> ## Documentation Index
> Fetch the complete documentation index at: https://opinionai.mintlify.site/llms.txt
> Use this file to discover all available pages before exploring further.

# Media Generation

> Create realtime images and videos inside the Hub

# Realtime Media Generation

Aivah’s Hub doubles as a creative studio. Within the same chat window you use for Q\&A, you can prompt high-quality images, turn them into videos, and keep everything organized automatically. This guide covers the full workflow, the models involved, and best practices for repeatable results.

## How It Works

* **Unified Input**: Use the chat bar at the bottom of any scene. There is no separate tool—just natural language prompts.
* **Realtime Rendering**: Responses arrive in-line with the conversation. Image thumbnails and inline video players appear as soon as generation finishes.
* **Automatic Storage**: Every asset is saved to [AI Drive](/integration/ai-drive) so your team can preview, filter, and download media later.
* **Agent-Aware**: Worker and Presenter Agents can both generate media while staying grounded in their knowledge sources.

## Supported Models

Aivah routes prompts to a media stack purpose-built for speed and fidelity:

* **Google Nano Banana** – optimized for rapid concept art and illustrated styles.
* **Veo 3.1** – delivers cinematic video motion from still images or storyboard prompts.
* **Sora 2** – handles realism-focused scenes with advanced physics and lighting.

You do not need to select these models manually; the platform chooses the best option based on your request. Advanced users can reference the model in their prompt if a specific aesthetic is required (e.g., “using Veo 3.1, create…”).

## Step-by-Step Workflow

<Steps>
  <Step title="Confirm Agent Settings">
    Open the gear icon in the Hub to verify persona, LLM, and voice. This ensures generated content matches the agent’s context.
  </Step>

  <Step title="Prompt an Image">
    Describe what you need, such as “Create an image of a futuristic city skyline at night.” Press **Enter** to submit.
  </Step>

  <Step title="Review the Thumbnail">
    The agent responds with a compact preview. Click it to view the full-resolution image.
  </Step>

  <Step title="Request a Video">
    Follow up with “Generate a short video based on this image.” The resulting clip appears with a built-in player.
  </Step>

  <Step title="Open AI Drive">
    Go to Integrations → AI Drive to see the newly generated assets categorized as images 📷 or videos 🎥.
  </Step>
</Steps>

## Prompting Tips

* **Be Specific**: Mention mood, lighting, angle, or motion cues to steer the output.
* **Iterate Quickly**: Ask for variations (“Try a sunset version”) to produce alternate looks.
* **Chain Context**: Reference earlier messages (“Make a video sequel to the last clip”) to keep the creative thread.
* **Document Choices**: Note which model handled the request when you find a style you like.

## Managing Assets in AI Drive

* **Filtering**: Switch between “All”, “Images”, or “Videos” to zero in on what you need.
* **Previewing**: Click entries for full-size previews before downloading.
* **Downloading**: Use the download icon to save locally or share with collaborators.
* **Version Control**: Older iterations remain available, so you can compare revisions over time.

## Use Cases

* **Sales & Marketing**: Build product visuals or teaser videos during live demos.
* **Events & Training**: Generate supporting media for webinars or workshops on the fly.
* **Support & Success**: Provide tailored visual explanations while troubleshooting with customers.
* **Content Teams**: Rapidly prototype campaign assets before handing off to designers.

## Troubleshooting

* **Delayed Responses**: Complex scenes may take several seconds. The agent posts status updates if rendering exceeds normal timing.
* **Unexpected Style**: Adjust adjectives in your prompt or specify a model (e.g., “Render with Sora 2 for realistic lighting”).
* **Missing Asset in AI Drive**: Refresh the page; large videos might appear a few moments after the chat response.
* **Download Failures**: Ensure browser downloads are allowed and retry; assets remain stored until you delete them.

Ready to create? Open the Hub, send your first prompt, and watch the media appear in real time.