Creator Stories

Léo Cannone (@leo_cannone) keeps telling his AI films from a kid's point of view: a boy's camcorder summer in Corsica, a teenage idea about giant grandmothers, and now La Nonna Gigante, an Instagram series "filmed" by a boy named Bruno Serra who has his own account.

Vicky Clarke, who releases music as SONAMB (@sonamb__), trains a small neural network on sounds she recorded herself, from the cotton looms at Quarry Bank Mill to glass and electricity, then composes with what the model gives back. Her own notes cover every step.

Alex Naghavi spent more than 17 years as a creative director on brands and digital products, including the 2012 Myspace redesign. With Ezra Li she made Le Drip, a painterly short about a fish in a kitchen sink, built shot by shot in Flora. It took second place at the Reply AI Film Festival in

Mike Bennion (@mikebennion1) spent years directing TV commercials on film. His AI short Kev, a documentary-style portrait of an invented young blind man, started as a Midjourney face, played about 20 traditional festivals, then became a Reply AI Film Festival finalist and a MetaMorph winner.

Jacopo Reale (@jacopo_reale) edited films for other directors for more than a decade. Since late 2024 he has been generating his own footage and cutting it the same way, from a viral AI reel to Love at First Sight, which won the 2025 Reply AI Film Festival in Venice.

Matteo Bernardini spent years on film sets with Christopher Nolan and Peter Greenaway. For his first AI film, a silent dark fairy tale called Brine, he wrote the script himself, shot-listed it like live action, and went through about 1,280 generations to get eight minutes.

WARYAUME began as a game idea in 1991 and is now a ten-volume fantasy saga told through books, films, albums, documentaries and games. Its creator, Christophe Fertault, explains exactly where AI fits in, including the AI-assisted roles that keep six invented languages consistent.

Abandoned Films (@abandonedfilms) is the anonymous YouTube channel that turned "1950s Super Panavision 70" into a genre: Star Wars, The Matrix and Game of Thrones re-cut as Technicolor trailers with a booming announcer. The creator credits every tool in the descriptions, has since made the whole

Albert Barqué-Duran, who performs as Albert.DATA (@albert_data), sits on stage in a 16-channel EEG headset while a model turns his brain signals into sentences and the sentences trigger the visuals. SYNAPTICON won Best of New Media at Runway's 2026 AI Film Festival, and a version of the code is on

The Dor Brothers (@thedorbrothers), the Berlin-founded AI video studio run by Yonatan Dor, harvest their films: fifty generations for one shot, two survivors, then a film look chosen to cover what the model got wrong. Here is the workflow as Dor has described it, from The Drill to the Logan Paul

I gave GPT-6 Astra phone photos of my basement and asked if it could do this in Blender. It installed Blender, built the room, and spent the rest of the day taking my corrections. Here is what it got right, what it faked, and the render bill I can't pay yet.

Dorian & Daniel took years to make one five-and-a-half-minute AI short, and the tools changed three times while they were making it. It won Gold at Runway's 2026 festival, and the pair still refuse to be called AI filmmakers.

Sougwen Chung (@sougwen) has been drawing alongside robotic systems since 2015, trained on two decades of archived mark-making. The V&A collected the model itself. At Art Basel Hong Kong this year the machine was driven by brainwaves.

Robert Gaudette (@RobertGaudetteAI) won the Runway Grand Prix and the Reply AI Film Festival at Venice with an eight-minute film about a Parisian dancer. He got there by discarding technically perfect shots whenever the character moved in a way that wasn't his.

Spotify starts labelling AI Personas this month, and the rule tests the picture rather than the music. We checked the five AI musicians we have profiled to see which side of the line each one lands on.

David Ribble posts as art_from_the_aether — Dublin-based, fog to the horizon, tiny robed figures, enormous skulls.

In 2023 demonflyingfox deliberately used an outdated Midjourney model so the characters would never look real. In 2026 the same videos got remade with tools that no longer allow that choice.

Mickmumpitz builds local AI film pipelines in ComfyUI and Blender — modular movie workflows, consistent-character LoRA training, a panorama-to-3D-world node pack — and ships the workflow file with every video.

goatcigs makes sci-fi, horror and comedy shorts on Instagram, and keeps a hard line between the AI work and the art they make by hand.

xstonedape posts gods arriving on enormous animals — Saraswati on a swan, Varuna on a makara, Poseidon on a hippocampus. The captions are titles only. No tools, no process, no explanation.

minuit.art.ia lists the exact tool stack under every fantasy reel they post — Midjourney, GPT Image, Seedance, Suno, CapCut. Over 1,300 posts, the receipts are the process notes.

forbiddendimension posts AI horror as recovered camera footage — a GoPro in the tree roots, a bull on a crane, something in the river. Here's how the reels get built, in the creator's own words.

ANIMUSE TV animates Klimt, Bouguereau and Parmigianino into short films — and writes the artist, date, movement and collection into every caption, with the AI labelled.

edward_dreamcast makes AI reels that look like an 80s creature workshop — jewelled monsters, a Star Witch, a dog called Buckley — and ends every caption by asking the reader what happens next.

Thought Palace posts seven-second animations in red, blue and green at full blast, and every caption ends the same way: the title, then the track ID for a song the same person wrote.