# Virtual Try-On with customer photos

Generate an image of a visitor wearing selected products using their photo and product images.

*Release 3.3 · admin, operator, visitor*

**Settings › AI › Functions** · **Settings › AI › Features**

## What it is

**Virtual Try-On** combines a visitor's photo with one or more product images to generate a new image. The agent can attach the result to the conversation. It is a generated visualisation, not a measurement of fit or a guarantee of how a product will look.

## Who uses it and why

- Visitors supply a photo and identify the products they want to see.
- Operators help visitors use the result during a shopping conversation.
- Administrators configure the image service and make the function available to the agent.

## Where to find it

Configure the **Virtual Try-On** function in **Settings › AI › Functions**, then make it available on the relevant feature in **Settings › AI › Features**. Visitors receive the generated image in a supported conversation attachment view.

## How it works

1. The visitor supplies a photo in the conversation and selects the products to try on.
2. The agent calls the configured function with the visitor photo, product image references and any extra styling instructions.
3. The image service generates the result. Some providers finish through a background job, so the result may take time.
4. INSIDE saves the generated image and attempts to attach it as a public conversation message.

The built-in prompt asks the model to preserve the person's identity, pose and proportions, retain the scene where possible and keep product details faithful to the supplied images. These are generation instructions, not guaranteed image properties.

## Configure it

With the **Settings** role and the **AI** permission, or the **Developer** role, configure the function's **Imagine API Key**, **Image Provider** and **Image Model**. Ask Powerfront for the service details needed for your setup. The provider choices are **Google AI Studio**, **OpenAI** and **xAI**.

Leave **Imagine URL** blank to use the default service. **Provider API Key** is an optional override. Leave **Prompt Template** blank for the built-in instructions, or supply instructions suited to your products.

Make the function available to the intended AI feature and explain when the agent should use it.

### Settings

| Setting | Where | Applies to | Default | Who can change it | What it does |
|---|---|---|---|---|---|
| Imagine URL | Settings › AI › Functions | website | Not set; use the default Imagine service | Settings role, Developer role | Selects the image generation service used by this function. |
| Image Provider | Settings › AI › Functions | website | Not set | Settings role, Developer role | Chooses Google AI Studio, OpenAI or xAI for image generation; a provider must be selected before saving. |
| Image Model | Settings › AI › Functions | website | Not set | Settings role, Developer role | Selects the image generation model for the provider; a model must be selected before saving. |
| Prompt Template | Settings › AI › Functions | website | Not set; use the built-in fashion editing instructions | Settings role, Developer role | Replaces the instructions used to combine the visitor photo and product images. |
| Imagine API Key | Settings › AI › Functions | website | Not set | Settings role, Developer role | Supplies the image generation service credential; a key must be supplied before saving. |
| Provider API Key | Settings › AI › Functions | website | Not set; use the service's configured provider credential | Settings role, Developer role | Overrides the service's configured credential for the image provider. |

### Permissions

| Permission | Lets you |
|---|---|
| Settings | Manage the function when AI is allowed under Settings Permissions. |
| AI | Allows AI configuration within Settings Permissions. |
| Developer | Manage AI functions and attach them to features. |

## Limits and good to know

- The function needs a visitor photo and product image references.
- Uploaded and downloaded input images are checked against a 10 MiB size limit. Downloaded content must be an image.
- Image generation can wait up to five minutes and can fail or time out.
- Automatic attachment delivery needs a visitor conversation with an assigned operator. Image creation can succeed even when attachment delivery fails.
- The result is public when attached to the conversation, so the visitor can see it.

## Depends on and works with

### This feature uses

- **Requires** [AI functions and required inputs](AI-016.md): Virtual Try-On must be available as a function on the AI feature.
- **Configured in** [Roles and permissions](ADM-008.md): AI configuration permissions control function setup.
- **Configured in** [AI feature settings and change history](AI-001.md): Feature configuration determines which agent can call the function.
- **Works better with** [Attachments in conversations](CHAT-018.md): Conversation attachments can supply the visitor photo and deliver the generated image.
- **Works better with** [File attachments for visitors](VX-021.md): Supported visitor chat interfaces display the generated image attachment.

### Used by

- **Enhances** [AI functions and required inputs](AI-016.md): Virtual Try-On is available as a configured function type.

## FAQ and troubleshooting

### Why does the agent say Virtual Try-On is not configured?

Check **Imagine API Key**, **Image Provider** and **Image Model** on the function. All three are required.

### Why was a product image rejected?

The reference may be unavailable, too large or return content that is not an image. Use an accessible product image and try again.

### Why was an image created but not shown in chat?

Automatic delivery needs an active visitor conversation and an assigned operator. Ask Powerfront to investigate a reported attachment failure.

### Will the result show exact sizing?

No. The function generates an image from photos and instructions. It does not measure the person or validate product sizing.