a new character or creature, such as a dragon, appears. The point OLM most wanted to stress, he said, is that it fits into existing anime workflows without disrupting them. At proof-of-concept stage, Yotsukura estimated it has cut painting workload by around 30%, though the checkers who review every colored frame have somewhat more to do.<\/p>\n

\n\tToon Boom Animation, the Canadian maker of animation software Harmony and Storyboard Pro, is tackling the same bottleneck. Product marketing director Harry Ravelomanantsoa said the beta of the coloring tool in Ember, the company\u2019s AI suite, is wrapping up, and in-betweening is expected to enter beta within a couple of weeks. One Manila studio that tested the coloring tool cut the painting time on a scene of 90 seconds to two minutes from 15 hours to two or three.<\/p>\n

\n\tOLM employs around 600 staff across Japan, Malaysia, Thailand and the U.S. \u201cAs you probably saw passing through the first floor, we make the \u2018Pokemon\u2019 anime \u2013 not the games \u2013 including the TV series and the theatrical films,\u201d Yotsukura said, referring to the 2026-27 Pokemon Asia Championship Series: Taiwan Premier Ball League, held Nov. 7\u20138 on a lower floor of the Taipei Nangang Exhibition Center, where TCCF is running Nov. 6\u20139.<\/p>\n

\n\tThe group\u2019s slate also includes kids\u2019 titles \u201cBeyblade\u201d and \u201cHanakappa\u201d and the adult-skewing \u201cThe Apothecary Diaries.\u201d Yotsukura, who also serves as CTO for the group\u2019s Malaysian and Thai arms, noted that an in-house R&D division is unusual for a Japanese animation company.<\/p>\n

\n\tThe group is experimenting with AI for in-betweening as well. But Yotsukura said the bigger gains may come from indirect uses across productions that involve more than 100 staff. With a startup partner from the GENIAC project, OLM is exploring anime-specific AI agents that could help new directors or production managers joining a series that has run for a decade or more, for instance by recalling what a director intended for a given cut. Agents could also speed up approvals and license management for secondary uses of its titles, which Yotsukura called the most revenue-critical part of the business. Because those areas are sensitive, OLM is proceeding with care, he said.<\/p>\n

\n\tYotsukura likened the gap between AI demos and production reality to gacha, the Japanese capsule-toy machines where each JPY500 ($3.30) turn may or may not yield the prize you want \u2013 roughly what an AI generation credit costs, he said. AI gets close enough to 80% to raise expectations. \u201cRaising it from there to 90 or 100% turns into work so hard it can make you heartsick.\u201d<\/p>\n

\n\tHobbyists may be happy with 80%, he said, but professional animators need tools that return their direction 100% of the time and let them make corrections. Anime is also highly sensitive to color, and AI tools that edit an image can shift its RGB values even when the result looks fine.<\/p>\n

\n\t\u201cThe creative stays with the artist, and the mechanical time goes to AI, to Ember,\u201d Ravelomanantsoa said.<\/p>\n

\n\tYotsukura said he agreed almost entirely with that split. \u201cAI can only do what it has learned,\u201d he said, adding that the kind of images that move audiences in cinemas, and the never-before-seen ideas behind hit anime, still have to come from people.<\/p>\n

\n\tHe compared AI\u2019s arrival to that of 3DCG, which took around 20 years to be accepted in anime. \u201cTwenty years ago, I kept rejecting it, saying, \u2018How could this ever be used in anime?\u2019 Now it\u2019s unthinkable for anime not to have 3DCG,\u201d he said. AI will move far faster, Yotsukura said, because its quality is already high from the start.<\/p>\n

\n\tCharles Beirnaert, chief creative officer of Blue Spirit, the French studio behind Netflix\u2019s \u201cBlue Eye Samurai,\u201d said the studio uses no AI at all on that series. \u201cWe\u2019re not allowed to use AI for the Netflix projects,\u201d he said. Blue Spirit asks every time at what stage AI might be permitted, he added, and the answer is always no \u2013 though he acknowledged the rule may not apply to every Netflix partner. On an Amazon series, the studio may use AI only to pre-visualize characters and robots, with none of it reaching the final image. Netflix\u2019s published guidelines for production partners, issued in August 2025, require them to disclose any intended use of generative AI, with written approval needed for certain uses, including generative AI material in the final product.<\/p>\n

\n\tAway from those shows, Beirnaert\u2019s team is testing AI on a series it is greenlighting as a conventional CG show, with a shorter YouTube version under consideration. A team of around 20 spent roughly three months on the traditional pilot. Using AI, one person recreated that work, covering everything from modeling and layout to animation, within days. The animation fell short of the original, Beirnaert said, but the backgrounds came out better.<\/p>\n

\n\t\u201cFor me, AI is not meant to replace animators. The day every animator is replaced, I will change job and do something else,\u201d Beirnaert said.<\/p>\n

\n\tThe two diverged when an audience member asked how much money AI could save. Beirnaert offered what he called a random estimate of \u20ac1 million on an \u20ac8 million budget, and said AI may help projects that stall at 80%\u201390% financed find the rest. Yotsukura said OLM is in a different position: it is brought in only after a budget is fixed, and sees AI as a way to gain more creative freedom within that budget rather than to shrink it. A year or two ago, the question was how many staff AI could replace, he said. Now it is shifting toward new shots that AI tools made possible.<\/p>\n

\n\tModerator Kao Chieh, a lecturer at Shih Chien University, put the split down to a difference between Eastern and Western animation companies.<\/p>\n

\n\tJapanese anime already has creativity in abundance, Yotsukura said, but technology could play a bigger part, with the two driving the industry forward like a pair of wheels. \u201cI think \u2018not using AI\u2019 is probably no longer an option from here on,\u201d he said.<\/p>\n","protected":false},"excerpt":{"rendered":"OLM, the Japanese animation group behind the \u201cPokemon\u201d anime series and films, has developed an AI coloring tool…\n","protected":false},"author":2,"featured_media":678359,"comment_status":"","ping_status":"","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[20],"tags":[220,218,219,61,60,40571,286542,80],"class_list":["post-678358","post","type-post","status-publish","format-standard","has-post-thumbnail","category-artificial-intelligence","tag-ai","tag-artificial-intelligence","tag-artificialintelligence","tag-ie","tag-ireland","tag-pokemon","tag-taiwan-creative-content-fest","tag-technology"],"_links":{"self":[{"href":"https:\/\/www.newsbeep.com\/ie\/wp-json\/wp\/v2\/posts\/678358","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.newsbeep.com\/ie\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.newsbeep.com\/ie\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.newsbeep.com\/ie\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/www.newsbeep.com\/ie\/wp-json\/wp\/v2\/comments?post=678358"}],"version-history":[{"count":0,"href":"https:\/\/www.newsbeep.com\/ie\/wp-json\/wp\/v2\/posts\/678358\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.newsbeep.com\/ie\/wp-json\/wp\/v2\/media\/678359"}],"wp:attachment":[{"href":"https:\/\/www.newsbeep.com\/ie\/wp-json\/wp\/v2\/media?parent=678358"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.newsbeep.com\/ie\/wp-json\/wp\/v2\/categories?post=678358"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.newsbeep.com\/ie\/wp-json\/wp\/v2\/tags?post=678358"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}