Working with the Reality Composer Pro Assistant
Connect an AI model to Reality Composer Pro to generate assets and get contextual help building scenes.
Overview
The Reality Composer Pro Assistant connects an AI model to Reality Composer Pro so you can get feature guidance, generate 3D assets and materials, and manage your scene hierarchy. You can ask the Assistant how to use Reality Composer Pro features, request that it generate assets from prompts or attached images, and direct it to reorganize entities in your scene.
Add a model provider
In the Inspector, click the Assistant tab (top right) to open the Assistant panel. Click the Settings (gear) icon at the bottom of the panel to open Assistant Settings, then click Add a Model Provider. In the dialog, enter a URL, an optional API Key, an API Key Header, and a Description, then click Create.
[Image]
To add more AI models, repeat this process with any compatible AI model provider.
Use the Assistant
At the top of the Inspector, click the Assistant tab. In New Conversation, enter your prompt and press Return. The Assistant responds in the conversation panel. To start a fresh conversation, click New Conversation. You can also attach image files to your prompt by dragging them into the conversation panel.
Explore prompt ideas
Experiment with different prompts to discover what the Assistant can do.
Ask about features
Ask the Assistant how to use specific Reality Composer Pro features.
“What is the Shader Graph and how do I use it?”
“What is a Script Graph used for?”
“How do I create a prototype?”
Generate assets from a prompt
Ask the Assistant to generate assets. Add as much or as little detail as needed in your prompt.
“Make a living room scene.”
Attach an image of a real-world object or space and ask the Assistant to create a scene or model from it. For example, “Make a living room area based on the attached image.”
Generate an image, then create a model from it
You can also ask the Assistant to generate a new image itself, then use that AI-generated image to build a 3D model — no user-supplied photo required.
“Create an image of a moose.”
“Create an image of a moose, and then create a model based on the image.”