- Ai For Real Life
- Posts
- The JSON Prompt Pack Everyone’s Asking For
The JSON Prompt Pack Everyone’s Asking For
Inside Veo 3’s JSON mode, Gemini’s advice, and the prompt pack that unlocked it all.
In partnership with
🎬 The Prompt That Changed Everything
Veo 3 JSON Mode Is a Whole New Level of Filmmaking
So here’s what happened.
I’d been seeing people online talking about Veo 3’s JSON prompt mode, how you could call out exact camera movements, pick your shot types, and actually direct a scene with just structured text. It sounded wild… but I wasn’t sure if it was real or just hype.
Then I tried it.
And the level of control I got back? Changed everything.
I'm talking cinematic-level shots, pans, push-ins, locked-off wides, character tracking. Done in seconds. And not in a soft, "AI guess what you mean" kind of way. Veo actually executed the direction, just like a camera crew working off a shot list.
It’s like film school in JSON.
If you’ve ever played with natural language prompts, you know they’re fun. They give you beautiful surprises. But when you switch to structured JSON, every frame becomes intentional. You’re no longer just playing with style. You’re shaping a story.
🧠 My Take on JSON vs Natural Prompts
Now here’s the truth from my side…
JSON is incredibly powerful, especially when you need precision. The kind of work you see in high-end AI ads with IKEA boxes exploding in sync, smooth camera rolls, all that technical choreography? JSON is where that kind of stuff shines.
That’s the side of JSON that I’m sold on.
But when it comes to cinematic storytelling, those emotionally charged shots, subtle push-ins, and rule-of-thirds framings, I’m still testing. I’ve gotten some promising results, but I haven’t seen enough from the community (or even my own work yet) to fully say JSON is there creatively.
So I’m in the lab, working on it. I’ll keep you posted.
And I’m also not throwing natural prompts out the window. I still use them, especially when I want the AI to bring something unexpected to the table. If you want creativity, happy accidents, or you're trying to discover a visual tone you don’t already have in your head, natural language can actually surprise you in the best ways.
Both have their place.
But JSON? That’s the move when you’re ready to direct.
💬 Gemini’s Full Response on Structured Prompts
So I asked Gemini, Google's own model, what works best in Veo 3: structured JSON or natural prompts?
Here’s the full, unedited response:
Structured prompting, including JSON, is demonstrably more effective and is a core part of how models like Veo are intended to be used for advanced applications. This isn’t a “trick” or a “hack” it’s a natural evolution of prompt engineering.
Models Are Trained on Structured Data
I’m trained on structured formats like JSON, XML, YAML. When I see JSON, I don’t interpret it like text. I read it as key-value logic with clear hierarchy and intent. That makes it easy to execute with precision.
Official Documentation Encourages Structure
Google’s developer docs (including Veo’s) recommend JSON-style key-value inputs — like aspectRatio, cameraMotion, negativePrompt, seed. This is how you speak directly to the model.
Gemini as a Prompt Generator
Most creators now use Gemini to turn natural language into JSON. That synergy is intentional — Gemini builds the instruction, Veo executes it.
Precision vs Creativity
Natural language is great for playful results. But when you know what you want a tracking shot at golden hour, a wide dolly-out with rain JSON is the way. It’s how creators move from exploration to execution.
Bottom line: Veo 3 was built to understand structure. The community didn’t stumble on a cheat code. We finally unlocked the user manual.
And if you want to play around with the exact JSON prompts I used, HERE THEY ARE
|
Big News: GPT‑5 Has Arrived
Today, OpenAI released GPT‑5—the smartest, fastest, and most reliable ChatGPT yet.
Available to all ChatGPT users—free, Plus, Pro—with tiered usage limits.
CEO Sam Altman called it a “PhD-level expert” that “rethinks responses when needed,” using a smart router between fast and deep reasoning modes.
Performance is strong across the board—especially in coding, writing, health, and complex reasoning tasks. It even adjusts style for reduced flattery (“sycophancy”) and handles multi-step prompts with more truth and fewer hallucinations.
Why this matters for us creators: GPT‑5 can generate more polished scripts, help debug code for your AI workflows, and elevate your prompt engineering across the boarMore on that soon.
Just wanted you to be the first to know.
✨ Other Highlights This Week
🧠 Consistent Characters, Cinematic Results
Still struggling to keep your AI character’s face consistent across different scenes?
I’ve been testing a new combo: Ideogram for facial consistency, and HiggsField Soul for cinematic quality. It’s been the most stable workflow I’ve found so far — especially when working from just one photo.
🎶 ElevenLabs Just Launched AI Music — And It’s Legit
They dropped Eleven Music, a full-on AI song generator.
You can create studio-grade tracks with vocals, isolate instruments, control BPM and key, and even time drops and voiceovers to match your visuals. It’s built for creators and ready for commercial use.
Closing Note from Me
We’re in a moment where tools like Veo’s JSON mode and GPT‑5 aren’t just tools—they’re creative collaborators. They’re letting us push past solo experiments into intentional, cinematic storytelling.
You’re not just generating clips anymore. You’re directing scenes, building stories, leveling up your content workflow.
I’m here experimenting and sharing in real time. Let’s build this wave together.
Best,
Khalil
🙌 Support the Project
One of the best ways to support me and everything I’m building here is simple:
Click on our sponsors below and check out their newsletter.
Every click helps this project grow and keeps the tools and tips coming your way. Appreciate you!
Your career will thank you.
Over 4 million professionals start their day with Morning Brew—because business news doesn’t have to be boring.
Each daily email breaks down the biggest stories in business, tech, and finance with clarity, wit, and relevance—so you're not just informed, you're actually interested.
Whether you’re leading meetings or just trying to keep up, Morning Brew helps you talk the talk without digging through social media or jargon-packed articles. And odds are, it’s already sitting in your coworker’s inbox—so you’ll have plenty to chat about.
It’s 100% free and takes less than 15 seconds to sign up, so try it today and see how Morning Brew is transforming business media for the better.