JSON Prompt Engine

JSON is your
product’s DNA

Plain-text prompts guess. Re-roll and your product changes. A JSON prompt is a structured blueprint of the shot: product, scene, lighting, camera. Change one field and everything else stays anchored to the same product intent.

Reference-firstField-level controlListing-ready
Full guide

Control preview

Change the field, not the product.

anchored
{
"product": "same reference image",
"lighting": "flat" -> "soft studio",
"platform": "Amazon / Shopify",
}

3,174

hosted references

JSON

prompt control

2K

marketplace examples

A one-minute explainer, narrated by NotebookLM.

Why JSON beats plain text

Plain text — guesses

“amber serum bottle on marble, but make the background sage green”

Re-roll and the bottle shape, label and lighting drift too. You fight the model.

JSON — controls
{
  "product": { "item_name": "amber serum", "material": "glass" },
  "environment": {
    "background_style": "marble" → "sage green"
  },
  "studio_lighting": { "mood": "soft, clean" }
}

Change one field. Everything else stays identical.

Reference library

Start from a visual system, not a blank prompt.

ProductJSON Studio includes a reviewed ecommerce reference library and a generated marketplace-format upgrade for main images, clean Shopify shots, lifestyle scenes, exploded views, feature callouts, scale images, box contents, and use-case comparisons. Use the references to choose the image type first, then use JSON to control product identity, lighting, camera, and platform constraints. Open Inspiration to see the 240 generated examples listed first inside their matching ecommerce categories.

See the method in motion

Full control AI image editing with JSON prompting · Design Life with AI

JSON prompts: write templates & edit anything · Design Life with AI

How the platform works

1Inspire / Convert

Start from 3,000+ prompts, or turn any idea into JSON.

2Edit

Tweak 7 plain fields — or the raw JSON.

3Lock product

Save your product so it stays identical.

4Generate

Nano Banana or GPT · A/B · batch.

5Refine

Variations & white-background swap.

Same JSON every time — re-generate, and only what you changed changes.

The DNA of AI imagery

The DNA of AI imagery — mastering JSON prompting for e-commerce: structured visual fields, the rule of one-change, and the extract-modify-generate workflow.
Generated with NotebookLM from the explainer videos.

You don’t need to know JSON.

Convert turns any idea or prompt into JSON. The form shows 7 plain fields (product, background, lighting, shot, avoid…). The JSON assistant edits it in plain English, and the JSON is auto-polished before every generation.

Quick start

Browse first screenshot
1Browse first

Explore the library, then add provider keys only when you create.

Pick or convert screenshot
2Pick or convert

Open an inspiration card, or paste any idea into Convert to get JSON.

Lock your product screenshot
3Lock your product

Save a Product so its shape, colour and label stay identical across scenes.

Generate & compare screenshot
4Generate & compare

Generate, A/B the two models, and batch up to four at once.

Provider setup

Provider setup

Direct API keys run inside the app. MCP providers run through your coding agent.

ProductJSON Studio can call direct APIs from localhost. Magnific and Higgsfield are different: they are agent-assisted provider workflows, so the buyer sets them up in Claude Code or Codex and lets the agent generate files for the Studio workflow.

Use Settings for direct keys

  • Gemini for Nano Banana / NB-Pro style generation
  • OpenAI for GPT Image generation
  • OpenRouter for JSON editing and prompt operations

Use Claude Code or Codex for MCP

  1. 1Install the Magnific or Higgsfield MCP/local tool in Claude Code or Codex.
  2. 2Authenticate the provider in the agent terminal using the buyer account.
  3. 3Ask the agent to generate from the ProductJSON prompt and product references.
  4. 4Import the saved outputs back into ProductJSON Studio as references or final images.

Agent prompt

Use my Magnific or Higgsfield MCP/local provider to generate ecommerce product images from the current ProductJSON Studio JSON prompt. Keep product identity locked to my reference image, save the outputs locally, and tell me which files to import back into ProductJSON Studio.

Amazon and Shopify listings

Listings workflow

Amazon compliance or Shopify product pages, from the same product brief.

The Listings tool works in two phases: create a trusted main image first, then use that approved image as the anchor for every secondary listing shot.

Full workflow
1

Load the product facts

Upload reference photos, then add the brand, product name, listing title, bullets, and accuracy notes.

2

Approve the main image

Check shape, scale, material, colour, label placement, and silhouette before building the rest of the set.

3

Build the listing gallery

Generate the secondary images from the approved main, regenerate weak slots, then download the final set.

Amazon

Compliance-first main image, marketplace-safe secondary shots, and zoom-ready export sizes.

Shopify

Brand-led PDP gallery images with a style field for store mood, colours, and art direction.

Which model should I pick?

Beauty & Personal CareGPT Image 2Biggest win (+14) — texture & label fidelity.
Food & BeverageGPT Image 2Slight edge (+3) on appetising detail.
Fashion & ApparelNano Banana ProTie — NB-Pro for speed & cost.
General ProductNano Banana ProTie — NB-Pro by default.
Text on the imageGPT Image 2Best at legible in-image typography.
Everything elseNano Banana ProFast starting point through your configured NB2 provider.

Use the model-eval loop when a reference style matters: generate or compare with Nano Banana 2 / Pro and GPT Image 2, then keep the model that preserves product identity, composition, text behaviour, and marketplace format most reliably.

Full guide