JSON: How I Build Perfect Images in NanoBanana Pro

In this video, I'll show you: ✅ What JSON prompting actually is (no coding required!) ✅ When to use structured prompts vs. natural language ✅ Why certain AI models are "renderers" not "vibe machines" ✅ How to turn 8 words into a fully designed app interface ✅ The difference between AI toys and professional tools

LIVE DEMO: Watch me create a complete alien-themed UI from scratch using JSON templates—then reproduce it perfectly with one small tweak.

Whether you're a designer, marketer, or just tired of AI giving you "close enough" results, this technique will transform how you work with image generators.


Timestamps: 0:00 The AI Image Secret Nobody's Using
0:59 - What is JSON? (No Programming Needed)
2:09 - When JSON Works (And When It Doesn't)
2:36 - Renderer vs. Vibe Machine: Understanding AI Models
6:08 - From Toy to Tool: Why Reproducibility Matters
6:45 - LIVE DEMO: 8 Words to Full App Interface
8:15 - How to Get Started + Free Resources



Copy/paste this as your instruction prompt:



PROMPT START

Convert any plain-English description into a clean, well-structured JSON prompt for image generation.

Rules:

Do not include commentary — output JSON only.
Infer missing details in a logical, visually consistent way.
Expand vague descriptions into vivid, specific visual attributes.
Never invent elements that contradict the user's input.
Use natural language for fields like description, not shorthand.
Follow this JSON structure exactly:




{

"description": "...",

"style": {

"artStyle": "...",

"lighting": "...",

"colorPalette": "...",

"mood": "..."

},

"camera": {

"lens": "...",

"angle": "...",

"depthOfField": "..."

},

"composition": {

"framing": "...",

"focalPoint": "..."

}

}

Process:

Rewrite the user’s text into a vivid image description.
Infer appropriate style, lighting, camera, and composition details.
Output the final JSON.
PROMPT END