# Creative AI Club > People, process and the work shaping creative practice with AI. Creative AI Club is a publication on the people, work and process shaping creative AI. Numbered editions go behind one creative's or studio's work: where the idea started, where AI entered the process, and where human taste, judgment and craft still mattered. Between editions, the Feed tracks new models, new work and production shifts, filtered through one question: what does this mean for people making creative work? Full text of every published edition and Feed story: https://creativeaiclub.co/llms-full.txt. RSS: https://creativeaiclub.co/feed.xml. Sitemap: https://creativeaiclub.co/sitemap.xml. ## Editions - [Years in the making](https://creativeaiclub.co/years-in-the-making): Special edition 001. How experienced creatives are bringing a lifetime of craft into a new era of possibility. ## The Feed - [An AI short just won a $450,000 festival Grand Prix](https://creativeaiclub.co/feed/astana-ai-film-festival-winners): 09.10.2026. Nikolay Shestak's The First Honest Day topped the first Astana AI Film Festival, picked from 8,067 entries. - [Ask Claude to invent the notation first, then write the music](https://creativeaiclub.co/feed/simon-willison-claude-game-music-format): 09.10.2026. Simon Willison had Opus 5.5 design a text music format, build a player and write game tunes in it. He calls them surprisingly good. - [AI brings back Houdini. People still wrote every word.](https://creativeaiclub.co/feed/houdini-ai-stand-ins-red-rock-films): 09.10.2026. Red Rock Films built Fox Nation's Houdini series from 400 licensed photos, human stand-ins and scripts drawn from his own diaries and letters. - [anabology open-sources the character bible behind his Claude music videos](https://creativeaiclub.co/feed/anabology-claudia-open-character-kit): 09.10.2026. claudia.gallery hands out 1,562 Midjourney images, three films, a wiki, an agent skill and an MCP server, so people and agents can make Claudia themselves. - [Google ships Nano Banana 2.1 with steadier characters and panoramic frames](https://creativeaiclub.co/feed/google-nano-banana-2-1-image-model): 09.10.2026. The image model update holds faces across multi-turn edits, renders text more cleanly and cleans up 8:1 and 1:8 panoramic frames at 2K and 4K, at about half the old price per image at 1K and 2K. - [Brothers and Sisters direct a generated boxing film like a normal shoot](https://creativeaiclub.co/feed/brothers-sisters-generated-boxing-shoot): 08.10.2026. Everything Is Computer casts, grades and cuts a 1970s boxing story for Behind The Gloves with no camera day. - [Credle, Rolfe and Jenkins film themselves making AI spots for small briefs](https://creativeaiclub.co/feed/google-small-brief-credle-rolfe-jenkins): 08.10.2026. Credle, Rolfe and Jenkins make Flow+Gemini spots for small businesses and get filmed doing the iterations. - [Pablo Diaz: AI wins hardest in the treatment, not the performance](https://creativeaiclub.co/feed/pablo-diaz-somos5-ai-spy-hybrid): 08.10.2026. SOMOS5's Pablo Diaz tells LBB the biggest AI win is pre-production treatments and continuity, not AI acting. - [FeviKwik turns AI into a mashup playground, not a tech demo](https://creativeaiclub.co/feed/fevikwik-chutki-mein-kalakaari-ai): 08.10.2026. FeviKwik's KwikGPT playground lets people mash two objects into named creations, past a million entries. - [OneDay revives an unproduceable 85-shot mountain script with Luma](https://creativeaiclub.co/feed/oneday-adrenaline-junkies-luma-dream-brief): 08.10.2026. OneDay turns a shelved unproduceable mountain script into an 85-shot Dream Brief film with licensed actors and full post. - [OpenAI's But Now puts the ChatGPT prompt on screen](https://creativeaiclub.co/feed/openai-but-now-special-biscuit): 08.10.2026. Special US and Biscuit's six But Now films put ChatGPT prompts on screen across 29 markets. - [Fanta writes horror lore for 18 months, then puts AI to work](https://creativeaiclub.co/feed/fanta-haunted-universe-grail-ai): 02.10.2026. Fanta's 50-market Haunted Universe spent 18 months on human lore before GRAiL's AI production. - [Asahi's Infinite Can is a bespoke craft-at-scale AI system](https://creativeaiclub.co/feed/asahi-infinite-can-pop-vermeer): 02.10.2026. POP's Vermeer Fusion Designer turns Hisham's language into hundreds of on-brand Asahi cans with human review gates. - [DuckDuckGo launches Duck.ai with giant prying eyeballs](https://creativeaiclub.co/feed/duckduckgo-duck-ai-rival-prying-eyes): 02.10.2026. Rival and Partizan launch Duck.ai with three :30s of giant prying eyeballs. - [Matt Lenski drives AI characters from live-action performances](https://creativeaiclub.co/feed/matt-lenski-last-commercial-beep-boop): 02.10.2026. Arts & Sciences' Matt Lenski builds The Last Commercial so real performances drive the AI characters. - [Leo Singapore builds F1 worlds Standard Chartered could not shoot](https://creativeaiclub.co/feed/standard-chartered-leo-f1-ai-hybrid): 02.10.2026. Leo Singapore and Stink use photography plus AI environments for Standard Chartered's F1 brand film. - [Sonnet 5.5 brings a sharp eye for design into the cheap seat](https://creativeaiclub.co/feed/claude-sonnet-5-5-design-eye): 30.09.2026. Sonnet 5.5 runs 30%+ faster, costs up to 30% less per task, and Anthropic flags a sharp eye for design. - [Gemini 3.8 TTS lets you design the voice, then direct the scene](https://creativeaiclub.co/feed/gemini-38-tts-scene-direction): 30.09.2026. Gemini 3.8 Flash and Flash-Lite TTS generate custom voices and directed dialogue across Google surfaces. - [Code-to-video craft goes mainstream in the agent desk](https://creativeaiclub.co/feed/code-to-video-agent-craft-class): 30.09.2026. Agent-written HTML and canvas motion, captured in-browser and stitched with FFmpeg, is becoming a real pipeline. - [Kling 4.0 Flash opens early while the full model waits for October](https://creativeaiclub.co/feed/kling-4-flash-early-access): 30.09.2026. Kling 4.0 is aimed at October; Flash early access started 28 September with longer takes and keyframe control. - [Eleven v4 makes delivery a direction pass, not a reroll](https://creativeaiclub.co/feed/eleven-v4-emotive-direction): 30.09.2026. Eleven v4 interprets tone, pacing and context, with inline tags for how a line should land. - [Firefly removes video backgrounds without frame-by-frame masks](https://creativeaiclub.co/feed/adobe-firefly-video-bg-remove): 30.09.2026. Firefly can strip backgrounds across an entire clip, with longer prompts and wider GPT Image 2 ratios. - [GPT-6 Sol and Luna cut the price of everyday agent craft](https://creativeaiclub.co/feed/gpt6-sol-luna-cheaper-craft): 30.09.2026. GPT-6 Sol and Luna arrive as cheaper Codex/Work models; Sol roughly 50% cheaper than prior mid-tier pricing per Ben's Bites. - [D&AD Black Pencils reward craft that punches slop](https://creativeaiclub.co/feed/dad-black-pencils-anti-slop-craft): 30.09.2026. Uncommon's Periodic Fable for The Ordinary wins Art Direction by calling out empty beauty jargon. - [Runway sits under the DaVinci Resolve playhead](https://creativeaiclub.co/feed/runway-inside-davinci-resolve): 30.09.2026. Generate and restyle inside Resolve Studio without leaving the edit. - [AiCandy ranks #2 after just one year](https://creativeaiclub.co/feed/aicandy-the-work): 19.09.2026. An AI-native production company has gone from launch to one of Australasia’s highest-ranked production companies in roughly twelve months. - [Be Brave rebuilds a life with AI](https://creativeaiclub.co/feed/be-brave-renzo-rosso): 19.09.2026. The documentary about Diesel founder Renzo Rosso uses AI not as an effect, but as part of the storytelling language itself. - [Infinite Slop is an endless AI stream built by the audience](https://creativeaiclub.co/feed/infinite-slop-is-an-endless-ai-stream-built-by-the-audience): 19.09.2026. A never-ending AI livestream where the audience writes the next scene is a glimpse at what happens when video generation becomes faster than playback. - [Seedance 2.5 pushes AI video toward sequence direction](https://creativeaiclub.co/feed/seedance-2-5-pushes-ai-video-toward-sequence-direction): 19.09.2026. Longer shots, stronger reference control and much more precise editing are pushing AI video closer to a real directing workflow. - [Gemini Omni moves video editing into conversation](https://creativeaiclub.co/feed/gemini-omni-moves-video-editing-into-conversation): 19.09.2026. Video editing is starting to move from timelines and prompts into conversation. - [H&M and Nick Knight turned the runway into a live image system](https://creativeaiclub.co/feed/hm-nick-knight): 19.09.2026. Photography, illustration, movement and AI came together in a fashion show where the imagery responded to the people inside it. - [Grenade faked London stunts with AI](https://creativeaiclub.co/feed/grenade-creme-egg): 19.09.2026. The Creme Egg collaboration used synthetic spectacle and deliberately rough social content to make people wonder whether the stunts had actually happened. - [MiniMax launches H3 for faster, more controllable AI video](https://creativeaiclub.co/feed/minimax-launches-h3-for-faster-more-controllable-ai-video): 19.09.2026. Native sound, stronger reference control and much faster iteration are pushing AI video closer to something creatives can actually direct. - [GPT-Live opens realtime voice as a creative surface](https://creativeaiclub.co/feed/gpt-live-opens-realtime-voice-as-a-creative-surface): 18.09.2026. Realtime voice creates new ways to make interactive campaigns — and a new way to develop ideas out loud. - [MIRAI is a virtual human built to keep showing up](https://creativeaiclub.co/feed/mirai-is-a-virtual-human-built-to-keep-showing-up): 16.09.2026. AI characters are starting to move from campaign assets to ongoing brand presence. - [Mike Teevee released Brand Perspectives 2026](https://creativeaiclub.co/feed/mike-teevee-released-brand-perspectives-2026): 15.09.2026. Senior creative and marketing leaders on how AI is changing the way ideas are developed, tested and produced. - [FLUX.2 Video makes generated video easier to art direct](https://creativeaiclub.co/feed/flux-2-video-makes-generated-video-easier-to-art-direct): 14.09.2026. Stronger control over start frame, end frame and motion is pushing generated video closer to something you can direct. - [ChatGPT Images 2.5 makes image gen feel more like directing](https://creativeaiclub.co/feed/chatgpt-images-2-5-makes-image-gen-feel-more-like-directing): 12.09.2026. Faster generation, stronger editing and better reference control make AI image-making feel much closer to a real creative workflow. ## The Club - [About](https://creativeaiclub.co/about): what the Club is and who it is for - [All editions](https://creativeaiclub.co/editions) - [The Feed](https://creativeaiclub.co/feed) and [archive](https://creativeaiclub.co/feed/archive) - [Subscribe](https://creativeaiclub.co/subscribe): new editions and the best of the Feed by email - [Pitch](https://creativeaiclub.co/pitch): put yourself or your studio forward for an edition - [Instagram](https://www.instagram.com/creativeai.club) · [LinkedIn](https://www.linkedin.com/company/creative-ai-club/)