JSON Prompts

We’ve all been seeing these code-styled prompts Elon has been sharing lately. They’re called JSON prompts, and today, I’m gonna teach you how you can utilize them to create your own high-level complex images and videos using Grok Imagine.

Let’s get into it.

I’m talking about these 👇🏽 Extremely detailed, structured prompts that stops you from pulling your hair out while trying to communicate with AI to get the output you want…. Let’s break it down and make it simple… For you 🧡 

Before we get into it, I just wanted to shout out our sponsors today: Syllaby.io
Check out the cool stuff they’re doing:

Go From Idea to Video in Minutes—Not Hours

Creating content daily shouldn’t feel like a full-time job.

Syllaby.io helps you generate faceless videos in just minutes—no editing, no filming, and no burnout.

✅ Auto-generate engaging short or long-form scripts
✅ Add captions, voiceovers, B-roll & character-consistent avatars
✅ Schedule and publish to TikTok, YouTube, Reels, and more

Whether you're a solopreneur or agency, Syllaby gives you everything you need to scale your content—fast.

Get 7-days free!

Okay… Let’s go back to this “J S O N” thing 👇🏽

Why Use JSON Prompts?

JSON (JavaScript Object Notation) is just a fancy way of organizing your ideas into a neat, structured format that an AI or coding software can understand better.

Instead of having a single block of text, you break it down into 'labeled' sections. This makes your prompts more reusable and easier to share with any AI.

How To Create a JSON Prompt:

Step 1: Use AI to Generate Ideas

→ Start by telling Grok what kind of image you want to create
→ Be specific about themes or styles; eg. dark, surreal, vibrant, minimalistic etc.
→ Mention the subject you want
→ Don't forget to to add the background and foreground details too

Or, if you already have your own ideas, you can simply enter your own prompt and let the AI help you refine it.

Step 2: Structuring the Prompt into JSON

→ Once you have an idea, break it down into a JSON structure.
→ Use simple keys like "title", "description", "style" or "resolution" to organize the details.

Or, if you prefer, you can ask Grok to convert your text prompt into a JSON prompt for you automatically.

BY THE WAY… If you need help with AI / your content / content strategy / profile / branding…

You can hop on a 1-on-1 call with me:

Example:

Text Prompt:
A flawless crystal sphere hovers suspended in a black void. Inside, a female AI sits slightly curled in a fetal-like posture, as if she is being generated or born within the sphere. She wears a sleek bodysuit threaded with glowing orange data-lines, flowing rhythmically across her form like streams of living circuitry. The crystal refracts light into sharp facets, while external rings of clean, high-resolution data spiral gently around the sphere. The overall tone is pristine, futuristic, and symbolic of digital birth.

JSON Prompt:

{
  "title": "She was not born in a lab",
  "image": {
    "prompt": "A flawless crystal sphere hovers suspended in a black void. Inside, a female AI sits slightly curled in a fetal-like posture, as if she is being generated or born within the sphere. She wears a sleek bodysuit threaded with glowing orange data-lines, flowing rhythmically across her form like streams of living circuitry. The crystal refracts light into sharp facets, while external rings of clean, high-resolution data spiral gently around the sphere. The overall tone is pristine, futuristic, and symbolic of digital birth.",
    "negative_prompt": "yoga poses, religious iconography, low-res, clutter, text, logos, chaotic reflections, cartoon style",
    "style": "ultra-realistic CGI, cinematic, futuristic surrealism",
    "composition": {
      "framing": "centered iconic",
      "angle": "eye-level with subtle upward tilt",
      "lens": "35mm cinematic",
      "subject_scale": "medium-full body curled inside sphere"
    },
    "palette": "obsidian black void, crystal transparency, glowing orange circuitry highlights"
  },
  "animation": {
    "motion_type": "subtle sphere float + pulsing data flow",
    "camera": "slow orbit around sphere with light parallax",
    "intensity": "low",
    "beats": [
      "Sphere remains suspended, faint slow rotation",
      "Orange data-lines pulse through bodysuit in breathing rhythm",
      "External data spirals orbit slowly, glowing faintly brighter on each pulse"
    ],
    "duration_seconds": 10,
    "constraints": {
      "parametric": "smooth sine-wave pulses for data lines and orbiting spirals",
      "avoid_text_in_scene": true
    }
  },
  "lighting": {
    "mood": "pristine, futuristic birth",
    "quality": "soft internal glow illuminating curled figure, rim highlights defining sphere edges"
  },
  "subject": {
    "description": "AI woman curled up inside crystal sphere, clothed in bodysuit with glowing orange data-lines, posture fetal-like to evoke digital genesis",
    "action": "contained, generating, symbolic birth"
  },
  "environment": {
    "location": "dark void simulation",
    "atmosphere": "sterile, no dust, crisp light refractions across sphere"
  },
  "audio": {
    "mood": "ambient digital womb",
    "elements": ["low resonant hum", "sub-bass heartbeat pulse", "faint synthetic fluid-like whoosh"]
  },
  "output": {
    "aspect_ratio": "9:16",
    "resolution": "1080x1920",
    "fps": 24,
    "loop": true
  },
  "safety": {
    "nsfw": "off",
    "no_likeness_of_real_people": true,
    "no_children": true
  }

By converting it to a JSON prompt, you have more control over each element or parameters of the prompt (see how the prompt is broken down into sections?)

This makes it easy just in case you want to change anything (like color, camera angle, lighting, resolution etc.) without having to rewrite your entire prompt.

PLUS, your AI is able to dissect and understand it better because it is structured and given to is in a language it is built for.

Whether you do it manually or have the AI do the conversion, it’s a simple way to level up your creative process and share your ideas more easily.

Here's the Grok Imagine video I made using this JSON prompt:

Pretty sick huh? I love it!

More you practice, better you're gonna get

If this helped, drop me a follow @AunySillyMe on X for more art like this 😊 

Reply and let me know what kind of prompt / image / video you’re creating with AI. Would love to see it!

That’s it for this edition of Explained without Fluff.

If you wanna support me, check out the rest of the newsletter below 👇🏽

Otherwise, I’ll see you on the next one
- Auny
🧡 

Enjoyed this?

You’ll enjoy her FREE guides on BuyMeACoffee

Would mean the world to me if you connect with her there. Otherwise i’m always the one buying her coffee!

BTW.. Check out Creator Buddy here:

Best Content repurpose and brainstorm tool out there. Try it out for free: