Rafomac News, Tech & Trends That Matter
What if you could turn a childhood passion and a bold vision into a full cinematic commercial—without a massive budget or fancy studio? Using AI tools like Claude and Higgsfield, this creator built a blockbuster-style football ad with robots from scratch, showing how accessible high-end filmmaking has become.
Turning a Football Dream Into AI-Powered Reality
Watching football ads as a kid felt like witnessing magic—slick, cinematic, and hugely ambitious. The creator behind this AI project was convinced it took a massive budget and a real studio to pull something like that off. But, it turns out, all you really need is the right set of AI tools—and the patience to iterate.
This project started with a twist: football, yes, but mixed with a love for old-school cartoons where robots come to life. That led to a bold concept—robots playing football, bathed in a cinematic style, all built from AI-generated assets and scenes.
Building the Foundations: Characters, Locations, and Props
First up was crafting the assets. Using Higgsfield’s Colab environment, all images and video models were kept in one easy-to-share place. The product—a soda can central to the ad—was transformed into a detailed 3D product sheet showing front, back, and top views to ensure consistency across scenes.
Creating the main character meant generating a virtual avatar modeled on the creator’s own face. The AI model named Soul Cinema analyzed 20 images of the creator taken from different angles and in various lightings to lock in the likeness flawlessly. Then came outfit options—forty batches worth—to find the perfect casual look that would contrast with the looming futuristic robot armor.
Next, the location had to feel real and cinematic. Soul Cinema produced an empty New York street bathed in bright sunlight and shot through an anamorphic lens to deliver that wide, cinematic feel. Multiple versions were tested until one nailed the vibe of a tall business district with wide streets and just the right film grain to keep it believable.
Using Claude to Write Perfect Prompts and Avoid Wasted Credits
Writing prompts is an art form here. Instead of spending time fiddling endlessly with AI input, the creator leveraged Claude, an AI that turned simple descriptions into highly structured, precise prompts. For instance, for robots—one meant to be the protagonist and the other the rival—Claude prepared detailed character sheets specifying colors, armor style, and emblem numbers. Soul Cinema then generated a wide variety of options, letting the creator lock in on the designs that meshed with the world’s aesthetic.
Props like the football and the goal went through the same treatment. Claude drafted futuristic prompts, while Soul Cinema generated metallic footballs with red accents and glowing, electric-infused goals. Consistency was key; these assets had to stay unchanged throughout every scene to maintain the illusion.
How the AI-Driven Scene Workflow Works
With characters, locations, and props nailed down, the next phase was about bringing everything together. The creator used a special skill called Cloth Scale to manage prompts and maintain visual style across multiple shots. Each scene’s prompt included headers, lighting setups, lens choice, and instructions for natural camera movements like handheld shots or Dutch angles—adding to the cinematic feel.
The first scene—our hero walking down the empty city street before discovering the floating soda can—was generated and tested with multiple batches. This flexible approach gave plenty of footage to choose from, enabling the creator to assemble the most cinematic version by cutting together the best seconds.
Iterating Like a Director, Not a Prompt Engineer
Many generations didn’t work at first. Smiles were broken, movements awkward, or physics unrealistic—like a robot dropping without weight or armor sliding off unnaturally. Instead of manually rewriting prompts, this creator directed the AI as if guiding actors and cinematographers: specifying camera angles, movement speed, and emotional beats.
For example, when robots failed to show competitive tension, the creator broke down every shot, from close-ups of legs sprinting to dynamic, shaky low-angle shots to capture the real sports energy. Adjustments like these turned flat animations into riveting, dramatic moments.
Scheming the Location to Lock Down Object Placement
One tricky part was ensuring the football goals stayed anchored in the same spot throughout the video. Early attempts saw goals jumping around or turning into full-screen shots unexpectedly. The solution? Creating a visual location scheme using GPT Image 2 with drawings marking exact goal positions on the street image. This map guided the AI precisely, solving the problem better than elaborate verbal descriptions could.
From Defeat to Comeback: The Story Unfolds
The ad follows a classic sports narrative; our robot hero loses his first round, his armor breaking as he falls, but then recharges with the soda’s power, transforming back to full strength mid-run. Every scene was generated with attention to these emotional cues, even dividing complex scenes into smaller parts so each moment had room to breathe and the physics felt grounded.
Technically, some once-off elements like a flashing watch were left to the model’s creativity rather than being separately created assets—saving time without losing quality.
Final Touches: The Action-Packed Match and the Comedic Close
The AI produced fast-paced, chaotic football battles between the two robots, complete with feints, nutmegs, and slide tackles sparking on the legs. The creator creatively reused unused phases from earlier renders, mixing them into the comebacks and victories, building flow without generating everything anew.
To cap it off, the ad added a funny visual gag featuring the victor robot standing comically alone on the empty street—humanizing the robotic star with a bit of cringe-worthy charm. Claude even pitched the joke lines, freeing the creator from trying to be funny.
The Pack Shot That Sells the Product
The grand finale—the product pack shot—was carefully crafted. The soda can falls from the sky, cracking the asphalt with chunks flying, before the brand name “Top Up” assembles from locking metal pieces with an electrifying lightning strike. Iterative notes ensured the camera movement matched the entire video’s vibe and that the lighting remained consistent.
Why This Workflow Will Change How Commercials Are Made
This project reveals a striking truth: the final cinematic commercial isn’t made from perfect first takes but from the best snippet stitched from a hundred AI-generated tries. The workflow leans heavily on iteration, directing the AI like a film director rather than a coder. With all assets, prompts, and visual references in one place, the system can produce stunning, cohesive ads without traditional film studio resources.
Whether swapping out the soda can for any product or replacing robots with other characters, the process remains solid: craft assets, prepare refined prompts with Claude, generate scenes with Soul Cinema and Seedance, and assemble the best moments. This AI-powered method democratizes cinematic ad production, making it accessible to anyone willing to learn the craft and embrace iteration.
