Rafomac News, Tech & Trends That Matter
A fully AI-generated, 90-minute feature film is heading to Cannes—created in just 14 days with a $500,000 budget. Behind the scenes, a team of 15 discovered 28 essential hacks that will change how you make AI videos forever.
How a Cannes AI Feature Pushed Boundaries in Just Two Weeks
Imagine producing a full-length, 80-minute AI-generated film in less than a month with a tight team and an unclear outcome. This was the challenge facing a dedicated group of 15 creators armed with 10 million AI credits and 14 days on the clock. By day eight, their first rough draft was 110 minutes—far too long and full of scenes that needed heavy fixing.
They tackled over 40 problematic scenes, but the turnaround was brutal. Iteration and precision became their mantra. One scene showed plasticky faces drenched in light and erratic positioning. Another had a monster awkwardly holding what looked like a sword—actually his own arm. Yet another mistakenly used an old skin asset, ruining the image. Tackling these issues was just the start of their tight marathon.
From AI Slop to Cinematic Realism: The Fixes That Matter
By day 11, after three intense days of fixes, the changes were striking. The previously artificial lighting was recalibrated to reflect the danger in a scene, spreading cool blue and purple tints that added real mood. Characters’ expressions shifted from robotic to genuinely terrified, and physical positioning became consistent and believable. The monster’s clawed arm was clarified—no more accidental swords. Even props like skateboards were redesigned to feel authentic.
Each fix made these AI-generated frames breathe with life, shifting the film’s feel from a robotic re-enactment toward genuine storytelling. It’s this kind of craftsmanship that sets this project apart.
28 AI Filmmaking Hacks to Save You Months of Work
The team distilled their hard-earned knowledge into 28 definitive hacks, covering everything from prompting and pre-production to teamwork and editing. They shared powerful strategies like providing deep situational context and anatomical detail for emotional scenes—a prompt isn’t just about what happens, but how bodies and faces react minute-by-minute. For example, rather than saying “Roku watches the monster take Lulu,” the improved prompt detailed Roku’s desperate reach, terror, and physical posture in the moment.
Prompting the AI like a photographer also made a big difference. Specifying camera settings, lens types, lighting conditions, and angles improved the cinematic quality drastically. Not just vague words, but exact visual language, aperture stops, and natural light sources created believable scenes.
Crucially, they built a comprehensive system prompt to keep everyone aligned and maintain consistency. This included rules for on-screen text, positioning coordinates, and even ‘safe language’ filters to avoid unwanted results. The system prompt acted as a blueprint across 15 team members, streamlining the process and preventing wasted work.
Mastering Pre-Production and Collaboration in AI Film
Good artwork starts before the AI generates a single frame. The team applied classic cinematography rules, like the 60-30-10 color guide, to balance scenes visually. They avoided front-facing locations to preserve depth, and anchored characters to specific visual points—like a tree or stack of books—to maintain spatial coherence.
Top-down maps were crafted for complex scenes, detailing exactly where every character and prop should be. In action sequences, choreography was beat-by-beat scripted to avoid random or awkward AI movements. Each prop, like the monster’s claws, had detailed ‘prop sheets’ showing multiple angles to prevent confusion in the final render.
Teamwork hacks were just as vital. Splitting into pairs helped catch errors that individuals missed, and a dedicated ‘slop director’ ensured only clean, usable shots made the cut. Centralized asset management on a shared canvas kept elements organized across a large group, preventing costly mistakes and repeated work.
Editing AI Footage: Every Frame Counts
Unlike traditional film, an AI-generated video can have jagged frames or stutters. The team reviewed footage frame-by-frame, deleting duplicates to maintain smooth 24 FPS playback. They salvaged seconds of good footage from failed takes instead of discarding them entirely and used targeted re-generations on problematic segments, knowing they could try as many times as needed until the perfect shot emerged.
This flexibility is a game-changer: where a live-action bad take can mean costly reshoots, AI lets creators simply tweak prompts and regenerate first-rate visuals.
The Cost and Scale of an AI-Created Cannes Film
After 14 exhausting days, they locked the film—not without a jaw-dropping tally. Nearly 109,000 AI video generations churned out 9.5 million credits, costing roughly $400,000 in generation expenses alone. Total project cost, including team and post-production, came in around half a million dollars.
This is a striking 1% of what a live-action 90-minute film with similar visual effects would cost, estimated at $50 million. And unlike traditional filmmaking, the timeline was compressed to two weeks.
Industry veterans like Jama Jurabaev, a concept artist behind blockbusters such as The Mandalorian and Avengers, praised the film’s quality and storytelling. He highlighted how, beyond tools, talent and vision elevate AI-generated content from pretty pictures to compelling cinema.
The Future of AI Filmmaking Starts Here
This project offers a blueprint for filmmakers looking to harness AI’s power. It’s more than just technology; it’s about mastering the art through detailed prompts, smart pre-production, collaborative teamwork, and careful editing. The 28 hacks uncovered are a masterclass in turning AI’s raw potential into seamless storytelling.
If you want to see what happens when cutting-edge tech meets tireless creativity, this Cannes-bound feature reveals just how far AI in film has already come—and how far it still can go.
