Merry Christmas!
On this day filled with fireworks and blessings, let me share a tool that can add magic to your holiday content.
Type “2026”, “MERRY XMAS” or “LOVE”, pick a famous landmark, and watch AI paint your words across the night sky in glorious pyrotechnics — no real fireworks, no permits, no cleanup.
The Spark
New Year’s Eve. Sydney Harbour.
You know those legendary midnight fireworks that trend globally every single year?
A friend of mine runs a social media account. Every holiday season is the same headache:
New Year’s posts, Fourth of July content, Diwali celebrations — and what footage do they have? The same three stock fireworks clips recycled across millions of posts.
“I just want fireworks that spell out ‘Happy 2026’,” she told me. “Is that too much to ask?”
It’s not. But until recently, it was impossible.
The gap hit me: Content creators, event planners, brand marketers — everyone wants personalized fireworks visuals that spell out their own text or form custom shapes. But in reality, that’s either impossible or prohibitively expensive. Meanwhile, AI video generation had just gotten really good.
What if you could type any text — “2026”, “MERRY XMAS”, “LOVE U” — pick a famous landmark, and get back a cinematic fireworks show where thousands of glowing particles form your exact words high above the skyline?
What I Built
Fireworks Show generates professional 8-second cinematic videos where AI-controlled fireworks spell out your custom text or form shapes above real-world locations.
Try it: mulerun.com/@LinkAIBrain/fireworks-show
The specs:
| What You Get | The Details |
|---|---|
| Speed | 60-180 seconds, idea to finished video |
| Quality | 1080p HD, cinematic camera movements |
| Duration | 8 seconds of pure spectacle |
| Intelligence | AI auto-selects perfect color palette based on your content |
| Locations | Any landmark or setting you can describe |
A restaurant owner can generate “HAPPY 2026” exploding above the Statue of Liberty for their Times Square party invite.
A wedding planner creates “JASON + EMMA” glittering over Golden Gate Bridge.
A Christmas post with “MERRY XMAS” forming in red and green sparks above the London Eye.
All generated in the time it takes to grab coffee!
Who This Is For
I built this for people who need content that stops thumbs cold — but don’t have Hollywood budgets.
Social media creators — competing for attention in an ocean of sameness, needing that “how did they make THIS?” factor.
Event planners — who want teaser videos for weddings, corporate parties, festival promotions.
Small businesses — running holiday campaigns on bootstrap budgets.
Gift-givers — seeking something more memorable than a greeting card, more affordable than real fireworks.
The common thread? They all need personalized spectacle — the kind of content that used to require either real pyrotechnics or a VFX team.
How I Built It
The architecture leverages Google Veo 3.1’s video generation through MuleRun’s unified API:
What Actually Worked
Three decisions separated success from mediocrity.
Removing choice paralysis. My first version had 6 color palettes, 6 environments, 3 durations, 4 resolutions. Users stared at 12 dropdown menus and abandoned. I stripped it to two inputs: your text/shape, your location. The AI handles everything else intelligently. Conversions tripled.
Embracing the quote syntax. I wrestled with how users specify “I want literal text” versus “I want a shape.” Complex toggle buttons? Mode selection? The answer was elegant simplicity: quotes mean text, no quotes mean shape. "LOVE U" versus heart. Users got it immediately.
Leaning into landmarks. Early versions used generic settings (“lakeside”, “city skyline”). Adding support for specific landmarks — Statue of Liberty, Eiffel Tower, Sydney Opera House — transformed the output from “nice video” to “wait, that’s MY city!” Emotional connection skyrocketed.
The Mistakes I Almost Made
Letting users control too much. I almost shipped with manual color palette selection. Testing revealed users agonized over choices, second-guessed themselves, and produced worse results than the AI’s automatic selection. Sometimes the kindest UX is taking choices away.
Ignoring firework physics. First versions had fireworks exploding at weird heights — sometimes practically at ground level, sometimes floating impossibly high. Adding explicit altitude requirements (“realistic launch trajectory, explosion at proper aerial height”) made output dramatically more believable.
Over-explaining the limitations. 2-6 characters work best for legibility. I initially buried this in fine print. Moving it to the placeholder text (“Keep text short — 2-6 characters work best”) prevented 90% of “why is my 15-character sentence unreadable?” complaints.
Open Source: The Veo 3.1 Video Generation Workflow
I’m releasing the core n8n workflow that powers video generation with Google Veo 3.1 through MuleRun — async polling, API integration, error handling, binary video download. One JSON file. Import it, add your credentials, start building.
This isn’t Fireworks Show itself — it’s the foundation. Swap in your own creative prompts for drone shows, cinematic trailers, product videos, whatever sparks your imagination.
Grab it here:
github.com/LinkAIBrain/mulerun-n8n-examples/…/Veo3_1-Video-Generation.json
Two seconds. Zero cost. 
Star the repo, 
hit the heart on this article — it helps other builders discover these resources, and honestly, it keeps me motivated to open-source more.
Quick Start
-
Download the JSON file
-
Import into n8n (Settings → Import from File)
-
Add your MuleRun API credentials
-
Test with a simple prompt — a landscape scene or basic animation
-
Iterate from there
If you’ve touched n8n before, you’ll be comfortable immediately. If you haven’t, video generation makes a surprisingly satisfying first project.
Why MuleRun Made This Possible
I built Fireworks Show in one day. Not a hackathon sprint. Not cutting corners. One productive day.
That timeline exists because MuleRun absorbed everything that wasn’t the creative core:
| Without MuleRun | With MuleRun |
|---|---|
| Subscribe to Google AI video APIs |  Handled |
| Build async polling infrastructure | Handled |
| Manage API rate limits & errors | Handled |
| Handle payments & billing | Handled |
| User authentication | Handled |
I spent my time on what matters: crafting the perfect prompt structure, testing color combinations, refining the user experience. The infrastructure simply worked.
The Formula
Fireworks Show isn’t breakthrough technology. It’s focused execution on a specific desire.
-
Find an emotional moment — Everyone wants personalized celebration visuals
-
Remove all friction — Two inputs, AI handles the rest
-
Deliver magic — 60-180 seconds to professional-quality results
-
Let infrastructure disappear — MuleRun handles everything that isn’t the product
The pattern is repeatable. Every cultural moment, every holiday, every celebration — there’s a visual experience people dream about but can’t easily create.
Find the moment. Build the bridge. Start shipping. 