Generate catalog concept PNGs with OpenAI gpt-image-1 and CSV manifests

Description

Quick Overview
This workflow runs manually to turn a small product catalog JSON, approved scene descriptions, and target aspect ratios into a validated matrix of product-image creative briefs, then exports the briefs as a downloadable CSV for review and handoff.

How it works
Starts when you run the workflow manually.
Loads your brand style, aspect ratios, product catalog JSON (SKU, subject, must-keep details, optional reference asset), and approved scene JSON.
Validates the inputs (required fields, unique SKUs, supported aspect ratios, and a maximum of 100 brief combinations).
Builds a creative brief for every product × scene × aspect ratio combination, including a standardized prompt, review checklist, status, and a unique brief ID.
Converts the brief matrix into a CSV (with spreadsheet-formula injection protection) and outputs it as a downloadable file.

Setup
Replace the sample product JSON with your own items, ensuring each product includes a unique sku, a subject description, and mustKeep constraints (and optionally referenceAsset).
Add or edit the approved scenes JSON (each needs a name and setting) and set brandStyle and aspectRatios to the formats you plan to produce.
Run the workflow and download the CSV file from the final step’s binary output; no credentials or external service configuration is required.