<aside> 🛡️

+10% Revenue Per Visitor in 90 days. Or we refund the engagement. Book a teardown

</aside>

One clean product photo in. A full listing set out: hero on white, lifestyle scenes, macro detail, angle variations. No studio, no retoucher, no week-long turnaround. This page is the exact stack, the setup, and the copy-paste prompts.

For most of what a DTC brand needs on a PDP or an Amazon listing, the shoot math stopped making sense. One photo of the actual product, fed to the right model with the right instruction, gives you a hero, lifestyle frames, and detail shots in the time it takes to brief a photographer.

The economics are the easy part. AI product imagery runs roughly $0.50 to $3 per image against $150 to $1,500 for a traditional shoot, and time-to-market drops from weeks to hours. The reason most operators get mediocre output is they treat it like a text-to-image toy instead of a compositing job. Get that one thing right and this becomes one of the fastest paybacks in the whole AI stack.


The stack

Layer Tool Role
The brain Claude (Desktop or Code), or Codex for parallel runs Interprets "make me a listing set," orchestrates the run, organizes output into folders
The hands (generate) Higgsfield (official hosted MCP) Fires the product-photoshoot skill: hero, lifestyle, detail, angles from one photo
The hands (composite) Nano Banana (Google Gemini image, via the Gemini API) Drops your real product into any scene with label, shape, and color locked
On-model / apparel Botika, Photoroom Virtual Model Flat-lay or ghost-mannequin to on-model across a range of models

Higgsfield ships an official hosted MCP server at https://mcp.higgsfield.ai/mcp plus a CLI. Nano Banana runs through the Gemini API. Photoroom and Botika are API/CLI tools. Wire Higgsfield as the MCP; the rest are API/CLI integrations.


Recipe 1: a full listing set from one product photo (Higgsfield + Codex)

Drop in one clean product photo, get back an organized folder of on-brand listing images. No studio. No retoucher.

You set the spread (how many hero, how many lifestyle, how many detail). Ask for 9 or ask for 30. The outcome is a complete listing set generated from a single source photo: hero on white, lifestyle scenes, macro and detail shots, angle variations, with lighting and reflections handled automatically, dropped into folders by type. The manual version is a studio booking and a retouching round. This is one prompt against one photo, finished in the length of a coffee break.

This is for anyone shipping to a PDP or an Amazon listing who does not want to book a shoot for every SKU refresh. Wide catalog, slim creative budget: that is where it pays off hardest.

Setup (about 5 minutes)

  1. Connect Higgsfield as an MCP:
claude mcp add --transport http --scope user higgsfield <https://mcp.higgsfield.ai/mcp>
  1. Verify it registered:
claude mcp list

(or /mcp inside the session)