ChatGPT Image Automation

Not selected
Not selected
Not selected
Start End
Checked before the run starts. If this page won't open a chat box (deleted or renamed project), the run falls back to a plain chat and says so instead of failing every product.
Sent to ChatGPT once per product together with its reference photo. {title} is replaced with the product's title from your CSV. You rarely need to change this.
Write as much or as little as you like. Everything above the first Image 1 heading is the shared brief and is sent once per product together with Step 1. Each heading line Image 1 … Image 5 starts that image's own instructions, and the five images are generated in that order. (The old NAME :: instruction five-line format still works too.) Click Save to keep your version for next time.
ChatGPT Login Not signed in — click to sign in automatically
Retry ignores the Start/End row range and checks the SKU's folder in your output folder on this computer: images still there are kept, missing ones are regenerated. To redo a bad image, delete its file (e.g. 2.jpg) from that SKU's folder, then Retry - only that image is created again.
Log console ready. Start local server.py and click 'Run Automation' to begin.

Skipped - Missing Source Images 0

Skipped - Already Generated 0

Failed - Click Retry to run again 0