Guide

AI model photo generator: virtual model shots from one photo

Turn a single product photo into on-model shots from inside Claude — no studio, no casting, no shoot to book.

The same shirt rendered on an AI-generated virtual model, full-body front view
A flat product photo of a light blue oxford shirt — the input image
Real · Your photo AI · Dreem ‹ ›

An AI model photo generator turns a single flat garment photo into polished on-model shots — no studio, model, or shoot to book. You upload your product image, the AI renders it on a realistic virtual model in seconds, and you publish. This guide shows how to do exactly that from Claude, using the Dreem MCP, step by step.

What are virtual model shots?

A virtual model shot — sometimes called an on-model or AI fashion model image — shows your garment worn by a digitally generated person. Instead of booking a model, a photographer, and a studio day, you generate the same hero image from a product photo you already have.

For apparel brands, on-model imagery is the shot that sells. Shoppers want to see how a piece sits on a real body: the fit through the shoulders, the length, the way a fabric falls. Flat lays and packshots show the product; an on-model shot shows the promise. Traditionally that meant a casting, a booking, and a few hundred dollars per look. An AI model photo generator collapses all of that into an upload.

Why generate them from Claude with the Dreem MCP

Dreem is a product-photography engine built for apparel and wearables. It exposes its tools through an MCP server — the Model Context Protocol, the open standard that lets AI assistants like Claude call external tools directly. Connect the Dreem MCP once, and you can ask Claude, in plain language, to generate on-model shots, ghost-mannequin images, packshots, or video from a single product photo.

The payoff is that generation happens inside your conversation. There's no separate dashboard to open and no export-import shuffle between tools. If you already work in Claude, you can describe the shot you want, hand it a product image, and get a finished model image back in the same thread — then iterate on pose, talent, or framing without leaving the chat.

How to generate virtual model shots, step by step

The whole setup takes a couple of minutes, and you only do the connection part once.

Step 1 — Connect the Dreem MCP server in Claude

In Claude, open Connectors and choose Add custom connector. Give it a name — dreem — and paste the server URL:

https://mcp.dreem.ai/mcp

Leave the OAuth fields blank unless you have client credentials, then click Connect.

Claude's Add custom connector dialog with the name dreem and the server URL https://mcp.dreem.ai/mcp
Add a custom connector in Claude and point it at the Dreem MCP server.

Step 2 — Grant access to your Dreem account

Claude opens a Dreem permission screen. Sign in — or create a free account — and approve the read and write scopes so Claude can send images to Dreem and pull the results back. Click Allow Access. You only do this once.

Dreem's Grant Permissions screen showing MCP Read and MCP Write access for Claude with an Allow Access button
Approve MCP read and write access so Claude can generate on your Dreem account.

Step 3 — Prompt Claude with your product image

Start a new chat, describe the shot in plain English, and point Claude at your product image. A public image URL works well — for example a hosted PNG of a base garment:

generate virtual model with this input image: https://…/baseshirts/putih.png

Be as specific as you like about pose, framing, or talent, or let Dreem choose for you.

A Claude chat with the prompt: generate virtual model with this input image, followed by a product image URL
Describe the shot and give Claude a link to your product image.

Step 4 — Get your virtual model shot back

Claude calls the Dreem tools — selecting a talent and a pose, then generating — and returns the finished image right in the conversation. In this run it picked Marcella Vento, a front, full-body standing pose, and rendered the white t-shirt on her in about 95 seconds. Not quite right? Ask for a different talent, a new pose, or a male model in the same thread and generate again.

Claude's reply showing the generated virtual model is ready, styled on Marcella Vento in a full-body front pose, generated in about 95 seconds
Dreem returns the finished virtual model shot inside the Claude conversation.
A real Dreem on-model result: a white t-shirt rendered on a virtual model, ready to publish
The payoff — a real Dreem on-model result. The white tee rendered on a virtual model from one product photo, ready for a product page.

Tips for better results

  • Start with a clean product image. A flat lay or packshot with even lighting and a plain background gives the generator the most to work with.
  • Be specific when consistency matters. If a whole catalog needs the same look, name the pose, crop, and talent in your prompt so every SKU matches.
  • Generate a few variations. It's quick — make two or three and pick the strongest rather than settling for the first.
  • Keep one thread per product. Context carries within a Claude conversation, so follow-ups like "same model, back view" just work.
  • Review the details. Fine logo detail and reflective surfaces are still improving — check the output before it goes live if a crisp logo is the hero of the shot.

Virtual model vs. ghost mannequin vs. flat lay

Not every product needs a person in the shot. Pick by what the shopper needs to see:

  • Virtual model — when fit on a body is the selling point: dresses, knitwear, outerwear, anything where drape and proportion matter.
  • Ghost mannequin — when you want to show structure cleanly without a person. The invisible-mannequin look reads great on a product grid, and you can generate it with Dreem too. See our ghost mannequin AI tool.
  • Flat lay — when the design is the story, like t-shirt graphics or flat-pack basics, or for accessories that don't need a body at all.

Most apparel stores run a mix: flat lays for the quick catalog grid, ghost mannequin or on-model shots for the hero image on the product page.

Good to know

Frequently asked questions

What is an AI model photo generator?
An AI model photo generator creates on-model product images from a flat photo of a garment. Instead of booking a model, photographer, and studio, you upload your product shot and the AI renders the piece worn by a realistic virtual model in seconds.
Do I need coding skills to use the Dreem MCP with Claude?
No. You connect the Dreem MCP server once through Claude's custom connector settings, then ask for shots in plain English. There's no code to write and no separate dashboard to learn.
What products does virtual model generation work best for?
Apparel and wearables — tops, t-shirts, knitwear, dresses, and outerwear — where fit on a body is the selling point. Dreem is built specifically for product photography of things people wear.
Can I choose the model, pose, or framing?
Yes. Describe the talent, pose, and crop you want in your prompt, or let Dreem choose and then ask for a different model or pose in the same thread until it's right.
How long does it take and how much does it cost?
A single virtual model shot generates in roughly a minute and a half. Compared with a traditional on-model shoot that runs into the hundreds of dollars per look, per-image cost on Dreem lands in the low single digits. See the pricing page for current plans.